article
stringlengths 0
745k
| abstract
stringlengths 0
92.7k
|
|---|---|
the statistics of minkowski functionals , including the euler number , as well as extrema counts requires the knowledge of the one - point joint probability distribution function ( jpdf ) @xmath1 of the field @xmath2 ( assumed to have zero mean ) , its first , @xmath3 , and second , @xmath4 , derivatives .
let us consider a random field @xmath2 defined on a 2d sphere @xmath0 of radius @xmath5 represented as the expansion in spherical harmonics @xmath6 where for the gaussian statistically homogeneous and isotropic field random coefficients @xmath7 are uncorrelated with @xmath8-independent variances @xmath9 of each harmonic @xmath10 the variance of the field is then given by @xmath11 when considering derivatives in the curved space , we use covariant derivatives @xmath12 , @xmath13 , @xmath14 , @xmath15 , @xmath16 where it will be seen immediately that mixed version for the second derivatives is the most appropriate choice .
the 2d rotation - invariant combinations of derivatives are @xmath17 where @xmath18 is linear in the field and @xmath19 and @xmath20 are quadratic , always positive , quantities .
the derivatives are also random gaussian variables , which variances are easily computed @xmath21 where the fundamental difference between a sphere and the 2d cartesian space is in the fact that @xmath22 . among the cross - correlations the only non - zero one is between the field and its laplacian @xmath23 .
from now on we rescale all random quantities by their variances , so that rescaled variables have @xmath24 . introducing @xmath25 ( where the spectral parameter @xmath26 ) leads to the following simple jpdf for the gaussian 2d fields @xmath27 \ , .
\label{eq:2dg}\ ] ] in @xcite we have observed that for non - gaussian jpdf the invariant approach immediately suggests a gram - charlier expansion in terms of the orthogonal polynomials defined by the kernel @xmath28 . since @xmath29 , @xmath19 , @xmath18 and @xmath20
are uncorrelated variables in the gaussian limit , the resulting expansion is @xmath30 , \label{eq:2dp_general}\end{aligned}\ ] ] where terms are sorted in the order of the field power @xmath31 and @xmath32 stands for summation over all combinations of non - negative @xmath33 such that @xmath34 adds to the order of the expansion term @xmath31 .
@xmath35 are ( _ probabilists _ ) hermite and @xmath36 are laguerre polynomials .
the coefficients of expansion @xmath37 are related ( and for the first non - gaussian order @xmath38 are equal ) to the moments of the field and its derivatives ( see @xcite for details ) . up to now our considerations
are practically identical to the theory in the cartesian space , which facilitates using many of the cartesian calculations .
we stress again the only , but important , difference being @xmath22 .
we shall see in the next sections how this difference plays out . here
we introduce the spectral parameter @xmath39 that describes this difference @xmath40 let us review the scales and parameters that the theory has . as in the flat space , we have two scales @xmath41 and @xmath42 and the spectral parameter @xmath43 ( which also describes correlation between the field and its second derivatives ) . on a sphere
we have a third scale , the curvature radius @xmath5 .
the meaning of the additional spectral parameter @xmath39 becomes clear if we notice that @xmath44 , thus @xmath45 , i.e describes the ratio of the correlation scale @xmath46 to the curvature of the sphere . as with @xmath47 , @xmath39 varies from @xmath48 to @xmath49 , with @xmath50 corresponding to the flat space limit . from eq .
( [ eq : beta ] ) we find that @xmath51 is achieved when the field has only the monopole and the dipole in its spectral decomposition .
there are three minkowski functionals that are defined for the excursion set above threshold @xmath52 of a 2d field , namely the filling factor , @xmath53 , i.e the volume fraction occupied by the region above the threshold @xmath52 , the length ( per unit volume ) of isofield contours , @xmath54 and euler characteristic @xmath55 .
statistics of the first two do not depend on the second derivatives of the field , and thus are identical on @xmath0 and the 2d cartesian space . here , for completeness
, we reproduce the non - gaussian expansions for these quantities from @xcite @xmath56 @xmath57 euler characteristic density of the region above a threshold @xmath58 is a more interesting case . it is given by the average of the determinant of the hessian matrix of the second derivatives of the field at the points where the first derivatives vanish @xcite @xmath59 it has been argued in @xcite that on @xmath0 the determinant should be that of the hessian of the mixed covariant derivatives @xmath60 . it is this choice that provides the density relative to the invariant volume element @xmath61 and has a scalar trace equal to the laplacian of the field . using scaled invariant variables @xmath62 where we have introduced another scaled quadratic invariant @xmath63 . in terms of the eigenvalues of the hessian , @xmath64 , while @xmath65 and @xmath66 . in the gaussian limit
the euler characteristic density becomes @xmath67 \left ( j_1 ^ 2- ( 1-\beta ) j_2 \right ) \label{eq : chi_gauss}\ ] ] it evaluates to @xmath68 which differ from the well known cartesian result by the @xmath69 term . on a sphere which has a finite volume
@xmath70 it is appropriate to quote the total euler characteristic in the whole volume , which , recalling the relation between @xmath47 , @xmath39 , @xmath5 and @xmath46 becomes @xmath71 which explicitly demonstrates that if @xmath72 .
i.e the whole space is included in the excursion set , the total euler characteristic is equal to that of a sphere , @xmath73 , as expected .
evaluation of the non - gaussian expansion for @xmath55 entails integrating eq .
( [ eq : chi_int ] ) with the distribution function @xmath74 given by eq .
( [ eq:2dp_general ] ) .
the procedure is similar to that in cartesian space as elaborated in detail in @xcite and which led to the complete expression for the euler characteristic to all orders first reported in @xcite .
indeed , the quantity @xmath63 that we then average over the whole range of @xmath20 can be rewritten as @xmath75 .
thus only the @xmath76 terms of the expansion , i.e containing @xmath77 or @xmath78 , do not vanish after integration . here
we should limit ourselves to presenting only the result of the most practical use - up to the first , cubic in the field , non - gaussian correction @xmath79\ , , \label{eq : euler}\end{aligned}\ ] ] where the gram - charlier moments of the non - primary variables @xmath2 and @xmath63 are understood as correspondent combinations of gram - charlier moments of the expansion variables @xmath29 , @xmath18 and @xmath20 .
the first term of eq .
( [ eq : euler ] ) is the gaussian result on the sphere that is responsible for the total euler number of the excursion set to be that of the total sphere when @xmath72 .
the last @xmath80 terms is a correction to the non - gaussian result due to the curvature of the sphere . in conclusion we as well write explicitly the result for the total euler number above threshold @xmath52 @xmath81\ , .
\nonumber \label{eq : euler_tot}\end{aligned}\ ] ]
the number density of extrema above a threshold @xmath52 is given by an integral very similar to the euler characteristic ( @xcite ) @xmath82 where the function @xmath83 chooses the regions of integration in the space of second derivatives with appropriate to the particular extremum type signs of the hessian eigenvalues .
in 2d , assuming @xmath84 , @xmath85 for maxima , @xmath86 for minima , and @xmath87 for saddle points .
in particular , a well - known topological relation gives @xmath88 the integral eq . ( [ eq : ext_int ] ) has a very transparent form when the hessian is described in invariant variables .
it is equivalent to eq .
( [ eq : chi_gauss ] ) , except that the limits of integration over @xmath18 are partitioned into the regions of fixed sign of the determinant @xmath63 .
namely , maxima correspond to the range @xmath89 , minima to @xmath90 and saddle points to @xmath91 .
calculations for the differential density of extremal points , @xmath92 , can be carried out analytically even for the general expression eq .
( [ eq:2dp_general ] ) ( see discussion for the flat case in @xcite ) .
the resulting expressions are cumbersome , and here we limit ourselves to presenting results for the total density of extrema to first non - gaussian order only .
the total number density of maxima is given by @xmath93 and , similarly , for the minima and the saddle points .
the result is @xmath94 where we immediately see that @xmath95 as expected .
the total number of saddles , as well as of all the extremal points , @xmath96 , are preserved at first order ( the latter following from the former ) , but the symmetry between the minima and the maxima is broken .
it is instructive to look how the gaussian extrema counts are modified by the properties of spherical space when the curvature radius is large relative to the typical extrema separation scale @xmath46 , i.e when @xmath39 is small .
up to first order in @xmath39 @xmath97 this shows that being on a sphere increases the number density of maxima and minima , but decreases ( and in a more significant way ) the number of saddles .
incidently , assuming large - angle cmb power spectrum , truncated at @xmath98 gives @xmath99 , i.e 1% correction to the count of extrema relative to the flat - sky approximation .
|
in the conference presentation we have reviewed the theory of non - gaussian geometrical measures for the 3d cosmic web of the matter distribution in the universe and 2d sky data , such as cosmic microwave background ( cmb ) maps that was developed in a series of our papers .
the theory leverages symmetry of isotropic statistics such as minkowski functionals and extrema counts to develop post- gaussian expansion of the statistics in orthogonal polynomials of invariant descriptors of the field , its first and second derivatives .
the application of the approach to 2d fields defined on a spherical sky was suggested , but never rigorously developed . in this paper
we present such development treating effects of the curvature and finiteness of the spherical space @xmath0 exactly , without relying on the flat - sky approximation .
we present minkowski functionals , including euler characteristic and extrema counts to the first non - gaussian correction , suitable for weakly non - gaussian fields on a sphere , of which cmb is the prime example .
random fields are ubiquitous phenomena in physics appearing in areas ranging from turbulence to the landscape of string theories . in cosmology , the sky - maps of the polarized cosmic microwave background ( cmb ) radiation a focal topic of current research is a prime example of such 2d random fields , specified on @xmath0 spherical space .
modern view of the cosmos , developed primarily through statistical analysis of these fields , points to a universe that is statistically homogeneous and isotropic with a hierarchy of structures arising from small gaussian fluctuations of quantum origin .
while the gaussian limit provides the fundamental starting point in the study of random fields @xcite , non - gaussian features of the cmb fields are of great interest .
indeed , cmb inherits a high level of gaussianity from initial fluctuations , but small non - gaussian deviations may provide a unique window into the details of processes in the early universe .
the search for the best methods to analyze non - gaussian random fields is ongoing . in the paper @xcite the general invariant based formalism for computing topological and geometrical characteristics of non gaussian fields
was presented .
the general formula for the euler characteristic to all orders has been derived , which encompasses the well known first order correction of @xcite , and which was later confirmed to the next order by @xcite .
this work was followed by the detailed exposition of the theory in 2d and 3d flat ( cartesian ) space in @xcite and @xcite , and generalized to the 3d redshift space where isotropy is broken in @xcite .
the goal of this paper is to extend these results to the fields defined on a finite curved spherical space @xmath0 without reliance on the flat field ( small angle ) approximation .
while these proceedings were being prepared , similar work has been done for statistics of peaks in the gaussian case within @xcite . here
our focus is on non - gaussian corrections .
we discuss how to compute exact minkowski functionals for the excursion sets of a scalar field on a @xmath0 sphere to all orders in non - gaussian expansion and provide an explicit expression for the euler characteristic to first order .
expressions for the total extrema counts to the first non - gaussian order are also given , while analytical formulas for differential extrema counts to the same order will be published elsewhere due to their length .
these results have a direct relevance to cmb data analysis .
|
the basic idea of this communication is to show that the well - known net redshift of transition region ( tr ) emission and the new observations of the ly-@xmath0 and ly-@xmath1 profiles obtained by soho - sumer are different manifestations of the same fundamental massflow process .
it is known since long that in tr emission all lines appear with a net redshift , which peaks at a temperature around log@xmath2/k=5 ( e.g. , brekke 1997 ) .
dammasch et al .
( 2008 ) have demonstrated that this redshift can be explained by the fact that closed loops higher up in the tr have both footpoints redshifted .
such coronal downflows have also been reported by hinode ( tripathi 2009 , del zanna 2008 ) .
already 30 years ago foukal ( 1978 ) mentioned such flows and introduced the term coronal convection. the common understanding of these observations is that in tr loops that are closed , material is draining down on both sides . near their footpoints
, the loops are not only redshifted , but they are also brighter than elsewhere , and this brightness - to - redshift relationship easily explains the net redshift of tr emission and the redshift in the network contrast ( curdt et al .
we show that clear signatures of dynamical processes are also seen in the optically thick lines of the hydrogen lyman series .
the entire lyman series and the long - wavelength portion of the hydrogen recombination continuum fall into the spectral range of the sumer spectrometer .
ly-@xmath0 by far the most important line and carrying 75% of the spectral radiance in this range plays an important role for the radiative energy transport in the solar atmosphere . despite of this importance ,
however , such observations were not completed , because the bright ly-@xmath0 line would saturate the detector .
recently the sumer team started unconventional observations , where 80% of the aperture is vignetted by the partially closed door .
several of these rasters were completed at various @xmath3 angles in summer 2008 , when the sun was very quiet . in april 2009
also coronal hole ( ch ) data was obtained .
later on , prominence , filament and sunspot targets were selected . in some of the rasters ,
the wavelength setting was within 23.3s switched back and forth between ly-@xmath0 and ly-@xmath1 , thus allowing quasi - simultaneous exposures for those lines ( for observational details see curdt et al .
2008b , 2010a , 2010b , tian et al .
2009a , 2009b ) .
some of the results that are relevant for our topic are described below . in all exposures we also recorded a tr line as line - of - sight ( los ) velocity indicator . in the case of ly-@xmath0 ,
the doppler flow in each pixel has been determined by the shift of the @xmath41206si line centroid .
we sort the ly-@xmath0 profiles by the doppler flow in each pixel and define six equally spaced velocity bins .
the different profiles of each bin are displayed in fig.1a .
there is an obvious and remarkable result : the blue horn is always stronger than the red horn , and the asymmetry increases significantly with the redshift . both peaks and the central reversal are offset towards the red with increasing downflow , as indicated by tracing lines .
similarly , we display in fig.1b the profiles of ly-@xmath1 .
also for ly-@xmath1 there is an obvious correspondence between asymmetry and redshift and between asymmetry and radiance .
but most surprisingly the asymmetries of ly-@xmath0 and ly-@xmath1 are reversed .
it is well known that higher order lyman lines become more optically thin .
our results clearly show that the trend towards self - reversed , red - horn dominated profiles is discontinued for ly-@xmath0 .
interestingly , this has already been noticed three decades ago in oso8 data ( gouttebroze et al . 1978 ) .
they clearly show in their fig.1 the red - horn dominance in ly-@xmath1 and the blue - horn dominance of ly-@xmath0 .
but they do nt give any explanation other than : the asymmetries are variable with time and location and are probably related to large - scale motions of the atmosphere. we now arrived at a heuristic model to explain the opposite asymmetries , where downflows and opacity are dominating the profile , but play a different role .
both lines are optically thick , but the opacity in ly@xmath5 is much less than that in ly@xmath6 . in fig.2 we show the average line profile of several chromospheric lines around ly-@xmath1 obtained in a quiet sun ( qs ) location .
we adopted the wavelength values of the sumer spectral atlas ( curdt et al .
2001 ) and completed a precision 5-component multigaussfit , assuming a negative gaussian to account for the absorption .
the fit with the least errors was achieved , if we assume the absorption to occur at zero velocity and the full curve at a redshift of 2 to 3 km / s relative to this .
this suggests that ly-@xmath1 ( and the higher - order lyman lines ) still behave like typical tr lines , even in ly-@xmath1 the redshifted footpoints are shining through . for ly-@xmath0 things change ,
the opacity of ly@xmath6 is so high , that any directional information is lost at @xmath7 because of the partial redistribution process .
in contrast to ly@xmath5 and all the higher lyman lines , which are still dominated by redshifted _ emission _ ( which narrows the blue peak ) , ly@xmath6 is dominated by redshifted _ absorption _ ( which suppresses the red peak ) . below we will give additional proof for this interpretation .
in fig.3 we show a mdi magnetogram of a qs area near disk center the overlaid contours indicate enhanced ly-@xmath0 radiance and show the top-25% level . here
we sort the profiles by the relative depth of their self - reversal . in panel a we marked pixels out of the bottom-25% regime , areas with flatter profiles ( reduced opacity ) obviously cluster along the network lane . in contrast , pixels from the top-25% regime - shown in panel b - cluster in the cell interior .
this suggests a lower opacity above the magnetic network as compared to the low - lying loops in the cell interior .
we assume that in the network magnetic loops open into over - arching structures , while the cell interiors are filled with low - lying loops .
less self - reversed profiles cluster along the network boundary , here the expanding magnetic field forms funnel - like structures with reduced opacity . in other words ,
chromosphere and transition region are extended above the network .
the arguments raised so far imply that in chs the effects should be even more expressed .
indeed , in the ch we found the largest peak separation in the ly-@xmath0 profile and the deepest self - reversals in the ly-@xmath1 profile , in both cases indication of extremely high opacity ( cf . ,
tian et al .
2009b ) . in fig.4
we show the profiles of ly-@xmath0 ( left panel ) and of ly-@xmath1 as observed in a polar ch .
it is most interesting that the ly-@xmath1 profiles in the polar ch have a blue peak dominance similar to ly-@xmath0 .
xia ( 2003 ) mentioned indications of such a behaviour in equatorial chs . in our work
we find very clearly that most ly-@xmath1 profiles are stronger in the blue peak in the polar ch .
ly-@xmath1 now starts to behave like ly-@xmath0 , and this in line with the argument used before and proves that our interpretation is consistent .
this difference in behaviour suggests that the ch is void of overarching loops . here
, the chromosphere is filled everywhere with low - lying loops , that are observed at an orthogonal view .
in the previous sections we have shown the signature of downflows in optically thin and in optically thick lines and that there is evidence for persistent downflows in closed loops . in foukal s notion this would be the red branch of the coronal convection . the blue branch is less clear .
there is evidence for continuous upflow in magnetically open structures from the spectrometers on soho and hinode . here
, we refer to the work of wiegelmann et al .
( 2005 ) . in their fig.2 ( not reproduced here ) open field lines are shown in brown , closed ones in yellow .
indeed a correlation exists between plasma outflow ( shown in blue ) and places with open field lines . , indicative of a quasi - continuous process that is guided by magnetic field lines ( marsch et al .
also transient scenarios like jets or spicular activity have been suggested by the community to be responsible for the material upflow .
this is not necessary in contradiction with the open - fieldline approach .
but at this time a clear answer can not be given .
gunr et al . ( 2008 ) claim that the orientation of a prominence axis relative to the los has an imprint on the self - reversal depth of ly-@xmath0 , such that the reversal is deeper , when the magnetic field is oriented vertical to the los .
the theoretical work is based on heinzel et al .
( 2001 , 2005 ) . and
observational evidence has recently been reported by schmieder et al .
( 2007 ) and curdt et al .
( 2010a ) .
we correlated the los - component of the extrapolated magnetic field in a qs area at different coronal heights versus the reversal depth the ly-@xmath0 profile .
fig.5 shows a small , but clear anti - correlation , which has its minimum around 6 mm .
this suggests that the orientation of the magnetic field relative to the los has an imprint on the source function and plays an important role not only in a prominence , but does it everywhere .
despite the importance of ly-@xmath0 for radiative transport and many other processes in the solar atmosphere , there is a clear disproportion if one compares ly-@xmath0 focussed articles to publications on optically thin lines .
we presented a set of empirical results , which we managed to explain in a consistent manner and which resulted into a heuristic model , where opacity effects and large - scale flows are the main elements .
we now started quantitative work along our heuristic model .
* brekke , p. , hassler , d. m. , wilhelm , k. : 1997 , @xmath8 * 175 * , 349 . * curdt , w. _ et al . _ : 2001 , @xmath9 * 375 * , 591 . * curdt , w. , tian , h. , teriaca , l. , schhle , u. : 2010a , @xmath9 * 511 * , l4 . * curdt , w. , tian , h. , b.n .
dwivedi , marsch , e. : 2008a , @xmath9 * 491 * , l13 .
* curdt , w. , tian , h. , teriaca , l. , schhle , u. , lemaire , p. : 2008b , @xmath9 * 492 * , l9 . * curdt , w. , tian , h. : 2010b , @xmath10 * 428 * , 81 . * dammasch , i.e. , curdt , w. , dwivedi , b.n . , parenti , s. : 2008 , @xmath11 * 26 * , 2955 .
* del zanna , g. : 2008 , @xmath9 * 481 * l49 . * foukal , p. : 1978 ,
@xmath12 * 223 * , 1046 .
* gouttebroze , p. , lemaire , p. , vial , j .- c . , artzner , g. : 1978 , @xmath12 * 225 * , 655 . *
gunr , s. , heinzel , p. , anzer , u. , schmieder , b. : 2008 , @xmath9 * 490 * , 307 .
* heinzel , p. , anzer , u. : 2001 , @xmath9 * 375 * , 1082 .
* heinzel , p. , anzer , u. , gunr , s. : 2005 , @xmath9 * 442 * , 331 .
* marsch , e. , tian , h. , sun , j. , curdt , w. , wiegelmann , t. : 2008 , @xmath12 * 685 * , 1262 . *
b. , gunr , s. , heinzel , p. , anzer , u. : 2007 , @xmath8 * 241 * , 53 . * tian , h. , curdt , w. , marsch , e. , schhle , u. : 2009a , @xmath9 * 504 * , 239 . * tian , h. , teriaca , l. , curdt , w. , vial , j .- c . : 2009b , @xmath12 * 703 * , l152 .
* tripathi , d. , mason , h.e . ,
dwivedi , b.n .
, del zanna , g. : 2009 , @xmath12 * 694 * , 1256 .
* wiegelmann , t. , xia , l .- d . , marsch , e. : 2005 , @xmath9 * 432 * , l1 . *
xia , l .- d . : 2003 , @xmath13 , georg - august - universitt gttingen .
|
we study the hydrogen lyman emission in various solar features now including ly-@xmath0 observations free from geocoronal absorption and investigate statistically the imprint of flows and of the magnetic field on the line profile and radiance distribution . as a new result
, we found that in ly-@xmath0 rasters locations with higher opacity cluster in the cell interior , while the network has a trend to flatter profiles .
even deeper self reversals and larger peak distances were found in coronal hole spectra .
we also compare simultaneous ly-@xmath0 and ly-@xmath1 profiles .
there is an obvious correspondence between asymmetry and redshift for both lines , but , most surprisingly , the asymmetries of ly-@xmath0 and ly-@xmath1 are opposite . we conclude that in both cases downflows determine the line profile , in case of ly-@xmath0 by absorption and in the case of ly-@xmath1 by emission .
our results show that the magnetically structured atmosphere plays a dominating role in the line formation and indicate the presence of a persisting downflow at both footpoints of closed loops .
we claim that this is the manifestation of a fundamental mass transportation process , which foukal back in 1978 introduced as the coronal convection.
|
modeling the dust and stellar content of spiral galaxies is a very crucial procedure needed for the correct interpretation of the observations .
the amount of interstellar dust embedded inside spiral galaxies , the way that dust is distributed within spiral galaxies and also the extinction effects of the dust to the starlight are some of the questions that can be answered by performing radiative transfer modeling of individual spiral galaxies .
one very important thing that needs consideration when doing such analysis is the right choice of the stellar and dust distributions . in particular ,
the galactic disk is a quite complex system , where stars and dust are mixed together usually in a spiral formation .
for this reason , one has to use realistic distributions able to reproduce quite accurately the observations . on the other hand , simple mathematical expressions for these distributions
are chosen in order to keep the free parameters to the minimum .
for the distribution of the starlight in the disk of spiral galaxies , the exponential function is very widely in use .
this simple mathematical expression is able to describe the distribution of stars in both directions , radially and perpendicular to the disk .
decomposition techniques used by different authors in order to separate the bulge and the disk component strongly support this argument . for galaxies
seen face - on ( and at moderate inclination angles ) , radial profile fitting ( e.g. freeman 1970 ) , fitting to azimuthally averaged profiles ( e.g. boroson 1981 ) , as well as ellipse fitting techniques to 2d images ( e.g. de jong 1995 ) show that the exponential in the radial distance @xmath0 is a good representation of galactic disks with only small deviations mainly due to the spiral structure of the galaxy ( see serna 1997 ) .
other works like those of shaw & gilmore ( 1989 ) and de grijs ( 1997 ) dealing with modeling of edge - on galaxies support the idea that exponential functions are good representations also for the @xmath1 ( vertical to the disk ) direction . performing radiative transfer modeling of edge - on galaxies , xilouris et al .
( 1997 , 1998 , 1999 ) found that exponential functions for the luminosity density of the stars in the disk as well as for the extinction coefficient give an excellent description of the observations .
the advantage of modeling galaxies in the edge - on configuration is that the integration of light along the line of sight is able to cancel out most of the structure of the galaxies ( i.e. spiral structure ) and therfore allows for simple functions such as exponentials to give good representation of the observations .
thus , although in the face - on configuration a large variation between arm and interarm regions might be present for both the stars and the dust ( white & keel 1992 , corradi et al .
1996 , beckman et al .
1996 , gonzalez et al .
1998 ) , in the edge - on case an average description of the galaxy characteristics can be obtained quite accurately .
we are going to investigate the validity of this argument by comparing the exponential distributions with more realistic distributions which include spiral structure . in sect .
2 we describe the method that we use to address this problem and in sect .
3 we present the results of our calculations . finally , in sect .
4 we summarize our work .
the method that we follow in this work consists of two basic steps . in the first step ,
model galaxies with realistic spiral structure are constructed . after a visual inspection of the face - on appearance of these models to see the spiral pattern ,
we create their edge - on images which are now treated as real observations . in the second step we fit these `` observations '' with a galaxy model where
now the galactic disk is described by the widely used plain exponential model . in this way ,
a comparison between the parameters derived from the fitting procedure and those used to produce the artificial `` observations '' can be made and thus a quantitative answer about the validity of the plain exponential model as an approximation to galactic disks can be given .
we adopt a simple , yet realistic , distribution of stars and dust in the artificial galaxy .
a simple expression is needed in order to keep the number of free parameters as small as possible and thus have a better control on the problem .
a realistic spiral structure is that of logarithmic spiral arms ( binney & merrifield 1998 ) .
thus , a simple but realistic artificial spiral galaxy is constructed by imposing the logarithmic spiral arms as a perturbation on an exponential disk . in this way ,
the azimuthally averaged face - on profile of the artificial galaxy has an exponential radial distribution . for the stellar emissivity we use the formula @xmath2 @xmath3\right\}\ ] ] @xmath4 in this expression
the first part describes an exponential disk , the second part gives the spiral perturbation and the third part describes the bulge , which in projection is the well - known @xmath5-law ( christensen 1990 ) .
here @xmath0 , @xmath1 and @xmath6 are the cylindrical coordinates , @xmath7 is the stellar emissivity per unit volume at the center of the disk and @xmath8 and @xmath9 are the scalelength and scaleheight respectively of the stars in the disk .
the amplitude of the spiral perturbation is described by the parameter @xmath10 . when @xmath11 the plain exponential disk is obtained , while the spiral perturbation becomes higher with larger values of @xmath10 .
another parameter that defines the shape of the spiral arms is the pitch angle @xmath12 .
small values of @xmath12 mean that the spiral arms are tightly wound , while larger values produce a looser spiral structure .
the integer @xmath13 gives the number of the spiral arms .
for the bulge , @xmath14 is the stellar emissivity per unit volume at the center , while @xmath15 is defined by @xmath16 with @xmath17 being the effective radius of the bulge and @xmath18 and @xmath19 being the semi - major and semi - minor axis respectively of the bulge . for the dust distribution we use a similar formula as
that adopted for the stellar distribution in the disk , namely @xmath20 @xmath21\right\},\ ] ] where @xmath22 is the extinction coefficient at wavelength @xmath23 at the center of the disk and @xmath24 and @xmath25 are the scalelength and scaleheight respectively of the dust . here
@xmath26 gives the amplitude of the spiral perturbation of the dust .
note that the angle @xmath6 here need not be the same as that in eq .
the stellar arm and the dust arm may have a phase difference between them .
.parameters used to describe a typical spiral galaxy . [ cols= "
< , < , < , < , < , < , < , < " , ] for the parameters describing the exponential disk of the stars and the dust as well as the bulge characteristics we use the mean values derived from the b - band modeling of seven spiral galaxies presented in xilouris et al .
( 1999 ) .
since the most dominant spiral structure in galaxies is that of the two spiral arms ( kennicutt 1981 ; considere & athanassoula 1988 ; puerari & dottori 1992 ) we only consider models where @xmath27 .
galaxies with strong one - arm structure do exist , but they constitute a minority ( rudnick & rix 1998 ) . for the parameter @xmath26
we take the value of 0.4 . with this value
the optical depth calculated in the arm region is roughly twice as much as in the inter - arm region .
this is in good agreement with studies of overlapping galaxies ( e.g. white & keel 1992 ) .
for @xmath10 we use the value of 0.3 resulting ( with the extinction effects included ) in a spiral arm amplitude of @xmath28 mag , which is a typical amplitude seen in radial profiles of face - on spiral galaxies and reproduces the desired strength for the spiral arms ( rix & zaritsky 1995 ) .
for the pitch angle @xmath12 we consider the cases of @xmath29 and @xmath30 , which give a wide variety of spiral patterns from tightly wound to loosely wound .
all the parameters mentioned above are summarized in table 1 .
the radiative transfer is done in the way described by kylafis & bahcall ( 1987 ; see also xilouris et al .
as described in detail in these references , the radiative transfer code is capable of dealing with both absorption and scattering of light by the interstellar dust and also of allowing for various distributions for the stars and the dust . using the model described above and the parameters given in table 1 we produce the images shown in fig .
1 . the top three panels of this figure show the face - on surface brightness distribution of such a galaxy for the three values of the pitch angle ( @xmath29 and @xmath30 ) from left to right .
the spiral structure is evident in these images with the spiral arms being more tightly wound for @xmath31 and looser when @xmath32 . in the middle three panels of fig .
1 we show the distribution of the optical depth when the galaxy is seen face - on for the three different values of the pitch angle mentioned above . in these pictures
one can follow the spiral pattern all the way to the center of the galaxy since the bulge is assumed to contain no dust . finally , in the last three panels of fig . 1
one can see the corresponding edge - on appearance of the galaxies shown face - on in the top three panels .
one thing that is very obvious from fig .
1 ( top and middle panels ) is that the galaxy is no more axisymmetric as it is the case in the plain exponential disk model .
the spiral structure that is now embedded in the model as a perturbation in the disk has broken this symmetry .
thus , in order to do a full analysis of the problem we have to examine the galaxy from different azimuthal views ( position angles ) .
to do so we have created nine edge - on model galaxies ( for each of the three different pitch angles considered here ) , covering the range from @xmath33 to @xmath34 with a step of @xmath35 for the position angle . for the definition of the position angle , see fig .
since the galaxy has exactly the same appearance in the interval from @xmath36 to @xmath37 , we only consider the range of position angles mentioned above . to demonstrate this asymmetry more quantitatively we have computed the central edge - on optical depth ( @xmath38 ) for all these nine model galaxies . unlike the plain exponential disk model where
@xmath38 can be calculated analytically ( @xmath39 ) , here we have to perform numerical integration of eq .
( 3 ) along the line of sight that passes through the center .
the value of @xmath38 is shown in fig .
3 as a function of the position angle . in order to have the full coverage in position angle ( from @xmath33 to @xmath37 ) , the values calculated in the interval ( @xmath33 - @xmath36 ) were repeated in the interval ( @xmath36 - @xmath37 ) . in this figure ,
the three models constructed with pitch angles @xmath29 and @xmath30 are denoted with circles , squares and diamonds respectively . in all three cases ,
a variation of the optical depth with position angle is evident .
the largest variation is found for the case where the pitch angle is @xmath30 and it is @xmath40 .
it is obvious that all the values are around the true value of 27 , used to construct the galaxy ( see table 1 ) .
the edge - on images created as described earlier are now treated as `` observations '' and with a fitting procedure we seek for the values of the parameters of the plain exponential disk that gives the best possible representation of the `` observations '' .
the fitting algorithm is a modification of the levenberg - marquardt routine taken from the minipack library .
the whole procedure is described in detail in xilouris et al .
( 1997 ) .
preliminary tests have shown that the derived values of the parameters describing the bulge are essentially identical to the real values used to construct the model images . in order to simplify the fitting process and since we are only interested in the disk , the bulge parameters were kept constant during the fit .
six parameters are now free to vary .
these are the scalelength and scaleheight of the stellar disk with its central surface brightness ( @xmath8 , @xmath9 and @xmath41 respectively ) as well as the scalelength and scaleheight of the dust and the central edge - on optical depth ( @xmath24 , @xmath25 and @xmath38 respectively ) .
figure 4 shows six graphs . the top left graph gives the variation of the deduced edge - on optical depth of the galaxy as we observe it from different angles . from this graph
one can see that the variation of the deduced optical depth from different points of view is no more than 6% different than the mean value of the optical depth . furthermore , a comparison with fig .
3 , which shows the real value of the optical depth , reveals that there is no systematic error in the derived value .
the deviations are equally distributed around @xmath42 , which is what we would have without the spiral structure .
the variation is of the same order of magnitude regardless of the pitch angle .
the top right graph in fig .
4 presents the deduced central luminosity of the disk .
this graph shows that the variation of the inferred central luminosity of the stellar disk is very small and it is weakly dependent on the pitch angle . in the middle left graph of fig .
4 the derived scalelength of the dust is presented .
the variation of the derived value is about 5% for the 10-degree pitch angle and goes up to 17% for the 30-degree pitch angle .
the increase of the variation with increasing pitch angle is expected , because for large pitch angles the spiral arms are loosely wound , thus causing the galaxy to be less axisymmetric . as a result ,
from some points of view the dust seems to be more concentrated to the center of the galaxy and from other points of view more extended . in the middle right graph of fig .
4 , that shows the scalelength of the stars , it is evident that the same effect occurs for the stars as well .
certain points of view give the impression of a more centrally condensed disk , while others of an extended disk .
the fact that we have taken the stellar and the dust spiral structure to be in phase ( i.e. the dust spiral arms are neither trailing nor leading the stellar spiral arms ) causes the deduced scalelengths of the dust and the stars to also vary in phase .
the case of a phase difference is examined below . the bottom left and bottom right graphs of fig .
4 show the variation of the scaleheight of the dust and the stars respectively .
the variation of both scaleheights is negligible .
this is an attribute of the formula we used for our artificial galaxy .
since the spiral variation we added to the exponential disks is not a function of @xmath1 it is expected that in the @xmath1 direction our artificial galaxy behaves exactly as the exponential model .
there are indications ( van der kruit & searle 1981;wainscoat et al .
1989 ) that the dust arms are not located exactly on the stellar arms .
thus , we re - created the edge - on images , but this time the stellar spiral arms were set to lead the dust arms by 30 degrees .
we then fitted the new images with the exponential model and the deduced parameters are shown in fig .
5 . as in fig.4 ,
the top left graph of fig .
5 shows the optical depth as a function of position angle . a comparison of this graph with the corresponding graph in fig .
4 reveals that the variation of the values derived from the new set of images is significantly larger .
the origin of this effect is the fact that the dust is either in front of the stars ( for some position angles ) or behind the stars ( for other position angles ) .
a strong dependence on the pitch angle is also evident .
note , however , that the mean of all the derived values is unaffected .
the same effect can be seen on the top right graph of fig . 5 , where the derived central luminosity of the disk is plotted .
the variation of the central luminosity is again larger than in the previous case , but the mean value is equal to the true one . in the middle left graph of fig .
5 we show the scalelength of the dust as a function of the position angle .
the variation of the derived dust scalelength can differ as much as 25% for a galaxy with pitch angle equal to 30 degrees .
but again the mean value for all position angles is identical to the one we used to create the images . in the middle right graph
one can see that the scalelength of the stars also varies as much as 30%,but the mean of all the derived values is the correct one .
the left and right bottom graphs show the variation of the scaleheights of the dust and the stars , which is about 10% for the worst case of a 30-degree pitch angle and is practically zero for a galaxy with more tight arms .
the most important conclusion of all the graphs is that the derived values of all quantities tend to distribute equally around the real value we used to create the artificial images .
in our attempt to investigate how significant the spiral structure is when doing radiative transfer modeling of spiral galaxies seen edge - on , we constructed a model galaxy with very prominent spiral arms in the disk .
this quite realistic image of the galaxy is now treated as observation and the widely adopted exponential model for the galactic disk is now fitted to the data .
this analysis shows that the plain exponential disk model is a very accurate description for galactic disks seen edge - on with only small deviations of its parameters from the real ones ( typically a few percent ) .
furthermore , the variation from the real parameters would be averaged out if we could see the same galaxy from several point of views .
this is of course impossible for an individual galaxy , but suggests that if the exponential model is used for a statistical study of many edge - on galaxies no systematic error is introduced .
thus , we conclude that the exponential model is a very good approximation of the galactic disks .
|
we investigate how significant the spiral structure is on calculations concerning radiative transfer in dusty spiral galaxies seen edge - on . the widely adopted exponential disk model ( i.e. both the stars and the dust are distributed exponentially in the radial direction and also perpendicular to the plane of the disk )
is now subject to a detailed comparison with a realistic model that includes spiral structure for the stars and the dust in the disk .
in particular , model images of galaxies with logarithmic spiral arms are constructed , such that the azimuthally averaged disk is exponential in radius and in height , as the observations suggest .
then , pure exponential disk models ( i.e. with no spiral structure ) are used to fit the edge - on appearance of the model images . as a result ,
the parameters derived after the fit are compared to the real values used to create the spiral - structured images .
it turns out that the plain exponential disk model is able to give a very good description of the galactic disk with its parameters varying only by a few percent from their true values .
|
observationally it is well established that elliptical galaxies have dark matter halos , and also host central supermassive black holes .
these empirical facts motivate the study of multi - component dynamical models .
when studying dynamical models of stellar systems ( single or multi - component ) , the minimal requirement to be met by a physically acceptable model is the positivity of the phase - space distribution function ( df ) of each distinct component .
a model satisfying this essential requirement ( which is much weaker than stability , but stronger than the fact that the jeans equations have a physically acceptable solution ) is called _ consistent _ ; moreover , when the total gravitational potential is determined by the total density profile through the poisson equation , the model is called _ self consistent_. in other words , we call self consistent a consistent self - gravitating system .
two general strategies can be used to construct a ( self ) consistent model , or check whether a proposed model is ( self ) consistent : they are commonly referred to as the `` @xmath4to@xmath5 '' and the `` @xmath5to@xmath4 '' approaches , where @xmath4 is the model df ( e.g. , see bertin 2000 , binney & tremaine 2008 ) .
an example of the first approach is the survey of self consistent two
component galaxy models carried out by bertin and co workers , where the stellar and dark matter components are described by two dfs of the @xmath6 family ( e.g. , bertin & stiavelli 1984 , bertin et al .
1992 ) ; other well known examples are the king ( 1966 ) models and the @xmath7 models ( bertin & trenti 2003 ) .
unfortunately , the resulting spatial densities obtained by solving the associated poisson equation are in general not expressible in terms of simple or even known functions , and so only numerical investigations are usually feasible . in the `` @xmath5to@xmath4 '' approach
the density distribution is given , and specific assumptions about the model internal dynamics are made ; in special cases inversion formulae from the density to the df can be obtained , usually in integral form or series expansion ( see , e.g. , fricke 1952 , lynden
bell 1962 , osipkov 1979 , merritt 1985 , hereafter om ; dejonghe 1986 , 1987 ; cuddeford 1991 ; hunter & qian 1993 , ciotti & bertin 2005 ) .
in particular , in order to recover the df of spherical models with orbital anisotropy , the om technique has been developed from the eddington ( 1916 ) inversion for isotropic systems , and widely used to study one and two component models ( see , e.g. , ciotti & pellegrini 1992 , hereafter cp92 ; hiotelis 1994 ; carollo et al .
1995 ; ciotti & lanzoni 1997 ; ciotti 1996 , 1999 , hereafter c96 , c99 ; baes & dejonghe 2004 ; buyle et al .
we remark that the om parameterization is not necessarily the best description of real systems , however its simplicity and the fact that it captures the main features of models of galaxy formation , that are generally found nearly isotropic at the center and increasingly radially anisotropic in the outer envelope ( e.g. , van albada 1982 ; trenti , bertin & van albada 2005 ; nipoti , londrillo & ciotti 2006 ; binney & tremaine 2008 . but see cuesta et al .
2008 and references therein ) , make it the natural choice for investigations as that presented in this paper . in many cases , the difficulties inherent in the operation of recovering analytically the df
prevent a simple consistency analysis , and phase space positivity must be investigated by numerical inspection of the inversion integral . in these cases
the reasons underlying consistency or inconsistency of a proposed model tend to be obscured by the numerical nature of the solution .
fortunately , informations about consistency of multi component om systems can be obtained without recovering their df , following the procedure described in cp92 .
this method uses the radial density profile of each component and the total potential of the system , and gives necessary and sufficient conditions for ( self ) consistency .
moreover , since only spatial differentiation and inequality checks are required , this method is best suited for analytical investigations .
for example , c96 and c99 applied the cp92 technique to the general family of two component , spherically symmetric and radially anisotropic @xmath8 models .
this family is made of the superposition of two @xmath1 models ( dehnen 1993 , tremaine et al . 1994 ) with different total masses , scale lengths , inner density slopes , and om radially anisotropic velocity dispersions . the possibility to investigate the combined effects of radial anisotropy and inner density slope on multi component systems made the study of @xmath8 models interesting , as it is well known that the inner density profile sets important constraints on the amount of admissible radial anisotropy ( e.g. , richstone & tremaine 1984 ) , and indeed in c96 and c99 analytical limitations on anisotropy as a function of the density slopes @xmath0 and @xmath9 were obtained .
these models clarified the reasons behind the numerical findings of cp92 , i.e. the difficulty of consistently superimposing a centrally peaked distribution such as the de vaucouleurs ( 1948 ) profile to a centrally flat one , such as the king ( 1972 ) or quasi isothermal density profile ( even in the isotropic case ) .
in fact , it was shown that the df of the @xmath0 component in isotropic @xmath8 models is nowhere negative , independently of the mass and concentration of the @xmath9 component , whenever @xmath10 and @xmath11 .
on the contrary , a @xmath12 component becomes inconsistent when adding a @xmath13 halo with a small core radius .
thus , in two component isotropic models , the component with the steeper central density distribution is usually the most robust against inconsistency .
more recently , the importance of the central density slope in limiting the amount of possible radial anisotropy has also been quantified with the so called `` cusp slope - central anisotropy '' theorem ( an & evans 2006 , hereafter ae06 ; see also eq . [
28 ] in de bruijne et al .
1996 ) .
the previous investigations left however unexplored the importance of the _ external _ density slope in determining the model consistency .
in fact , the phase space density can not be identified , in general , with any specific spatial position in the system , as ( for example ) stars of a given energy can span a large range of radial positions as a proxy for phase
space density , where @xmath14 is the local value of the velocity dispersion , has no assignable meaning without an appropriate discussion .
for example , in power
law isotropic systems with @xmath15 , because the functions @xmath16 ( for @xmath17 ) and phase space density @xmath18 ( for @xmath19 , where @xmath20 is the so called relative energy , see sect . 4 and baes et al .
2005 ) are both power laws with respect to their arguments , the exponents are related in a simple way . however , the converse statement is not true : for example , in the plummer ( 1911 ) sphere @xmath21 is a power law at all energies , but @xmath22 is not a power law of radius . ]
( systems made of circular orbits are an obvious exception )
. therefore also the external regions of a density distribution can be important in limiting the maximum allowable anisotropy , but the @xmath8 models are of no help in the study of this issue , because the external density profiles of both components all decrease as @xmath23 . for these reasons here we focus on the phase space properties of models , i.e. models similar to the standard @xmath1 models in the inner regions , but with a density profile proportional to @xmath24 ( instead of @xmath23 ) in the external regions ; remarkably , several properties of models can be obtained from those of @xmath1 models by differentiation with respect to their scale length . in this notation , the 4-@xmath1 models coincide with the standard @xmath1 models .
we also study the larger class of ( @xmath25,@xmath0 ) models , i.e. two component systems in which a @xmath0 halo is added to a component .
thus , here we further explore the trends emerged in cp92 , c96 , and c99 , determining the limits imposed by phase
space consistency on the parameters describing ( , @xmath0 ) models , and models with a central bh [ hereafter ( , bh ) models ] , with particular focus on the effects of the external slope parameter @xmath2 . in specific cases
( that we call _ halo - dominated _ models ) , the calculations are performed under the assumption that the mass of the halo component ( or of the central bh ) is dominant over the mass of the visible one .
this assumption is mainly motivated by mathematical simplicity ( see also sect . 4 ) , although this is not the only reason .
in fact , for any given two component model , it is expected that the df properties are bracketed by those of the one component model and by those of the halo - dominated model ( corresponding to the formal case of an infinite halo mass ) . of course ,
while the case of dark matter dominated systems can be considered a viable representation of some real astrophysical systems , the case of a dominant bh is less natural , and it just gives the strongest possible limitations for consistency of systems with a central bh .
the paper is organized as follows . in section 2
we briefly review the technique developed in cp92 , and we prove that the necessary condition for consistency derived in cp92 for om systems can be rewritten _
exactly _ as the ae06 `` cusp slope - central anisotropy '' theorem , holding however at all radii and not just at the center . in section 3
the one and two component models are introduced , and the necessary and sufficient conditions imposed on the model parameters by phase space consistency are derived for different values of the logarithmic density slope @xmath2 . in section 4 the df of the component of
halo - dominated ( @xmath25,1 ) models with @xmath3 , and of similar models with a dominant central bh , are derived explicitly for arbitrary ( but integer ) values of @xmath2 , and the true boundaries of the consistency region in the parameter space are obtained .
the main results of the investigation are summarized in section 5 .
finally , in the appendix an easy method to solve analytically the jeans equations in the general om case for the wider class of ( @xmath26-@xmath0,@xmath27-@xmath9 ) models is presented .
a stellar system made of the sum of the density components @xmath28 is called consistent if each df ( @xmath29 ) is non negative over the whole accessible phase space ; a consistent self gravitating system is called self consistent .
the technique developed in cp92 permits us to check whether the df of a multi component spherical system , where the orbital anisotropy of each component is modeled according to the om parameterization , is positive ; in practice , only information about the radial trend of each density component and of the total integrated mass are used . in the om formulation the df of each component is obtained assuming @xmath30 , with @xmath31 where @xmath32 and @xmath33 are respectively the relative energy and the angular momentum modulus per unit mass , @xmath34 is the so called _ anisotropy radius _ , and @xmath35 for @xmath36 ( e.g. see binney & tremaine 2008 ) .
the velocity dispersion tensor associated with eq .
( [ eq : q ] ) is characterized by radial anisotropy increasing with the radius @xmath37 , while in the limit @xmath38 the system becomes globally isotropic .
as well known , the df of the density component @xmath28 is given by @xmath39 where the _ augmented density _ is @xmath40 @xmath41 is the total relative potential , @xmath42 , and in the integrals above it is assumed that the radius is eliminated from @xmath43 in favour of @xmath44 .
it can be proved that the second equivalence in eq .
( [ eq : fom ] ) holds for untruncated systems with finite total mass , as the models discussed here .
quite obviously , only a very limited number of density profiles among those expressible in analytic form admit an explicit df , so that the study of phase space consistency would appear restricted to such rare cases when conducted analytically , while all the remaining cases should be investigated numerically .
fortunately this is not true , as the cp92 technique is based on the verification ( numerical or analytical ) of the following : a _ necessary condition _ ( nc ) for the non negativity of @xmath29 is @xmath45 if the nc is satisfied , a _ strong sufficient condition _ ( ssc ) for the non negativity of @xmath29 is @xmath46\geq 0 , \quad 0\leq r\leq\infty .\ ] ] finally , a _ weak sufficient condition _ ( wsc ) for the non negativity of @xmath29 is @xmath47\geq 0 , \quad 0\leq r\leq\infty .\ ] ] : see cp92 , c96 , c99 .
the first consideration that follows from the conditions above is that the violation of the nc is connected only to the radial behavior of the augmented density @xmath43 , and so this condition applies independently of any other density component of the model .
obviously , a model failing the nc is _ certainly inconsistent _ , while a model satisfying the nc _ may be inconsistent _ , i.e. the @xmath29 may be negative even for values of model parameters allowed by the nc .
the second consideration is that a model satisfying the wsc ( or the more restrictive ssc ) is _ certainly consistent _ , while a model failing the wsc ( ssc ) _ may be consistent_. therefore the consistency of an om model satisfying the nc and failing the wsc ( or the ssc ) can be proved only by direct inspection of its df . the nc can be recast into a simple inequality between the value of the density slope @xmath48 and the value of the orbital anisotropy indicator @xmath49 that must hold at each radius for consistent om systems . in the expression above @xmath50 and @xmath51 are the radial and tangential components of the velocity dispersion tensor of the system ( e.g. binney & tremaine 2008 ) , and in the following proof we restrict to natural systems , i.e. systems with monotonically decreasing density profile , so that @xmath52 .
the proof is trivial : in fact , it suffices to express the nc in terms of the logarithmic density slope as @xmath53 and from eq .
( [ betaom ] ) one obtains the necessary condition @xmath54 which must be verified by each om component of a consistent multi
component system .
curiously , the condition above is formally _ identical _ to the inequality appearing in the `` cusp slope - central anisotropy '' theorem derived in ae06 .
this latter theorem was proved by using constant anisotropy systems ( i.e. , @xmath55 , and in this case as well the inequality above holds at all radii in consistent models .
[ 10 ] in ae06 ) , and then it was convincingly argued that the inequality it holds asymptotically for the central regions of spherical systems with generic anisotropy distribution . in the specific case of density distributions (
[ [ eq : ng ] ] ) , the logarithmic density slope @xmath56 and therefore eq . ( [ cuspslope ] ) shows that in general the nc is satisfied near the center of om systems ( where @xmath57 and @xmath58 ) , and at large radii ( where @xmath59 and @xmath60 ) , while critical behaviors may be expected at intermediate radii ( see sect . 4 ) .
interestingly ( but not unexpectedly ) the particular functional form of the augmented density characteristic of om parametrization , limits the possible manifestations of phase space inconsistency to few general cases , that can be illustrated as follows . from eqs .
( [ eq : fom])-([eq : rhoom ] ) it is apparent that the df of each density component can be written as @xmath61 where @xmath62 is the df of the considered density component in the isotropic case ( for simplicity , from now on we avoid the use of the subscript @xmath63 ) .
let @xmath64 be the subset of phase space defined by the property that @xmath62 is positive @xmath65 .
then , from eq .
( [ eq : fsa ] ) a first condition to be satisfied for consistency is @xmath66_{q\in a_+}\right\}}.\ ] ] obviously , when @xmath67 over all the accessible phase
space ( the common situation for realistic density distributions ) , inequality ( [ eq : sa- ] ) is also the _ only _ condition to be satisfied for the model consistency , and there is at most a lower bound for the anisotropy radius . for example
this is the case for the families of one component anisotropic @xmath1 models ( c99 ) and sersic ( 1968 ) models ( ciotti 1991 ) . when the complement of @xmath64 is not empty , i.e. @xmath68 over some region @xmath69 of the accessible phase
space , a second inequality again derived from eq .
( [ eq : fsa ] ) must be verified for consistency : @xmath70_{q\in a_-}}.\ ] ] therefore , if there exists some @xmath71 for which @xmath72 , then the proposed model is inconsistent @xmath73 .
all the results presented therein are however correct . ] . if @xmath74 is positive over @xmath69 , from conditions ( [ eq : sa-])-([eq : sa+ ] ) it follows that @xmath75 for om consistency , so that if @xmath76 the proposed model is inconsistent . note that formally _
identical _ considerations hold when discussing the inequalities ( [ eq : nc ] ) and ( [ eq : wsc ] ) , because from eq .
( [ eq : rhoom ] ) it follows that the nc and wsc can be written in the same functional form of eq .
( [ eq : fsa ] ) : of course , in these cases the sets @xmath64 and @xmath69 are to be intended as subsets of the radial range @xmath77 . in the following section , after presenting the one and two component models , the necessary and sufficient conditions for model consistency will be derived , also for the case of models with a massive central black hole .
the one component models are a natural generalization of the widely explored family of the so called @xmath1 models ( dehnen 1993 , tremaine et al . 1994 ) , whose density , cumulative mass within @xmath37 , and relative potential are given by @xmath78 @xmath79}\cr \displaystyle{\ln\frac{r+\rc}{r}},\quad(\gamma=2),\cr \end{cases}\ ] ] respectively . in the models the logarithmic density slope for @xmath80 is not 4 , but it is a free parameter @xmath81 ( to ensure that their total mass is finite ) , so that these density profiles belong to the family considered by zhao ( 1996 ) . in the following , in order to exploit a useful analytical property of the density profiles , we will often assume @xmath2 restricted to integer values @xmath82 .
in fact , the generic density profile of a model is given by @xmath83 where @xmath84 is the total mass of the density distribution , @xmath85 and @xmath86 are the complete euler beta and gamma functions , respectively , and the first expression holds for any real number @xmath81 .
the second expression , based on repeated differentiation of eq .
( [ eq : rhog ] ) with respect to @xmath87 , holds instead for integer @xmath88 . of course , for @xmath89 the standard @xmath1 model density profile is recovered .
the radial behaviour of @xmath90 is shown in fig .
[ f1 ] ( top panel ) for @xmath2 - 1 models with increasing @xmath2 . in the bottom panel
we show the corresponding logarithmic density slopes , calculated accordingly to eq .
( [ gammarhong ] ) .
in particular , it is apparent how the inner slope is @xmath1 ( for @xmath91 ) , and the external is @xmath2 ( for @xmath92 ) . as a consequence , while the density profile outside @xmath87 becomes more and more steep at increasing @xmath2 , the innermost region where the density slope is @xmath1 contracts near the center . of the @xmath2 - 1 model ( top panel ) , and its logarithmic slope ( eq . [ [ gammarhong ] ] , bottom panel ) , for @xmath93 , and @xmath94.,scaledwidth=52.0% ] as anticipated , the possibility to write the density @xmath95 as a derivative with respect to the scale
length @xmath87 provides an easy way to determine analytical properties of the models for @xmath2 integer .
indeed , since the density enters linearly in the integrals of the total mass inside the radius @xmath37 and of the potential , from eq .
( [ eq : ng ] ) it follows that @xmath96 instead , when @xmath2 is not at integer , @xmath97 and @xmath98 are in general given by hypergeometric @xmath99 functions .
expressions similar to eqs .
( [ eq : mng])-([eq : phing ] ) hold for any quantity that can be written as a linear functional of @xmath95 , so that the surface density profile of models with @xmath2 integer can be obtained by repeated differentiation of the surface density profile of @xmath1 models ( when analytically available , e.g. see binney & merrifield 1998 ) . this property will be exploited in section 4 to obtain the explicit df of halo dominated ( , @xmath0 ) models ; moreover , in appendix b we show how quadratic functionals of @xmath95 ( such as the gravitational energy and the velocity dispersions ) can be also evaluated by using repeated differentiation with respect to @xmath87 .
we are now in the position to introduce the two component models used in this work .
the most general family of two component models , i.e. the ( @xmath26-@xmath0,@xmath27-@xmath9 ) models , is made of the superposition of two models with different total masses , scale lengths , internal and external density slopes , and finally two different anisotropy radii . for simplicity here we restrict to the case of ( , @xmath0 ) models , where the `` halo '' density distribution is a standard @xmath0 model : we note however that some of the presented results can be generalized without much effort to the family of ( @xmath26-@xmath0,@xmath27-@xmath9 ) models ( see appendix b ) . in the following , the total mass @xmath100 and the characteristic scale
length @xmath101 of the @xmath0 halo are adopted as normalization constants , so that the physical scales for density and potential are given by @xmath102 and @xmath103 , while we define @xmath104 , @xmath105 , @xmath106 . with this choice , the ( , @xmath0 )
models are structurally determined by fixing the four independent parameters @xmath107 , with the obvious conditions @xmath108 and @xmath109 .
we conclude this introductory discussion by noticing that , for reasons that will become apparent in section 3.2 and 4 , the present normalization differs from that adopted in c99 , where the normalization mass and scale length were those of the first component . before discussing the case of the two component models , we consider the nc for anisotropic models , in order to determine analytically a critical value for the anisotropy radius such that a higher degree of radial om anisotropy ( i.e. , a smaller @xmath110 ) would produce a negative df for some permitted value of @xmath111 .
we recall that the obtained anisotropy limit holds also when a second component is added ( see section 2 ) , so that the present discussion is fully general .
moreover , we note that as the nc involves only the density distribution under scrutiny , we can use the first expression in eq .
( [ eq : ng ] ) , and the following results hold for any @xmath81 , not necessarily limited to integers . in the following the unit mass and unit length are the total mass @xmath84 and the scale length @xmath87 of the model , with @xmath112 . in c99 the analytical expression for the critical @xmath113 was obtained for the whole family of @xmath1-models , and here we derive its generalization @xmath114 .
in fact , eq . ( a1 ) shows that for @xmath115 and @xmath81 the nc is satisfied for @xmath116 , and the result of c99 is now obtained as the particular case for @xmath89 . in other words ,
the nc leaves open the possibility of making models with @xmath117 using radial orbits only . in the range
@xmath118 the nc requires instead that @xmath119 where the explicit expression of @xmath120 is given by eq .
( a2 ) , and the inequality above reduces to eq . ( 13 ) of c99 when @xmath89 .
the nc then proves that models with @xmath118 can not be made of radial orbits only , independently of the value of the external slope @xmath2 , and of the presence of any possible second component .
this result extends the list of cases for which it has been proved that a density cusp shallower than @xmath121 can not be supported by radial orbits only ( richstone & tremaine 1984 , cp92 , c99 , ae06 ) . as a function of @xmath1 in one component models , for @xmath122 .
the nc limit is represented by the solid lines : all models with the pair ( @xmath123,@xmath1 ) in the triangular region below them are inconsistent .
the dashed lines mark the wsc limit : all points above these lines correspond to consistent one
component models . for @xmath3 ,
the filled circles joined by the dotted line represent the more accurate limits obtained from the ssc . for the black - hole dominated ( @xmath2-@xmath1,bh ) models the long - dashed lines ( interrupted for @xmath124 , see section 3.2 ) , are the lower limit on @xmath123 given by the wsc , while the solid squares are the true limits derived from the df.,scaledwidth=50.0% ] in fig .
[ f2 ] the lower bound for the anisotropy radius given in eq .
( [ eq : ngnc ] ) is shown with the solid lines as a function of @xmath1 for @xmath122 : models ( one or multi component ) with the pair ( @xmath123,@xmath1 ) in the nearly triangular region under the solid curves are inconsistent .
in particular , for fixed @xmath1 an increase of @xmath2 produces a decrease of the minimum @xmath123 , i.e. we have here a first indication that a steepening of the density profile in the external regions of a system with fixed inner density slope can be effective in increasing the maximum amount of sustainable radial anisotropy .
this behaviour is quantified by substitution in eq .
( [ eq : ngnc ] ) of the asymptotic expansion for @xmath125 of @xmath120 in eq .
( a2 ) : @xmath126 as often happens in asymptotic analysis , even if the expansion above holds in principle only for very large values of @xmath2 , the substitution of ( [ eq : smas ] ) in ( [ eq : ngnc ] ) leads to percentual errors on @xmath123 less than 22% , 14% , and 10% for @xmath122 respectively ( for the inner density slope @xmath127 ) .
we now move to discuss the wsc for one component models : the obtained @xmath128 will mark a lower limit above which consistency ( for the considered @xmath2 and @xmath1 ) is guaranteed . unfortunately , the wsc can not be explored algebraically in the general case , because the resulting inequality ( that for simplicity we do not report here ) involves the solution of an equation of fifth degree for @xmath129 , and the degree increases for increasing @xmath2 . for @xmath89 , instead
, it is possible to treat the wsc analytically , since it reduces to the discussion of a cubic equation ( see c99 ) .
of course , the critical values for the anisotropy radius can be easily obtained solving numerically inequality ( [ eq : wsc ] ) in any specific case of interest , and the results are shown in fig .
[ f2 ] for @xmath122 with dashed lines : all one component models with the pair ( @xmath123,@xmath1 ) in the region above the dashed lines are consistent .
note how for increasing values of the external density slope @xmath2 more and more radially anisotropic orbital distributions can be supported , thus confirming the indications provided by the nc .
values of @xmath123 nearer to the limits set by the df are obtained by using the ssc .
inequality ( [ eq : ssc ] ) evaluated for a generic pair ( @xmath1,@xmath2 ) results in a transcendental equation that must be studied numerically ( as already done for @xmath89 in c99 ) , and the black dots joined by the dotted lines in fig .
[ f2 ] represent the critical lower values of @xmath123 for one component @xmath1 models with @xmath3 , and @xmath122 .
as expected , the dotted lines are contained between the solid ( nc ) and the dashed ( wsc ) lines in each panel , and again they shift downward for increasing @xmath2 .
thus , from this preliminary investigation of one component models we conclude that for fixed @xmath1 an increase of @xmath2 corresponds to a decrease of the minimum admissible value of @xmath123 , i.e. _ steeper density distributions in the external regions can support more radial anisotropy_. in order to proceed further with the preliminary discussion , we now apply the wsc to the component of a ( , @xmath0 ) model , extending to this class of systems some of the results obtained in c96 and c99 for two
component @xmath8 models .
in particular , the analytical study in c99 was restricted to some representative @xmath8 models , namely @xmath130 isotropic two - component systems with inner slopes in the ranges @xmath131 , and @xmath10 ( i.e. the @xmath9 component is shallower in the central regions ) ; @xmath132 isotropic two component ( 0,1 ) systems ( i.e. the @xmath9 component is steeper in the central regions ) ; and finally @xmath133 anisotropic @xmath0 profiles with @xmath131 in the gravitational field of a dominant central black hole . here , in order to obtain analytical results for the more general ( @xmath25,@xmath0 ) models , we assume that the @xmath0 component ( the `` halo '' ) dominates everywhere the gravitational field . under this simplifying assumption , the following three results , corresponding to the points @xmath130 , @xmath132 , and @xmath133 above , will be proved analytically : 1 . in the case of halo dominated _ isotropic _ ( , @xmath0 ) models , with @xmath134 , @xmath135 , and @xmath81 , the centrally more peaked component is consistent , independently of the value @xmath136 of its concentration relative to the centrally flatter @xmath0 halo . in the case of _ anisotropic _ ( , @xmath0 ) models , the determination of a minimum anisotropy radius for consistency as a function of @xmath2,@xmath1,@xmath0,@xmath137 reduces to the solution of an algebraic equation of sixth degree ( for generic @xmath2 ) , which is solved numerically . in the particular case of halo dominated ( @xmath2 - 2,1 ) models , the application of the wsc shows that for @xmath138 these models can be consistently assembled using radial orbits only . 2 . in the case of halo
dominated _ isotropic _ ( @xmath2 - 0,1 ) models , the wsc shows that for @xmath139 the @xmath2 - 0 density distribution is consistent . for broader @xmath2 - 0 density distributions ,
the models can be consistent only in presence of some amount of radial anisotropy .
3 . in the case of _ anisotropic _ @xmath25 models with a dominant black hole at their center it is possible to determine analytically a lower limit @xmath140 for consistency , and this limit decreases for increasing @xmath2 . ) as a function of the relative concentration @xmath137 , for consistency of halo dominated ( @xmath25,1 ) models .
the external density slope @xmath2 increases as @xmath93 , and @xmath94 from left to right panels , while the internal density slope as @xmath141 , and @xmath142 from top to bottom panels .
solid lines represent the limits derived from the wsc , while dashed lines are the @xmath143 derived from the df .
note that for ( @xmath2 - 0,1 ) models also the @xmath144 limits appear for @xmath145 ( see sections 3.2 and 4.3 for details).,scaledwidth=50.0% ] the proof of the first result , which is an extension of the study mentioned in point @xmath130 , is conceptually straightforward but algebraically cumbersome , and only the main steps are reported in appendix a ( eq . [ a3 ] ) .
in particular , as this result holds also for @xmath81 , while the external logarithmic slope of the @xmath0 halo is @xmath146 , it means that a centrally steep density profile can be consistent in the gravitational field of a more massive ( but centrally shallower ) component , even when its external regions are less peaked .
the situation is illustrated for some representative cases in fig .
[ f3 ] ( bottom and middle panels ) , where all points above the solid lines correspond to consistent halo dominated ( @xmath2 - 1,1 ) and ( @xmath2 - 2,1 ) models , and apparently the isotropic limit ( @xmath147 ) is allowed for any choice of @xmath137 ; the solid lines are obtained by solving numerically the corresponding wsc .
a few additional trends are apparent .
first , for a given central logarithmic density slope @xmath1 and for a given relative concentration @xmath137 , an increase of @xmath2 corresponds to a better and better ability to sustain radial anisotropy , and thus the trend found in one component models is confirmed also in the two component case .
second , at fixed @xmath2 and @xmath1 , the minimum anisotropy radius for consistency increases for increasing @xmath137 .
this trend was already found in c96 and c99 : in practice , in a fixed potential broader density distributions are less and less able to sustain radial anisotropy .
third , at fixed external slope @xmath2 and relative concentration @xmath137 , more centrally peaked systems are better able to support radial anisotropy .
an additional comment concerns the specific case of ( @xmath2 - 2,1 ) models ( fig .
[ f3 ] , bottom panels ) .
in fact , it is apparent how in presence of a dominant @xmath148 halo the wsc limits flatten to zero for relative concentration less than some critical value @xmath149 , and accordingly the model may be purely radially anisotropic . remarkably , it is easy to show that the @xmath2 - 2 component can support purely radial orbits for relative concentrations @xmath150 ( see eq . [ a4 ] ) , which are exactly the critical points in fig .
[ f3 ] . of course
, this is just a sufficient condition , so we expect that the existence of a larger critical concentration for consistency in case of purely radial orbits will be revealed by direct inspection of the df .
a final comment is due .
in fact , the adopted range of values for the internal density slopes @xmath151 imply that the halo density distribution is centrally less peaked than the @xmath152 component . therefore , for large but _ finite _ total halo mass
, the integrated mass of the halo is subdominant with respect to the stellar one for vanishing @xmath37 , and the assumption of a dominant halo breaks down at the very center .
this means that in an asymptotic sense , the analysis presented in the previous section should be applied at the center of these finite - halo mass models .
result ( ii ) extends the study mentioned in point @xmath132 above , and it is proved in eq .
the obtained limitation @xmath153 , shown by the solid line in fig .
[ f4 ] , suggests that the concentration of the @xmath154 component must `` adapt '' to the dominant @xmath148 halo for phase space consistency .
note that @xmath149 increases for increasing @xmath2 , i.e. a steeper external density profile can ( partially ) compensate for the effect of a central shallower density distribution .
the appearence of the concentration limit @xmath149 in the isotropic case manifests in the top panels of fig .
in fact , as discussed in section 2 , in systems with positive isotropic functions ( wsc or df ) only a lower limit for @xmath123 exists , as in the case of halo dominated ( @xmath2 - 1,1 ) and ( @xmath2 - 2,1 ) models . in the present case , instead , when for increasing @xmath137 the critical value @xmath155 is crossed ( in the @xmath89 panel )
, the additional condition @xmath156 corresponding to the radial domain @xmath69 appears as the vertical asymptote of the solid line , and the parameter space for consistency progressively reduces , shrinking to zero for the value of @xmath137 where the two solid lines ( obtained numerically ) cross . a similar configuration repeats in the panels representing the @xmath129 and @xmath157 cases , where the consistency limits move , as expected , downward and rightward .
of course , these limits represent only a sufficient condition for the consistency , and when considering the df in the limit @xmath158 ( see section 4 ) we expect to determine larger critical values of @xmath149 , and larger consistency regions in the ( @xmath137,@xmath123 ) plane .
the result concerning the models with central black hole , related to point @xmath133 above , is proved as follows .
as well known , for @xmath159 the potential of the @xmath148 model becomes that of a point mass , and from the first result proven in this section it follows that a black hole of any mass @xmath160 can be consistently added at the center of a globally isotropic model , when @xmath161 and @xmath81 .
for this reason in the following we restrict to this range of slopes @xmath1 , and accordingly the long dashed curves in fig .
[ f2 ] interrupt for @xmath124 .
having reduced the study to models with a positive isotropic df , we only have to estimate a lower limit of @xmath110 . in appendix a the wsc is applied to the anisotropic ( , bh ) models with @xmath134 ; it can be discussed analytically in the special case of a _
dominant _ black hole , i.e. assuming in eq .
( [ eq : wsc ] ) @xmath162 ( and so @xmath163 ) . as shown in eq .
( a6 ) , under these assumptions @xmath164 } { n(n-1)\sm^2 + 2n(\gamma-1)\sm+\gamma(\gamma-1)},\ ] ] where @xmath165 is obtained by solving an algebraic equation of fourth degree : the obtained limits are shown in fig .
[ f2 ] with the long dashed lines , for @xmath122 . for
@xmath89 equation above coincide with that given in eq .
( 18 ) of c99 , while asymptotic analysis proves that for @xmath125 @xmath166 where @xmath167 is the larger real root of eq .
we found that the substitution of eq .
( [ eq : smbhas ] ) in eq .
( [ eq : sabh ] ) leads to estimates of the minimum @xmath123 discrepant from the values obtained by the full formula by less than 50% , 33% , and 23% for @xmath122 respectively ( for the inner density slope @xmath127 ) . as in the other cases , an increase of the external density slope
@xmath2 makes it possible to sustain more radial orbits .
it should be clear that the df of two component ( , @xmath0 ) models can not be obtained analytically , except for very special combinations of the values of @xmath168 and @xmath0 , as for example the ( 1,1 ) models discussed in c96 ( that would be referred as ( @xmath146-@xmath169,@xmath169 ) models in the present notation ) .
however , the search is not hopeless .
in fact , from c99 it is known that the halo dominated df of two component ( 1,0 ) and ( 0,1 ) galaxy models can be obtained in terms of elementary functions , and a simple argument shows that this is also the case of ( 2,1 ) and ( 2,0 ) models .
this fact , the linearity of the om inversion formula with respect to @xmath5 in the non self gravitating case ( see eq .
[ 2 ] ) , and finally the formula ( [ eq : ng ] ) for @xmath2 integer , prove that also the df of the halo dominated ( , 1 ) and ( , 0 ) models can be expressed in terms of elementary functions , for any integer @xmath88 and @xmath3 . note that , albeit their special nature , the study of halo dominated models is useful because the formulae
expressible using elementary functions can be studied very easily , making apparent the effects of the relative distribution of the investigated component and of the halo . in particular , in this section we will determine the df of the component of halo dominated ( , 1 ) models , and the exact phase
space constraints will be derived and compared with those obtained using the nc , wsc , and ssc in section 3 .
we restrict to the @xmath148 case because the hernquist ( 1990 ) potential is the simplest in the class of @xmath1 models ; furthermore , this choice allows us to investigate the consistency of models with central density slopes flatter ( @xmath154 ) , equal ( @xmath127 ) , and steeper ( @xmath170 ) than that of the halo . as a consequence
, the results obtained for ( , 1 ) models should be representative of the whole situation . with the normalization scales introduced at the beginning of section 3 ,
the relative potential of the hernquist halo needed for the recovery of the df is @xmath171 while from eq .
( [ eq : ng ] ) @xmath172 as for the density and the potential , also for the df it is useful to work with dimensionless functions , and we define @xmath173 with @xmath174 , and @xmath175 : note that , at variance with c99 , the normalization quantities used here are those of the halo , consistently with the halo dominated nature of the present models .
in other words , the dominant density distribution sets the natural scales . the easiest way to compute the df is to change the integration variable from the total potential to the radius in the first of the identities in eq .
( [ eq : fom ] ) .
after normalization we obtain @xmath176 where @xmath177 and from eq .
( [ eq : her ] ) @xmath178 in the equations above , a negative sign appears in front of the integral ( [ eq : fnu ] ) due to the monotonic decrease of the relative potential with increasing radius ; @xmath179 is the normalized augmented density associated with eq .
( [ eq : rhonor ] ) , while @xmath180 is the normalized anisotropy radius ; finally , the subscripts of the functions @xmath181 refer to the isotropic and anisotropic parts of the df , respectively .
it is now evident how in halo dominated ( @xmath25,@xmath0 ) models the external slope parameter appears in the density @xmath179 only , so that the df of the component can be obtained by repeated differentiation with respect to @xmath137 of the simpler df of the halo dominated ( @xmath1,@xmath0 ) model . in the following , we will also discuss black hole dominated ( @xmath25,bh ) models : then , we will adopt as natural normalization scales the black hole mass @xmath160 and the scale
length @xmath87 of the component ; moreover , @xmath182 , since in this case @xmath183 , so that in eq .
( [ eq : qnu ] ) @xmath184 .
it is not difficult to show that , for generic values of @xmath1 , the integral ( [ eq : fnu ] ) can be expressed as a combination of hypergeometric @xmath99 functions when the halo is a @xmath12 , or @xmath169 model , or a black hole .
however , the resulting expressions are not more illuminating than the integral itself , so that we do not report them here . instead , we prefer to show simple asymptotic expansions relative to a couple of interesting situations .
the first concerns the behaviour of the df of ( @xmath25,bh ) models for @xmath185 , for which the leading term of the expansion ( normalized to @xmath186 ) is @xmath187 thus , the slope @xmath188 marks the different behaviour of the models .
we note also that @xmath189 for @xmath190 and the model is inconsistent ( tremaine et al .
1994 ) ; furthermore , we note how in the leading term above the function @xmath191 is producing no contribution .
the other case of interest is the behaviour of the df of isotropic models for @xmath192 .
we recall that this case is very general , as it applies to self - gravitating models but also to models in generic halos of finite total mass , because in all systems @xmath193 for @xmath194 . in this case
we found @xmath195 an expression ( obviously ) formally identical to eq .
( [ asymbh ] ) .
finally , we notice that the asymptotic expansion for @xmath196 of the velocity dispersion of isotropic models ( both single component and embedded in dark matter halos of finite total mass ) is @xmath197 , normalized to @xmath198 , so that for @xmath196 @xmath199 , in accordance with the comment in footnote 1 .
following the preliminary discussion , here we derive the explicit expression for the df of the @xmath2 - 2 model with an arbitrary degree of om orbital anisotropy , in the gravitational field of a dominant hernquist halo .
after partial fraction decomposition of eq .
( [ eq : fnu ] ) , the isotropic and anisotropic components of the df are @xmath200 and @xmath201 respectively , where for @xmath202 we define @xmath203 for @xmath204 the function @xmath205 assume a very simple form : @xmath206 while for @xmath207 the formula @xmath208 holds , with @xmath209 note that the second expression in eq .
( [ dfang2 ] ) has been obtained from the relation ( [ recg ] ) . of the fully isotropic halo - dominated ( @xmath2 - 0,1 ) model , as a function of the external density slope @xmath2 .
the solid line is the limit derived from the wsc ( section 3.2 ) , while the dotted line is the true limit obtained from the df ( section 4.3).,scaledwidth=40.0% ] -2,1 ) , ( @xmath2 - 1,1 ) , and ( @xmath2 - 0,1 ) models , as a function of @xmath210 .
the solid lines represent the isotropic dfs , while the dotted lines are the anisotropic dfs , with @xmath123 approaching the critical value for consistency . in all cases ,
@xmath211.,scaledwidth=55.0% ] from the derived formulae we determined the true anisotropy limit for the ( @xmath2 - 2,1 ) models .
we found that @xmath67 for all the explored values of @xmath2 , i.e. the isotropic @xmath2 - 2 model in a dominant hernquist halo is consistent independently of the halo scale length , in accordance with the result in point @xmath212 of section 3.2 , obtained with the wsc .
therefore the set @xmath64 in phase
space coincides with the whole accessible phase space , and only the anisotropy limit @xmath143 exists . in fig .
[ f5 ] ( top panel ) we show the isotropic ( solid ) and strongly anisotropic ( i.e. , near to the consistency limit ) dfs for @xmath213 and @xmath211 .
note how the isotropic dfs are very similar , and monotonically decreasing .
instead , in the anisotropic cases the depression leading to inconsistency is apparent , clearly showing how phase
space inconsistency is always set outside the center .
this feature is qualitatively similar to the others explored om systems ( e.g. , see fig . 2 in ciotti & lanzoni 1997 , figs . 2 and 3 in c99 , fig
. 3 in ciotti , morganti & de zeeuw 2008 ) ;
in addition , the systematic shift of the df depression towards high @xmath111 values for increasing @xmath2 is also apparent : an argument supporting this phenomenon is given in section 2.1.1 .
we notice that the cuspy dips shown by the dfs for values of @xmath123 near the consistency limit could be the source of kinetic instabilities , whose investigation is of course well beyond the framework of this paper . in any case , it is almost certain that any system would develop radial orbit instability for values of the anisotropy radius larger than the consistency limit ( e.g. , see the n - body experiments discussed in nipoti , londrillo & ciotti 2002 ) : therefore , the critical dfs in fig .
5 describe equilibrium unstable systems .
the trend of the anisotropy limit @xmath143 for different values of @xmath2 and increasing @xmath137 is shown in fig .
[ f3 ] ( bottom panel ) with the dashed lines .
first , for fixed @xmath2 and increasing @xmath137 , @xmath143 increases : in practice , for a given halo , a broader stellar density distribution is less and less able to sustain radial anisotropy .
second , at fixed @xmath137 a steeper external density slope ( i.e. larger @xmath2 ) corresponds to higher amount of admissible radial anisotropy , thus showing that not only the inner density slope , but also the external density profile is important for phase space consistency . as expected , the true limitation on @xmath143 is less stringent than the corresponding limit derived from the wsc , and the dashed line is everywhere below the solid line . in particular , while from the wsc the critical concentrations @xmath149 in order to have a purely radial model are @xmath214 and @xmath215 ( for @xmath216 and @xmath94 ) , from the df we obtain @xmath217 and @xmath218 .
it is also possible to determine the df of ( @xmath2 - 2,bh ) models in presence of a dominant central black hole .
the resulting formulae are identical to eqs .
( [ dfisg2])-([dfang2 ] ) , where the quantity @xmath219 at numerator is replaced by @xmath220 and the @xmath221 functions are replaced by the corresponding @xmath222 functions of same index , defined for @xmath223 as @xmath224 and @xmath225 for @xmath226 the formula @xmath227 holds , where @xmath228 -2,1 ) and ( @xmath2 - 1,1 ) models , as a function of the relative energy @xmath210 , for @xmath216 and @xmath94 .
normalization scales are the mass of the black hole and the scale length of the component .
solid lines represent the isotropic dfs , while dotted lines the anisotropic dfs , with @xmath123 approaching the critical value for consistency.,scaledwidth=52.0% ] note that after performing the required differentiations one must set @xmath204 . in fig .
[ f6 ] ( upper panel ) the obtained dfs are shown in the isotropic ( solid lines ) and strongly anisotropic ( dotted lines ) cases .
again , inconsistency due to om anisotropy affects the df at intermediate energies .
the true anisotropy limits derived from the df are given by solid squares in fig .
[ f2 ] . finally , note also the characteristic log - log linear trend of the df for very high values of @xmath111 .
this trend is easily explained by considering the asymptotic expansion given in eq .
( [ asymbh ] ) .
we now derive the explicit df for the family of halo dominated ( @xmath2 - 1,1 ) models , so that with this class of models we can study the effect of the external logarithmic slope when the two density profiles have the same inner slope .
of course , the ( 4 - 1,1 ) models are the halo dominated of ( 1,1 ) models discussed in c96 .
integration of eq .
( [ eq : fnu ] ) reveals that the functions involved in the expression of the df are the same as for the ( @xmath2 - 2,1 ) models .
in particular , @xmath229 and @xmath230 in fig .
[ f5 ] ( middle panel ) , the df in the isotropic and strongly anisotropic cases is shown , for @xmath122 .
the behaviour is qualitatively similar to that of ( @xmath2 - 2,1 ) models , i.e. while the isotropic systems display a monotonically decreasing df , in the anisotropic case ( when @xmath143 approaches the consistency limit ) the df has a significant depression , that moves inwards for increasing @xmath2 . in fig .
[ f3 ] ( middle row ) we show the minimum value of the anisotropy radius normalized to @xmath87 as derived from the df ( dashed lines ) , and from the wsc . again , in accordance with the preliminary study of section 3.2 , we found that @xmath67 independently of the relative concentration value @xmath137 and of the external slope @xmath2 , so that only the limit @xmath143 exists .
all the trends exhibited by the ( @xmath2 - 2,1 ) models are confirmed : in particular , @xmath143 increases with @xmath137 for fixed @xmath2 , while it decreases at fixed @xmath137 for increasing @xmath2 .
note that the maximum radial anisotropy admissible for given @xmath137 and @xmath2 is smaller than for ( @xmath2 - 2,1 ) models , as expected from the shallower central profile of @xmath2 - 1 models .
finally , at variance with the ( @xmath2 - 2,1 ) models ( and in agreement with the expectations of the preliminary analysis ) no purely radial orbital configurations can be supported , independently of the value of the concentration @xmath137 . as for ( @xmath2 - 2,bh )
models , also in the case of the black hole dominated ( @xmath2 - 1,bh ) models the functions @xmath231 and @xmath191 are formally identical to those in eqs .
( [ fisg1])-([fang1 ] ) , with each @xmath221 function replaced by the corresponding @xmath222 function , and with the substitution @xmath232 ; again , it is necessary to set @xmath204 after performing the required differentiations . while the true limits on the anisotropy radius are plotted in fig .
[ f2 ] with the solid squares , the obtained dfs are shown in the isotropic ( solid ) and strongly anisotropic ( dotted ) cases in the bottom panel of fig .
[ f6 ] ; not unexpectedly , the inconsistency due to anisotropy manifests itself at intermediate energies . finally , the behaviour of the df for very high values of @xmath111 is in accordance with the asymptotic expansion of the df in eq .
( [ asymbh ] ) .
this case is expected to be the more complicated , because from c99 it is known that the ( 0,1 ) model presents a peculiar behaviour : in practice , there exists a range of @xmath137 where anisotropic models can be consistent when instead isotropic models are not , and this is due to the fact that @xmath62 becomes negative over the non empty set @xmath69 , so that two limits on @xmath123 must be considered .
note that the discussion of the wsc in the case of ( @xmath2 - 0,1 ) models in section 3.2 already showed a similar behaviour . as in the other cases , the integration of eq .
( [ eq : fnu ] ) can be done by using the @xmath205 functions only : @xmath233 @xmath234 where the second expression for @xmath231 is obtained from the relation ( [ recg ] ) of the @xmath235 functions .
motivated by the remarks above , we start to study the sign of @xmath62 , in order to determine the condition for the existence of the sets @xmath64 and @xmath69 .
indeed , we found that for @xmath236 ( in the @xmath89 case , see c99 ) , @xmath237 ( for @xmath129 ) , and @xmath238 ( for @xmath157 ) , @xmath62 is positive everywhere over the accessible phase space , a situation which is graphically represented with the dotted lines in fig .
it follows that for sufficiently concentrated models only the anisotropy limit @xmath143 exists , and in fig .
[ f5 ] ( bottom panel , solid lines ) the isotropic df is shown for ( @xmath2 - 0,1 ) models with @xmath211 and @xmath213 : note how , at variance with the other cases , the df of ( @xmath2 - 0,1 ) models decreases for increasing @xmath111 .
the trend of @xmath239 is shown in fig .
[ f3 ] ( top panels , dashed lines ) : as for the models previously discussed , the @xmath143 curve is similar ( but displaced ) with respect to that derived from the wsc .
it is apparent how the @xmath143 line lies above those of @xmath2 - 1 and @xmath2 - 2 models , as expected from the shallower density profile of @xmath2 - 0 models ; moreover , we note how the minimum admissible values for @xmath143 decrease for increasing @xmath2 , as in the other cases . in fig .
[ f5 ] the dfs for @xmath123 near the consistency limit @xmath143 are shown with dotted lines for models having @xmath211 , and as in the previous @xmath127 and @xmath170 models , inconsistency appears at intermediate values of @xmath210 . for values of @xmath137 larger than the critical values of @xmath149 reported above
, the set @xmath69 is not empty , because the isotropic component of the df becomes negative at high relative energies .
in fact , at variance with the ( @xmath2 - 2,1 ) and ( @xmath2 - 1,1 ) models , the df of the @xmath2 - 0 model converges to a finite limit ( that can be easily calculated analytically from eq . [ [ fig0 ] ] ) for @xmath240 ; when increasing @xmath241 , the value @xmath242 monotonically decreases , till it becomes negative for @xmath137 greater than @xmath149 .
a similar behavior was found in the numerical investigation of consistency of king ( 1972 ) and quasi
isothermal halos added to a de vaucouleurs ( 1948 ) density distribution ( see cp92 ) .
thus , while the effects of om radial anisotropy appear to affect the df at intermediate energies , the inconsistency of isotropic models embedded in a peaked halo seems to characterize the df at very high ( relative ) energies . according to the discussion of section 2 ,
since the function @xmath74 is positive on @xmath69 , the upper bound @xmath144 can be determined : the appearence of this new value is manifested by the dotted lines in fig .
[ f3 ] , with the vertical asymptote corresponding to @xmath149 .
then , for @xmath137 greater than the critical limit , a range of concentrations exists where an anisotropy radius @xmath243 corresponds to consistent models ; finally , for @xmath137 larger than the value where the dotted ( @xmath144 ) and dashed ( @xmath143 ) lines cross each other , @xmath244 , and the models become inconsistent . summarizing , consistent models correspond to points placed above the @xmath143 line and on the left of the @xmath144 line . in the case of a central dominant black hole , the same comments as those for ( @xmath2 - 2,bh ) and ( @xmath2 - 1,bh )
models apply , and the functions @xmath231 and @xmath191 are obtained accordingly . of course ,
the isotropic model is inconsistent , independently of the values of @xmath2 , because the model with black hole would correspond to the formal limit @xmath245 in ( @xmath2 - 0,1 ) models , which is already excluded by the previous discussion .
what happens when considering the effect of anisotropy ?
is there any compensating effect ?
a numerical inspection shows that the anisotropic part of the df is negative on @xmath69 , so that not even orbital anisotropy can make the models consistent .
again , this result was expected from the study of ( @xmath2 - 0,1 ) models , due to the crossing of the @xmath143 and @xmath144 critical lines at finite values of @xmath137 .
in this paper , in a natural extension of previous investigations ( cp92 , c96 , c99 ) , we focused on the importance of the _ external _ density logarithmic slope on the determination of the phase space consistency of one and two component stellar systems with om orbital anisotropy .
in particular , the considered ( @xmath25,@xmath0 ) models generalize ( @xmath0,@xmath9 ) models , as the density component is characterized by logarithmic slope @xmath2 instead of @xmath146 outside the core radius , while in the inner regions the logarithmic density slope is @xmath1 .
the main results can be summarized as follows : 1 .
it is shown that , for @xmath2 integer , several structural and dynamical properties of the models can be obtained by differentiation , with respect to the scale length , of the corresponding formulae valid for the standard @xmath1 models . for example , it is shown how to construct explicit solutions of the jeans equations for the class of two component anisotropic ( @xmath26-@xmath0,@xmath27-@xmath9 ) models once the solution is known for ( @xmath0,@xmath9 ) models . 2 . in one - component models , a lower limit for the anisotropy radius ,
so that smaller values certainly produce inconsistent models , is analytically derived following the technique introduced in cp92 .
it is found that for @xmath246 this minimum anisotropy radius is strictly positive , but it decreases by increasing @xmath2 , i.e. a larger amount of radial anisotropy can be supported by externally steeper density profiles . for @xmath247 instead
the necessary condition for consistency is satisfied @xmath248 , independently of the value of @xmath1 and @xmath2 . the minimum anisotropy radius so that larger values correspond to consistent models
is then determined by using the strong and weak sufficient conditions for model consistency derived in cp92 . as for the necessary condition
, we find that the minimum value of the anisotropy radius decreases for increasing @xmath1 and fixed @xmath2 , and for increasing @xmath2 and fixed @xmath1 .
3 . a similar analysis
is then performed for two - component ( @xmath25,@xmath0 ) systems , in order to extend the study of c99 . for simplicity
we restrict to halo dominated models , i.e. we take into account only the gravitational field of the @xmath0 halo .
in particular , we show that in the isotropic case , when @xmath161 and @xmath135 ( i.e. , the halo density is centrally less peaked than the stellar component ) , the df of the component is nowhere negative , independently of the mass and concentration of the @xmath0 halo .
this is true even when in the external parts the component is less peaked @xmath249 than the @xmath0 component . as a special application of this result
, it follows that a black hole of any mass can be consistently added at the center of a globally isotropic model , when @xmath250 .
moreover , in the case of ( @xmath2 - 0,1 ) models , where instead the halo is centrally steeper than the stellar component , the sufficient condition applied to the @xmath2 - 0 density distribution reveals the existence of an upper limit @xmath251 of the relative concentration @xmath241 for model consistency in the isotropic case .
finally , in the case of anisotropic models with a dominant black hole at their center , we determined analytically a lower limit of the critical anisotropy radius for consistency as a function of @xmath2 and @xmath1 ; this value decreases for increasing @xmath1 and/or @xmath2 .
4 . the analytic expression for the df of om anisotropic halo dominated ( @xmath25,1 ) models , and (
, bh ) models with a dominant central bh is recovered in terms of elementary functions in the special cases when @xmath252 .
it is found that , while for @xmath253 the isotropic df is positive independently of the model concentration , the @xmath154 component becomes inconsistent when the halo is sufficiently concentrated , even in the isotropic case .
in addition , the trend of the minimum value of the anisotropy radius as a function of the halo concentration confirms that steeper density profiles in the external region are consistent with a larger amount of radial anisotropy . in conclusion , the explored family of models provide a direct indication that the density slope in the external regions of stellar systems can be important in determining the admissible radial anisotropy , and this in addition to the well known relevance of the central density slope .
a quantification of this argument is embodied in a _ necessary _ inequality for model consistency , that we derived by expressing a previous finding ( cp92 ) in terms of the logarithmic density slope .
the obtained result holds for _ all _ consistent om anisotropic systems , and relates the logarithmic density slope and the om anisotropy indicator at _ each radius _ , in a way formally identical to the `` cusp slope - central anisotropy '' theorem by ae06 , which is known to apply at the center of stellar systems with generic anisotropy distribution , and everywhere in constant anisotropy systems .
we wish to thank the anonymous referee for helpful suggestions that greatly improved the presentation of the paper .
simple algebra shows that the nc applied to anisotropic models can be written as @xmath254 for @xmath255 and @xmath256 the r.h.s .
is everywhere negative , so that inequality ( a1 ) is trivially satisfied for @xmath257 . instead ,
for @xmath246 the parameter @xmath123 must be larger than the maximum of the expression of the r.h.s .
, which is reached at @xmath258}}{2n(n-2)},\ ] ] that after substitution in eq .
( a1 ) gives eq .
( [ eq : ngnc ] ) . of course , for @xmath89 eq .
( a2 ) in c99 is reobtained . here
we apply the wsc to the component of globally isotropic , halo dominated ( , @xmath0 ) models , where @xmath134 , @xmath135 , and @xmath81 . under the assumption of a dominant halo , eq .
( [ eq : wsc ] ) reduces to investigate the positivity of an expression which factorizes in a strictly positive function and in an algebraic factor that after the natural substitutions @xmath259 and @xmath260 ( with @xmath261 and @xmath262 ) becomes @xmath263s^2\\ & & + \xi\{[(n+1)(2+\epsilon+\epu ) + \epsilon\epu-\epsilon^2]+\xi\epsilon\gamma\}s+ \xi^2\gamma(2+\epu),\nonumber\end{aligned}\ ] ] whose positivity @xmath264 is easily proved .
therefore the wsc is verified for any value of the concentration parameter @xmath137 .
note that the variables @xmath137 and @xmath265 adopted here correspond to @xmath266 and @xmath267 respectively in eq .
( a13 ) of c99 , due to the different normalization length .
being the wsc of the isotropic models satisfied , the existence of a critical anisotropy radius @xmath143 for consistency depends on the sign of the anisotropic part of the wsc , as discussed in section 2.2 . in particular
, the positivity of the anisotropic wsc for ( @xmath2 - 2,1 ) models reads @xmath268 and thus shows the existence of a critical value @xmath269 of the concentration parameter , marking an upper limit on the values of @xmath137 for which @xmath270 , i.e. purely radial orbital distributions are allowed .
the discussion of the wsc for globally isotropic @xmath2 - 0 models in a dominant @xmath127 halo is simple , reducing to the request that @xmath271 which is satisfied for @xmath153 . for @xmath89
we reobtain the result of c99 . in the case of central dominant bh
, we assume in eq .
( [ eq : wsc ] ) @xmath272 , and from the previous discussion we restrict to models with @xmath134 .
the wsc then requires that @xmath273 after the differentiation of the r.h.s . and the successive study of a quartic equation
, it can be proved that for @xmath274 eq .
( a6 ) admits only one maximum , located at @xmath275 . the general expression for @xmath120 is not reported here , but it was used to produce the long dashed lines in fig .
[ f2 ] . in any case ,
@xmath276 , @xmath277{s_0}}{3n(n-1)(n-3)}- \frac{2(n-4)}{3(n-1)(n-3)}}\cr + \displaystyle{\frac{4(n^3+n^2 - 20n+27)}{3 ( n-1)(n-3)\sqrt[3]{s_0 } } } , & \gamma=2,\cr \end{cases}\ ] ] where @xmath278,\end{aligned}\ ] ] and for @xmath89 and @xmath279 the values given in table 1 of c99 are reobtained .
we finally report the reduced asymptotic fourth order equation needed to determine @xmath120 in the limit @xmath125 : remarkably , after the scaling @xmath280 and the successive limit @xmath125 , one obtains @xmath281
here we present an easy way to express analytically the main dynamical quantities of the two component om anisotropic ( @xmath26-@xmath0,@xmath27-@xmath9 ) models for generic @xmath0 and @xmath9 , and @xmath26 and @xmath27 integer @xmath282 .
the method is based on the evaluation of the similar quantities for the two components @xmath8 models , and then on repeated differentiation with respect to the scale lengths .
we start by considering the radial component @xmath50 of the velocity dispersion @xmath283 of a system with a density component @xmath5 , which in the om parameterization can be written as @xmath284 where @xmath285 ( binney & mamon 1982 ) , and @xmath286 is the total mass within @xmath37 .
once @xmath50 is known , the tangential velocity dispersion is obtained from eq .
( [ betaom ] ) .
as the method is general , we focus without loss of generality on the component `` 1 '' of a two component model , for which the two functions @xmath287 and @xmath288 are given by the sum @xmath289 and @xmath290 , due to the linearity of mass in eqs .
( b2 ) , and where the meaning of the subscript indices is apparent .
other quantities of interest in applications are the global energies entering the scalar virial theorem , and for the component 1 the scalar virial theorem reads @xmath291 , where @xmath292 is the contribution due to the self interaction , and @xmath293 is the interaction energy with the `` halo '' ( e.g. , ciotti 2000 ) . the basic idea is to compute the integrals for the generic pair ( @xmath294 ) relative to standard two
component @xmath8 models with different scale lengths @xmath295 and @xmath296 , to perform the required differentiations as prescribed by eqs .
( [ eq : ng ] ) and ( [ eq : mng ] ) , and finally ( in the case of self interactions ) to set the two scale lengths and the two slopes to the same value . from eqs .
( [ eq : ng ] ) and ( [ eq : mng ] ) it follows @xmath297 for generic values of @xmath0 and @xmath9 the integrals above involve hypergeometric @xmath99 functions , while for integer values of @xmath0 and @xmath9 the resulting formulae can be expressed in terms of elementary functions ( e.g. , see c96 and c99 for ( 1,0 ) and ( 1,1 ) models ; ciotti et al .
1996 for ( 2,1 ) and ( 2,2 ) models ) .
even simpler expressions can be obtained without difficulty for to the halo dominated ( @xmath2 - 0,1 ) , ( @xmath2 - 1,1 ) , and ( @xmath2 - 2,1 ) models , by adopting as normalization constants the physical scales @xmath100 and @xmath101 of the @xmath127 dominant component , and by fixing @xmath298 , @xmath299 , @xmath300 , @xmath301 , @xmath302 , @xmath106 in eqs .
( b3)-(b6 ) , where @xmath84 and @xmath295 are the mass and the scale lenght of the @xmath25 component .
as the integration is trivial but the results not particularly illuminating , we do not show the explicit formulae .
99 an , j.h .
, & evans , w. 2006 , apj , 642 , 752 ( ae06 ) baes , m. , & dejonghe , h. 2004 , mnras , 351 , 18 baes , m. , dejonghe , h. , & buyle , p. 2005 , a&a , 432 , 411 bertin , g. 2000 , dynamics of galaxies ( cambridge : cambridge university press ) bertin , g. , & stiavelli , m. 1984 , a&a , 137 , 26 bertin , g. , & trenti , m. 2003 , apj , 584 , 729 bertin , g. , saglia , r.p . , & stiavelli , m. 1992 , apj , 384 , 423 binney , j. , & merrifield , m. 1998 , galactic astronomy , ( princeton : princeton university press ) binney , j. , & mamon , g. 1982 , mnras , 200 , 361 binney , j. , & tremaine , s. 2008 , galactic dynamics ( 2nd ed .
; princeton : princeton university press ) de bruijne , jos h.j . , van der marel , r.p .
, & de zeeuw , p.t . 1996 ,
mnras , 282 , 909 buyle , p. , hunter , c. , dejonghe , h. 2007 , mnras , 375 , 773 carollo , c.m . , de zeeuw , p.t . , & van der marel , r.p .
1995 , mnras , 276 , 1131 ciotti , l. 1991 , a&a , 249 , 99 ciotti , l. 1996 , apj , 471 , 68 ( c96 ) ciotti , l. 1999 , apj , 520 , 574 ( c99 ) ciotti , l. 2000 , lecture notes on stellar dynamics ( pisa : scuola normale superiore ed . ) ciotti , l. , & pellegrini , s. 1992 , mnras , 255 , 561 ( cp92 ) ciotti , l. , & lanzoni , b. 1997 , a&a , 321 , 724 ciotti , l. , lanzoni , b. , & renzini , a. 1996 , mnras , 282 , 1 ciotti , l. , & bertin , g. 2005 , a&a , 437 , 419 ciotti , l. , morganti , l. , & de zeeuw , p.t .
2008 , mnras in press ( arxiv:0809.4444 ) cuddeford , p. 1991 , mnras , 253 , 414 cuesta , a.j . ,
prada , f. , klypin , a. , & moles , m. 2008 , mnras , 389 , 385 de vaucouleurs , g. 1948 , ann.dap . , 11 , 247 dehnen , w. 1993 , mnras , 265 , 250 dejonghe , h. 1986 , _ phys . rep .
_ , 133 , no . 3 - 4 , 217 dejonghe , h. 1987 , mnras , 224 , 13 eddington , a.s .
1916 , mnras , 76 , 572 fricke , w. 1952 , astron .
nachr . , 280 , 193 hernquist , l. 1990 , apj , 536 , 359 hiotelis , n. 1994 , a&a , 291 , 725 hunter , c. , & qian , e. 1993 , mnras , 262 , 401 king , i. 1966 , apj , 146 , 372 king , i. 1972 , apjl , 174 , l123 lynden bell , d. 1962 , mnras , 123 , 447 merritt , d. 1985a , aj , 90 , 1027 nipoti , c. , londrillo , p. , & ciotti , l. 2002 , mnras , 332 , 901 nipoti , c. , londrillo , p. , & ciotti , l. 2006 , mnras , 370 , 681 osipkov , l.p .
1979 , pisma astron.zh . , 5 , 77 richstone , d.o .
, & tremaine , s.d .
1984 , apj , 286 , 27 sersic , j. l. 1968 , atlas de galaxias australes , cordoba , argentina : observatorio astronomico tremaine , s.d . ,
richstone , d.o .
, byun , y.i . , dressler , a. , faber , s.m .
, grillmair , c. , kormendy , j. , & lauer , t.r .
1994 , aj , 107 , 634 trenti , m , bertin , g. , & van albada , t.s .
2005 , a&a , 433 , 57 van albada , t.s .
1982 , mnras , 201 , 939 zhao , h. s. 1996 , mnras , 278 , 488
|
it is well known that the density and anisotropy profile in the inner regions of a stellar system with positive phase - space distribution function are not fully independent . here
we study the interplay between density profile and orbital anisotropy at large radii in physically admissible ( consistent ) stellar systems .
the analysis is carried out by using two - component ( , @xmath0 ) spherical self - consistent galaxy models , in which one density distribution follows a generalized @xmath1 profile with external logarithmic slope @xmath2 , and the other a standard @xmath0 profile ( with external slope 4 ) .
the two density components have different `` core '' radii , the orbital anisotropy is controlled with the osipkov - merritt recipe , and for simplicity we assume that the mass of the @xmath0 component dominates the total potential everywhere . the necessary and sufficient conditions for phase - space consistency
are determined analytically , also in presence of a dominant massive central black hole , and the analytical phase - space distribution function of ( , 1 ) models , and of models with a central black hole , is derived for @xmath3 .
it is found that the density slope in the external regions of a stellar system can play an important role in determining the amount of admissible anisotropy : in particular , for fixed density slopes in the central regions , systems with a steeper external density profile can support more radial anisotropy than externally flatter models .
this is quantified by an inequality formally identical to the `` cusp slope - central anisotropy '' theorem ( an & evans 2006 ) , relating at all radii ( and not just at the center ) the density logarithmic slope and the anisotropy indicator in all osipkov - merritt systems .
stellar dynamics galaxies : ellipticals dark matter
black holes
|
the kazhdan - lusztig theory deeply increases our understanding of coxeter groups and their representations and their role in lie representation theory .
the central concepts in kazhdan - lusztig theory include kazhdan - lusztig polynomial , kazhdan - lusztig basis , cell , based ring .
kazhdan - lusztig polynomials play an essential role in understanding certain remarkable representations in lie theory , for instances , the representations of quantum groups at roots of 1 , the modular representations of algebraic groups , the representations of kac - moody algebras .
kazhdan - lusztig basis and cells are very useful in understanding structure and representations of coxeter groups and their hecke algebras .
the based ring of a two - sided cell of certain coxeter groups is defined through kazhdan - lusztig basis by lusztig in [ l5 ] .
the based ring is very interesting in understanding the concerned hecke algebras and their representations ( see [ l5 , l6 , l8 ] ) .
moreover we can construct irreducible representations of the hecke algebras if we know the structure of the based ring explicitly ( see [ x3 ] ) .
this fact is remarkable since constructing computable irreducible modules of affine hecke algebras is in general difficult .
recently lusztig introduced periodic @xmath1-graphs for constructing finite dimensional modules of affine hecke algebras ( see [ l9 ] ) .
for the structure of the based ring of a two - sided cell in a weyl group or in an affine weyl group , lusztig has two nice conjectures , one for weyl groups ( see [ l7 ] ) and the other for affine weyl groups that says the based ring is isomorphic to a certain equivariant @xmath2-group ( see [ l8 ] ) .
the conjecture for weyl groups is proved by lusztig ( unpublished ) . the conjecture for affine weyl groups
is proved for rank 2 cases , for the lowest two - sided cell of an affine weyl group and for the second highest two - sided cell of an affine weyl group , see [ x1 , x3 ] . as an application
, the author gives a classification of irreducible modules of the hecke algebras of affine weyl groups of rank 2 for any non - zero parameter , see [ x3 ] .
( when the parameter is not a root of 1 , kazhdan and lusztig worked out the classification of irreducible modules of an affine hecke algebra , see [ kl2 ] . ) also the author computed the irreducible modules of affine hecke algebras associated with second highest two - sided cell , see [ x3 ] .
recently ram has developed an interesting combinatorical approach to study representations of affine hecke algebras , see [ r1 , r2 ] . in this paper we will prove lusztig conjecture on based ring for an extended affine weyl group associated with the general linear group @xmath3 or the special linear group @xmath4 , which is of type @xmath0 .
let us here briefly explain the idea of the proof of the conjecture for type @xmath0 .
for each two - sided cell of the extended affine weyl group associated with @xmath3 , we first show that the based ring of the two - sided cell is a matrix algebra over the based ring of the intersection of a left cell in the two - sided cell and its inverse ( a right cell ) .
then we show that the based ring of the intersection is isomorphic to the ( rational ) representation ring of a certain connected reductive algebraic group over complex numbers .
the main difficulty is to establish a bijection between the intersection ( of a left cell in the two - sided cell and its inverse ) and the set of isomorphism classes of irreducible rational representations of the algebraic group , and to show that the bijection leads to the isomorphism between the based ring of the intersection and the representation ring of the algebraic group .
the bijection is established in chapter 5 . in chapters 6 and 7
we prove several formulas in the based ring of the intersection , the corresponding formulas in the representation ring of the concerned algebraic group are obvious . using the formulas established in chapters 6 and 7 , in chapter 8
we show that the bijection in chapter 5 is the right one . in chapter 8
we also show that lusztig conjecture on based ring is true for the extended affine weyl group associated with @xmath4 and can not be generalized to arbitrary extended affine weyl groups .
the contents of the paper are as follows . in chapter 1
we recall some basic definitions and facts about cells and based rings . in section 1.4
we prove some simple properties about the structure constants of a based ring that are important to our proof of lusztig conjecture on based ring for an affine weyl group of type @xmath5 . in chapter 2
we discuss the structure of cells in the extended affine weyl group associated with @xmath3 .
the cells in the extended affine weyl group are explicitly described by shi and lusztig ( see [ s , l3 ] ) . in section 2.3 , we show that the based ring of a two - sided cell of the extended affine weyl group is isomorphic to a matrix algebra over the based ring of the intersection of any given left cell in the two - sided cell and its inverse ( a right cell ) . thus to understand the based ring of the two - sided cell , we only need to work out the structure of the based ring of the intersection of a left cell in the two - sided cell and its inverse . in section 2.4
we discuss chains and antichains defined by shi , which are essential for our bijection between the intersection of a left cell and its inverse and the set of isomorphism classes of irreducible rational representations of the corresponding algebraic group .
in chapter 3 we give some discussion to canonical left cells .
although we do not need the results in this chapter for the proof of our main result of the paper , but for other types canonical left cells maybe play a big role for some questions such as lusztig conjecture on based ring and properties of lusztig bijection between the set of two - sided cells of an affine weyl group and the set of unipotent classes of the corresponding algebraic group . in chapter 4
we describe the reductive algebraic group corresponding to a two - sided cell in the extended affine weyl group associated with @xmath3 and recall some needed results about the representation ring of a general linear group over complex numbers . in chapter 5
we establish a bijection between the intersection of a left cell and its inverse and the set of dominant weights of the corresponding reductive algebraic group .
we obtain this bijection by means of antichains and this bijection is one key to our main result . in chapters 6 and 7 we work with the based ring of the intersection of a left cell and its inverse .
believing in that lusztig conjecture is true , motivated by some simple multiplication formulas in the representation ring of the concerned reductive algebraic group , we prove some multiplication formulas in the based ring . using the formulas in chapters 6 and 7 , in chapter 8 we prove that the based ring of the intersection is isomorphic to the representation ring of the corresponding reductive algebraic group .
this completes our proof of lusztig conjecture on based ring for type @xmath0 . in chapter 8
we also discuss the based rings of extended affine weyl groups associated with the projective linear group @xmath6 and with the special linear group @xmath4 .
it is expected that the explicit knowledge on the based rings will have nice application to representation theory of the concerned affine hecke algebras .
it is interesting that we can use the explicit structure on based ring to find leading coefficients of some kazhdan - lusztig polynomials , the details will appear elsewhere .
i am very grateful to professors v. chari and j .- y .
shi for helpful discussions .
i thank professor le yang for his constant encouragement .
i would like to thank professor g. lusztig for many helpful comments .
i am greatly indebted to the referee for carefully reading , very valuable comments and for pointing out a serious gap in an earlier version of the paper .
part of the work was done during my visit to department of mathematics at university of california at riverside ,
i thank the department of mathematics for hospitality .
the author was in part supported by chinese national sciences foundation .
aa in this chapter we first recall some basic concepts such as cell , based ring , etc . , then in section 1.4 we give some discussions to star operations . in section 1.1
we recall the definition of kazhdan - lusztig basis of a hecke algebra . in section 1.2
we recall the definition of cell and of the @xmath7-function . in section 1.3
we recall some properties about the integers @xmath8 ( defined through the structure constants of kazhdan - lusztig basis ) , which are due to lusztig .
the star operation was introduced in [ kl1 ] and is a useful tool to study cells of coxeter groups . in section 1.4
we prove some interesting results about the relations between star operations and the integers @xmath8 .
the main result in this section is theorem 1.4.5 which says that the integers @xmath8 are invariant under star operations . in section 1.5
we recall the definition of based ring ( due to lusztig , see [ l5 ] ) and lusztig s nice conjecture about the structure of the based ring of a two - sided cell of an affine weyl group .
a few comments about the conjecture are given . in section 1.6
we discuss the relationship between the generalized star operations ( see [ l4 ] ) and the integers @xmath8 .
the results in section 1.6 should be helpful for understanding the structure of the based ring of a two - sided cell of an affine weyl group of type @xmath9 .
the basic references for this chapter are [ kl1 ] and [ l4-l8 ] . here
hecke algebras are defined for extended coxeter groups .
let @xmath10 be a coxeter system with @xmath11 the set of simple reflections .
assume that a commutative group @xmath12 acts on @xmath10 .
then we can consider the extended coxeter group @xmath13 .
the * length function * @xmath14 on @xmath15 and the * partial order * @xmath16 on @xmath15 are extended to @xmath1 as usual , that is , @xmath17 , and @xmath18 if and only if @xmath19 and @xmath20 , where @xmath21 are in @xmath22 and @xmath23 are in @xmath15 .
let @xmath24 be an indeterminate .
the * hecke algebra * @xmath25 of @xmath26 over @xmath27 $ ] with parameter @xmath28 is an associative algebra over @xmath29 , with an @xmath29-basis @xmath30 and relations ( 1 ) @xmath31 if @xmath32 , ( 2 ) @xmath33 if @xmath34 .
let @xmath35 be the involution of @xmath29 defined by @xmath36 .
then we have a bar involution of @xmath25 defined by @xmath37 for each @xmath38 there is a unique element @xmath39 in @xmath25 such that @xmath40 and @xmath41 where @xmath42 is a polynomial in @xmath24 of degree @xmath43 if @xmath44 and @xmath45 .
the basis @xmath46 is the famous * kazhdan - lusztig basis * of the hecke algebra @xmath25 and the polynomials @xmath42 are the celebrated * kazhdan - lusztig polynomials*. the basis and the polynomials have deep relations with the geometry of schubert varieties when @xmath1 is the coxeter group associated with a kac - moody algebra and the polynomials are essential in understanding some difficult irreducible representations ( such as the finite dimensional irreducible representations of quantum groups at roots of 1 , some irreducible representations of kac - moody algebras , irreducible rational representations of algebraic groups over an algebraic closed field of positive characteristic , etc . ) .
the basis also plays an important role in understanding the representations of the hecke algebra @xmath25 , see [ kl1 , l4-l8 ] . through the basis lusztig defined based ring for some coxeter groups ( including weyl groups and affine weyl groups ) . in this paper
we will determine the structure of the based ring of an affine weyl group of type @xmath0 .
applications to the representation theory of the concerned hecke algebras will appear elsewhere .
if @xmath47 and @xmath48 , we have @xmath49lower degree terms . we write @xmath50 if @xmath51 .
we then set @xmath52 .
we shall write @xmath53 if @xmath51 or @xmath54 .
we have \(a ) assume that @xmath47 and @xmath32 .
if @xmath55 , then @xmath56 and @xmath57 .
if @xmath58 , then @xmath59 and @xmath60 .
see [ kl1 ] .
cells of coxeter groups are defined in [ kl1 ] and are useful in understanding the structure and representations of @xmath1 and of the hecke algebra @xmath25 .
also we can systematically construct representations of hecke algebras by means of the cells ( see [ kl1 , l4 , l6 , l8 ] ) .
let us recall the definition of cells of coxeter groups
. for @xmath38 set @xmath61 @xmath62 let @xmath63 and @xmath64 be elements in @xmath15 .
we say that @xmath65 ( resp .
@xmath66 if there exists a sequence @xmath67 in @xmath15 such that for @xmath68 we have @xmath69 and @xmath70 ( resp .
@xmath71 ; @xmath70 or @xmath71 ) .
then for any @xmath21 in @xmath22 we say that @xmath72 ( resp .
@xmath73 ; @xmath74 ) if @xmath75 ( resp .
@xmath76 ) . for elements
@xmath23 in @xmath1 we write that @xmath77 ( resp .
@xmath78 ; @xmath79 ) if @xmath80 ( resp .
@xmath81 ; @xmath82 ) .
the relations @xmath83 are preorders on @xmath1 and the relations @xmath84 are equivalence relations on @xmath1 .
the corresponding equivalence classes are called * left cells , right cells , two - sided cells * of @xmath1 , respectively . the preorder @xmath85 ( resp .
@xmath86 ) induces a partial order on the set of left ( resp .
right ; two - sided ) cells of @xmath1 , denoted again by @xmath85 ( resp .
@xmath86 ) . the following properties ( see [ kl1 ] or [ l4 , l5 ] ) will be needed .
\(a ) if @xmath65 , then @xmath87 . in particular ,
if @xmath88 , then @xmath89 .
\(b ) if @xmath90 , then @xmath91 . in particular ,
if @xmath92 , then @xmath93 .
for any @xmath94 and @xmath38 we write @xmath95 @xmath96 @xmath97 then \(c ) @xmath98 ( resp .
@xmath99 ; @xmath100 ) if @xmath101 ( resp .
@xmath102 ; @xmath103 ) .
\(d ) assume that @xmath10 is crystallographic and that there exists a positive integer @xmath104 such that @xmath105 for all @xmath106 ( see below for the definition of @xmath107 ) .
if @xmath101 ( resp .
@xmath102 ) and @xmath108 ( resp .
@xmath109 ) , then @xmath110 .
the @xmath7-*function * on @xmath1 introduced by lusztig is a useful tool to study cells of @xmath1 and is also necessary for defining based ring of two - sided cells .
given @xmath23 in @xmath1 we write @xmath111 for @xmath106 we define @xmath107=the minimal non - negative integer @xmath112 such that @xmath113 is in @xmath114 $ ] for all @xmath23 in @xmath1 .
if such @xmath112 does not exist we set @xmath115 .
set @xmath116 for @xmath38 .
then @xmath117 c_x$ ] .
write @xmath118 then we also have @xmath107=the minimal non - negative integer @xmath112 such that @xmath119 is in @xmath114 $ ] for all @xmath23 in @xmath1 . see [ l4 ] .
we are mainly interested in extended affine weyl groups and their hecke algebras ( especially of type @xmath120 . in this section
we recall some properties about the integers @xmath8 defined through the kazhdan - lusztig basis of an affine hecke algebra .
the integers are actually structure constants of the based ring of the extended affine weyl group .
let @xmath121 be a connected reductive group over @xmath122 .
let @xmath123 be the corresponding weyl group , weight lattice , root lattice , root system , respectively .
then @xmath124 is an * affine weyl group * and @xmath125 is an * extended affine weyl group*. let @xmath11 be a set of simple reflections of @xmath15 such that @xmath126 generates @xmath127 and is a set of simple reflections of @xmath127 .
we can find an abelian subgroup @xmath22 of @xmath1 such that @xmath128 for any @xmath129 and @xmath130 .
lusztig proved that the number of left cells in @xmath1 is finite . moreover each left cell of @xmath1 contains a unique element of @xmath131 where @xmath132 is the neutral element of @xmath1 .
the set @xmath133 is a finite set of involutions of @xmath15 .
the elements in @xmath133 will be called * distinguished involutions*. see [ l5 ] .
let @xmath134 be the longest element of @xmath127 .
we have @xmath135 for all @xmath63 in @xmath1 ( see [ l4 ] ) .
thus for any @xmath136 we can define an integer @xmath8 by the condition @xmath137,\ ] ] see 1.2 for the definition of @xmath138 .
the following are some properties of @xmath139 ( see [ l5 ] for ( a)-(e ) and [ l4 ] for ( f ) ) .
\(a ) if @xmath8 is not equal to 0 , then @xmath140 and @xmath141 .
in particular we have @xmath142 if @xmath139 is not equal to 0 .
\(b ) @xmath143 and @xmath144 .
\(c ) let @xmath145 be in @xmath133 .
then @xmath146 if and only if @xmath147 and @xmath148 .
moreover @xmath149 .
\(d ) @xmath150 if and only if @xmath139 is not equal to 0 for some @xmath151 .
\(e ) if @xmath152 then @xmath153 . in particular
if @xmath79 then @xmath154 . if @xmath65 ( resp .
@xmath90 ; @xmath152 ) and @xmath154 , then @xmath77 ( resp .
@xmath78 ; @xmath79 ) .
\(f ) the positivity : @xmath155 for all @xmath156 in @xmath1 .
let @xmath26 be as in section 1.3 .
there is a very useful operation on @xmath1 , called star operation , introduced by kazhdan and lusztig in [ kl1 ] . in this section
we study the relations between star operation and the integers @xmath157 the main result is theorem 1.4.5 .
let @xmath158 and @xmath159 be in @xmath11 such that @xmath160 has order 3 , i.e. @xmath161 . define @xmath162 @xmath163 if @xmath63 is in @xmath164 , then @xmath165 contains exactly one element in @xmath164 , denoted by @xmath166 , here @xmath167 .
the map : @xmath168 , @xmath169 , is an involution and is called a * left star operation*. similarly if @xmath170 we can define the * right star operation * @xmath171 on @xmath172 , where @xmath167 .
the following are some properties proved in [ kl1 ] .
let @xmath167 .
denote by @xmath173 the subgroup of @xmath1 generated by @xmath158 and @xmath159 .
assume that @xmath174 are in @xmath164 .
we have \(a ) if @xmath175 is not in @xmath173 , then @xmath50 if and only if @xmath176 .
moreover @xmath177 .
\(b ) if @xmath175 is in @xmath173 , then @xmath50 if and only if @xmath178 .
moreover @xmath179 .
\(c ) @xmath180 if and only if @xmath181 .
\(d ) @xmath182 .
let @xmath167 .
assume that @xmath174 are in @xmath172 .
we have \(e ) if @xmath183 is not in @xmath173 , then @xmath50 if and only if @xmath184 .
moreover @xmath185 .
\(f ) if @xmath183 is in @xmath173 , then @xmath50 if and only if @xmath186 .
moreover @xmath187 .
\(g ) @xmath188 if and only @xmath189 .
\(h ) @xmath190 .
finally we have \(i ) if @xmath191 is a left cell of @xmath1 , then @xmath192 is a left cell of @xmath1 , here @xmath167 .
the following lemma shows that @xmath193 and @xmath194 have nice relationship with @xmath39 .
* lemma 1.4.2 . *
let @xmath195 be in @xmath196 such that @xmath197 has order 3 .
set @xmath167 .
_ \(a ) _ assume that @xmath63 is in @xmath164 and @xmath198 . then @xmath199 _
\(b ) _ assume that @xmath63 is in @xmath172 and @xmath200 .
then @xmath201 _ _ proof . _ ( a ) since @xmath198 , we have @xmath202 and @xmath203 .
thus ( see [ kl1 ] ) @xmath204 if @xmath205 with @xmath206 and @xmath207 , then @xmath208 . by 1.1 ( a ) , for @xmath50 with @xmath209 we have @xmath210 . in this case
( a ) is true . if @xmath211 with @xmath206 and @xmath207 , then @xmath212 and @xmath213 .
moreover , by 1.1 ( a ) , if @xmath214 , @xmath50 and @xmath209 then we have @xmath210 . in this case
( a ) is also true .
( a ) is proved .
the proof of ( b ) is similar . *
lemma 1.4.3 .
* _ let @xmath215 be in @xmath196 such that both @xmath160 and @xmath216 have order 3 .
assume that @xmath63 is in @xmath217 .
set @xmath167 and @xmath218 . then _
\(a ) _ @xmath166 is in @xmath219 and @xmath220 is in @xmath221 .
\(b ) _ @xmath222 .
we shall write @xmath223 for @xmath222 .
_ \(a ) by 1.4.1 ( d ) , @xmath182 , thus we have @xmath224 ( see 1.3 ( a ) ) , so @xmath166 is in @xmath219 .
similarly we see that @xmath220 is in @xmath221 .
\(b ) by ( a ) , @xmath225 and @xmath226 are well defined . since @xmath227 and @xmath228 , we have \(1 ) @xmath229 and @xmath230 .
thus we have \(2 ) both @xmath225 and @xmath226 are in @xmath231 .
it is no harm to assume that @xmath198 and @xmath232 . then @xmath233 and @xmath234 .
we also have @xmath235 and @xmath236 since @xmath237 and @xmath238 .
write @xmath239 by lemma 1.4.2 and 1.2 ( b - c ) , we get \(3 ) @xmath240 . moreover , if @xmath241 and @xmath242 , then @xmath243 is not in @xmath164 or @xmath243 is not in @xmath219 . noting that @xmath244 , using lemma 1.4.2 again we see @xmath245 is not equal to 0 . now using ( 2 ) and ( 3 )
we get @xmath222 .
( b ) is proved .
the lemma is proved .
\(a ) _ we have @xmath246 , so we can define @xmath247 .
_ \(b ) _ @xmath248 , see 1.2 for the definition of @xmath138 . _
\(a ) is trivial .
\(b ) we have @xmath249 and @xmath250 . by definition
we have \(1 ) @xmath251 .
we may assume that @xmath233 and @xmath252 .
using lemma 1.4.2 we get \(2 ) @xmath253 \(3 ) @xmath254 write \(4 ) @xmath255 then we have \(5 ) @xmath256 @xmath257 .
\(6 ) @xmath258 .
using ( 1 ) and comparing ( 5 ) with ( 6 ) , we get \(7 ) @xmath259 @xmath260 . using 1.2 ( a - b ) we see \(8 )
assume that @xmath261 and @xmath262 .
then @xmath263 when @xmath264 we must have @xmath265 . otherwise , noting that @xmath266 and @xmath267 , by ( 7 ) and ( 8) we have @xmath268 and @xmath269 for some @xmath270 .
assume that @xmath269 .
by 1.2 ( c ) and 1.3 ( e ) we have @xmath271 and @xmath272 . by the proof of lemma 1.4.3 we then have @xmath273 , and @xmath274 , @xmath275 if @xmath276 and @xmath277 . thus @xmath278 and @xmath279 .
this contradicts that @xmath280 .
therefore @xmath265 whenever @xmath264 . now suppose that @xmath281 . as above
we have @xmath282 whenever @xmath283 and @xmath284 . using ( 7 ) and ( 8) we then get @xmath285 in this case .
we proved ( b ) .
the proposition is proved . * theorem 1.4.5 .
* _ suppose that @xmath63 is in @xmath286 and @xmath64 is in @xmath287 .
set @xmath288 and @xmath289 .
let @xmath151 be in @xmath1 such that @xmath151 is in @xmath231 .
then we have _
@xmath290 by lemma 1.4.3 and proposition 1.4.4 ( a ) , the elements @xmath291 and @xmath247 are well defined .
if either @xmath292 or @xmath293 , then both sides of the wanted equality are 0 .
now suppose that @xmath294 and @xmath295 .
according to proposition 1.4.4 ( b ) we have @xmath248 .
therefore @xmath296 it is easy to see that @xmath297 , @xmath298 , and @xmath299 .
thus we have @xmath300 & = \ga_{{}u^\star,({}^*v^\star)^{-1},(^*w)^{-1}}\qquad \text { using prop .
1.4.4 } \\[3 mm ] & = \ga_{{}^\#u^\star,({}^*v^\star)^{-1},{}^\#((^*w)^{-1 } ) } \\[3 mm ] & = \ga_{{}^\#u^\star,({}^*v^\star)^{-1},(^*w^\#)^{-1 } } \qquad \text{using 1.3 ( b)}\\[3 mm ] & = \ga_{{}^*w^\#,{}^\#u^\star,{}^*v^\star } .
\end{array}\ ] ] the theorem is proved .
* proposition 1.4.6 .
* _ let @xmath1 be an extended affine weyl group .
_ ( a ) _ let @xmath301 be a subset of @xmath11 such that the subgroup @xmath302 of @xmath1 generated by @xmath301 is finite .
then the longest element @xmath303 is a distinguished involution .
_ in ( b ) and ( c ) @xmath145 is a distinguished involution .
\(b ) _ for any @xmath304 , the element @xmath305 is a distinguished involution . _
\(c ) _ suppose @xmath306 and @xmath197 has order 3
. then @xmath307 if and only if @xmath308 .
if @xmath307 , then @xmath309 is a distinguished involution . _
_ ( a ) is well known , see for example [ l5 ] .
\(b ) since @xmath310 and @xmath311 , for any @xmath156 in @xmath1 we have @xmath312 and @xmath313 in particular we have @xmath314 and @xmath315 for any @xmath316 . noting that @xmath317 for any @xmath63 in @xmath1 , we see @xmath318 by definition
, @xmath305 is a distinguished involution .
\(c ) let @xmath319 be the distinguished involution of the left cell containing @xmath320 .
we have @xmath321 so @xmath320 is an involution . using 1.3 ( c - d ) we get @xmath322 . using theorem 1.4.5
we then have @xmath323 . using 1.3 ( b )
we get @xmath324 . applying 1.3 ( c ) we have @xmath325 .
therefore @xmath326 .
the proposition is proved . following [ l5 ]
we define the based rings .
let @xmath327 be the free @xmath328-module with a basis @xmath329 .
the multiplication @xmath330 defines an associative ring structure on @xmath327 , see [ l5 ] .
the ring @xmath327 is called the * based ring * of @xmath1 , its unit is @xmath331 . according to 1.3(a ) , for each two - sided cell @xmath332 of @xmath1 , the @xmath328-submodule @xmath333 of @xmath327 , spanned by all @xmath334 @xmath335 , is a two - sided ideal of @xmath327 .
the ideal @xmath333 is in fact an associative ring with unit @xmath336 .
the ring @xmath333 is called the * based ring of the two - sided cell * @xmath332 . for a left cell @xmath337 of @xmath1 , the @xmath328-submodule @xmath338 of @xmath327 , spanned by all @xmath334 @xmath339 ) , is also an associative ring , its unit is @xmath340 , here @xmath145 is the unique distinguished involution in @xmath337 . * proposition 1.5.1 .
* _ let @xmath337 be a left cell of @xmath1 .
_ \(a ) _ for any @xmath129 , @xmath341 is a left cell of @xmath1 .
moreover the map @xmath342 induces an isomorphism between the based rings @xmath338 and @xmath343 . _
\(b ) _ let @xmath306 be such that @xmath160 has order 3 .
suppose @xmath344 .
then @xmath345 , here @xmath167 .
_ \(a ) obviously @xmath346 is a left cell of @xmath1 . from the proof of prop
1.4.6 ( b ) we see @xmath347 for any @xmath136 .
therefore the map @xmath348 induces an isomorphism between the based rings @xmath338 and @xmath343 .
\(b ) suppose @xmath349 .
then @xmath350 is in @xmath351 .
the map @xmath352 defines a bijection between @xmath353 and @xmath354 . according to theorem 1.4.5 we see that the map @xmath355 defines a ring isomorphism between @xmath338 and @xmath356 .
the proposition is proved .
the based rings @xmath333 are very interesting in understanding the structure and representations of hecke algebras of @xmath1 , see [ l5-l8 ] .
if we know the structure of @xmath333 explicitly we can construct modules of affine hecke algebras in a computable way , see [ l8 , x3 ] .
lusztig has a nice conjecture concerning the structure of @xmath357 .
assume that @xmath121 is connected and has a simply connected derived group .
there is a natural bijection between the set of two - sided cells of @xmath1 and the set of unipotent classes of @xmath121 , see [ l8 ] .
assume that @xmath64 is an element in the unipotent class corresponding to a two - sided cell * c * of @xmath1 .
denote by @xmath358 a maximal reductive subgroup of the centralizer @xmath359 of @xmath64 in @xmath121 .
lusztig conjectured that there exists a finite @xmath358-set @xmath360 and a bijection @xmath361 the set of isomorphism classes of irreducible @xmath362 vector bundles on @xmath363 such that @xmath364 defines a ring isomorphism ( preserving the unit element ) between @xmath333 and @xmath365 and @xmath366 , see [ l8 ] for the conjecture and for the definition of @xmath367 .
when @xmath358 is connected ( or equivalently @xmath359 is connected ) , @xmath358 must act on @xmath360 trivially . in this case
@xmath368 is the number of left cells contained in @xmath332 and @xmath365 is isomorphic to the @xmath369 matrix algebra @xmath370 over the rational representation ring @xmath371 of @xmath358 .
let irr@xmath362 be the set of isomorphism classes of rational irreducible representations of @xmath372 .
then lusztig conjecture says that there is a bijection @xmath373 and @xmath374 , where we identify @xmath375 with the matrix whose @xmath376-entry is @xmath377 and other entries are 0 , ( 2 ) @xmath378 if @xmath375 , here @xmath379 is the dual of @xmath377 .
lusztig conjecture has been proved for the following cases , ( 1 ) @xmath380 is the lowest two - sided cell of @xmath1 , ( that is , @xmath332 is the two - sided cell of @xmath1 containing the longest element @xmath134 of @xmath127 , ) see [ x1 ] , ( 2 ) rank 2 cases , see [ x3 ] , ( 3 ) the case @xmath381 , see [ x3 ] . when @xmath382 , each @xmath358 is connected .
the purpose of this paper is to show that lusztig conjecture is true for the extended affine weyl group associated with @xmath3 .
the star operation , introduced in [ kl1 ] , was generalized in [ l4 ] .
we are interested in the relationship between the ( generalized ) star operation and the structure constants @xmath8 of the based ring of an extended affine weyl group . in this section
we show that theorem 1.4.5 remains true for the generalized star operation .
we also give a few other identities about the constants @xmath8 .
the results in this section are not used in the sequent chapters but should be useful for understanding the based ring of an arbitrary extended affine weyl group .
let @xmath26 be as in section 1.3 .
assume that @xmath158 and @xmath159 are in @xmath11 and @xmath160 has order @xmath383 .
denote by @xmath384 the subgroup of @xmath1 generated by @xmath158 and @xmath159 .
each coset @xmath385 can be decomposed into four parts : one consists of the unique element @xmath386 of minimal length in the coset , one consists of the unique element @xmath243 of maximal length in the coset , one consists of the @xmath387 elements @xmath388 one consists of the @xmath387 elements @xmath389 .
the last two subsets are called left strings ( related to @xmath390 ) and shall be regarded as sequences ( as above ) rather than subsets . similarly we define right strings ( related to @xmath390 ) .
we have ( see [ l4 ] ) \(a ) a left string in @xmath1 is contained in a left cell of @xmath1 and a right string in @xmath1 is contained in a right cell of @xmath1 .
assume that @xmath63 is in a left ( resp .
right ) string ( related to @xmath390 ) of length @xmath387 and is the @xmath112th element of the left ( resp .
right ) string , we define @xmath166 ( resp .
@xmath391 ) to be the @xmath392th element of the string , where @xmath167 .
we have * lemma 1.6.1 . * _ let @xmath63 be in @xmath1 such that @xmath63 is in a left string related to @xmath167 and is also in a right string related to @xmath289 . then _
\(a ) _ @xmath393 is in a right string related to @xmath394 and @xmath220 in a left string related to @xmath390 .
_ \(b ) _ @xmath222 .
we shall write @xmath223 for @xmath222 . _ the proof is similar to that of lemma 1.4.3 although more complicated .
the following theorem is a generalization of theorem 1.4.5 * theorem 1.6.2 .
* _ let @xmath156 be in @xmath1 such that ( 1 ) @xmath63 is in a left string related to @xmath167 and also in a right string related to @xmath395 , ( 2 ) @xmath64 is in a left string related to @xmath395 and also in a right string related to @xmath396 , ( 3 ) @xmath151 is in a left string related to @xmath167 and also in a right string related to @xmath396 . then _
@xmath290 the proof is similar to that of theorem 1.4.5 .
* in this subsection we assume that @xmath158 and @xmath159 are in @xmath11 and @xmath160 has order 4 .
let @xmath156 be in @xmath1 such that @xmath397 and @xmath398 .
as in the proof of lemma 1.4.2 we have @xmath399 @xmath400 @xmath401 considering the products @xmath402 and using the associativity of multiplication of the hecke algebra @xmath25 , we get \(a ) @xmath403 \(b ) @xmath404 \(c ) @xmath405 \(d ) @xmath406
\(e ) @xmath407 \(f ) @xmath408 in particular , we have ( a ) @xmath409 ( b ) @xmath410 ( c ) @xmath411 ( d ) @xmath412 ( e ) @xmath413 ( f ) @xmath414 one may compare the above equalities with the equalities in [ l4 , ( 10.4.2 ) ] . using 1.3 ( b ) and the above equalities we can get more equalities for the structure constants of the based ring @xmath327 , we omit them .
when @xmath160 has higher order , there exist similar equalities .
we omit the discussion .
from now on we will concentrate on type @xmath0 .
the cells of an affine weyl group of type @xmath415 have been described in [ s , l3 ] . in this chapter
we first recall some facts about the cells , then we derive some new results for our purpose . in section 2.1
we recall an alternative definition ( due to lusztig ) for the extended affine weyl group @xmath1 associated with @xmath416 . in section 2.2
we recall the description of cells of @xmath1 in [ s , l3 ] . for this description and later use
we slightly refine the definition of chain and antichain in [ s ] .
more precisely we will define d - chain ( resp .
d - antichain ) and r - chain ( resp .
r - antichain ) . in this section
we also consider the intersection of left cells and right cells .
the intersections are important to our purpose . in section 2.3
we show that the based ring of a two - sided cell of @xmath1 is isomorphic to a matrix algebra over the based ring @xmath417 for any left cell @xmath337 in the two - sided cell ( see theorem 2.3.2 ) .
this is the main result of this chapter .
thus to understand lusztig s conjecture on the structure of the based ring of the two - sided cell we only need to understand the structure of @xmath417 .
we will do this in chapters 5 - 8 . for later use in section 2.4
we work out some results about chains and antichains .
the results will be needed in chapter 5 for defining the required bijection between @xmath418 and irr@xmath372 , see chapter 5 . in section 2.5
we prove a result about star operation . in this section
we recall a definition of lusztig for affine
weyl group of type @xmath0 , see [ l2 ] .
suppose that @xmath382 is the general linear group over @xmath122 of degree @xmath419 .
then the weyl group @xmath127 of @xmath121 is isomorphic to the symmetric group of @xmath419 letters .
let @xmath420 be the subgroup of @xmath121 consisting of all diagonal matrices in @xmath121 .
then @xmath420 is a maximal torus of @xmath121 and the weight lattice @xmath421 is the character group of @xmath420 .
let @xmath422 be the homomorphism @xmath423 .
we have @xmath424 and the extended affine weyl group associated with @xmath3 is @xmath125 .
we identify @xmath127 with the group of all the @xmath425 permutation matrices ( by a permutation matrix , we mean a monomial matrix whose nonzero entries are all equal to 1 ) .
let @xmath426 @xmath427 be the simple reflection of @xmath127 obtained from the identity matrix by interchanging the @xmath112th and the @xmath428th rows
. then @xmath429 and @xmath430 if @xmath431 .
another realization of @xmath1 , due to lusztig , is as follows .
consider the permutations @xmath432 that satisfy @xmath433 and @xmath434 all such permutations form a permutation group @xmath435 of @xmath328 .
* lemma 2.1.1 : * _ @xmath1 is isomorphic to @xmath435 . _
_ we can define a @xmath1-action on @xmath328 as follows , @xmath436 @xmath437 it is easy to check that this action defines an isomorphism between @xmath1 and @xmath435 .
the lemma is proved .
we shall identify @xmath1 and @xmath435 using the above isomorphism . *
lemma 2.1.2 .
* _ the permutation @xmath438 for all @xmath112 , is in @xmath1 . _ _ proof . _
it is clear from the definition .
the following are some simple properties of the extended affine weyl group @xmath1 .
\(a ) the center of @xmath1 is generated by @xmath439 .
\(b ) @xmath440 , where @xmath441 is defined by @xmath442 if @xmath443 , and @xmath444 if @xmath445 .
\(c ) let @xmath22 be the subgroup of @xmath1 generated by @xmath446 and @xmath15 the subgroup of @xmath1 generated by all @xmath426 .
then @xmath22 is an infinite cyclic group and @xmath15 is an affine weyl group of type @xmath0 .
moreover we have @xmath447 .
\(d ) we have @xmath448 if @xmath449 and @xmath450 @xmath451 note that the length of @xmath452 is @xmath453 .
\(e ) as a special case of the length formula in [ i m ] , we have @xmath454\right|,\ ] ] where [ h ] is the integer part of h ( recall that @xmath455<1)$ ] . for a direct proof of the formula , see [ s , lemma 4.2.2 ] .
\(f ) let @xmath38 .
then @xmath456 if and only if @xmath457 , see [ s , corollary 4.2.3 ] .
the cells in @xmath1 have a beautiful combinatorial description , conjectured by lusztig and proved by shi . in this section
we recall the description of cells of @xmath1 in [ s , l3 ] . following shi
we define chains and antichains .
more precisely we refine slightly his definition by defining d - chains , d - antichains , r - chains and r - antichains .
let @xmath63 be an element of @xmath1 and @xmath458 be integers .
we call @xmath458 a * d - chain * of @xmath63 of length @xmath459 and @xmath460 an * r - chain * of @xmath63 of length @xmath459 if \(1 ) @xmath461 , \(2 ) @xmath462 whenever @xmath463 , \(3 ) @xmath464 , here d - chain means domain chain and r - chain means range chain
. thus a d - chain of @xmath1 essentially is a subset of @xmath328 whose natural order is reversed by @xmath63 . a * d - chain ( resp .
r - chain ) family set * of @xmath63 of index @xmath24 is a subset @xmath360 of @xmath328 such that ( 1 ) elements in @xmath360 are noncongruent to each other modulo @xmath419 , ( 2 ) @xmath360 is a disjoint union of @xmath24 d - chains ( resp .
r - chains ) @xmath465 of @xmath63 .
we also call @xmath466 a * d - chain ( resp .
r - chain ) family * of @xmath1 .
we call @xmath458 a * d - antichain * of @xmath63 of length @xmath459 and @xmath460 an * r - antichain * of @xmath63 of length @xmath459 if \(1 ) @xmath467 , ( then @xmath468(mod @xmath419 ) whenever @xmath469 , ) \(2 ) @xmath470 . a * d - antichain ( resp .
r - antichain ) family set * of @xmath63 of index @xmath24 is a subset @xmath360 of @xmath328 such that ( 1 ) elements in @xmath360 are noncongruent to each other modulo @xmath419 , ( 2 ) @xmath360 is a disjoint union of @xmath24 d - antichains ( resp .
r - antichains ) @xmath465 of @xmath63 .
we also call @xmath466 a * d - antichain ( resp .
r - antichain ) family * of @xmath1 . obviously if @xmath360 is an r - chain ( resp .
r - antichain ) family set of @xmath63 of index @xmath24 , then @xmath360 is a d - chain ( resp .
d - antichain ) family set of @xmath471 of index @xmath24 .
we shall regard @xmath1 as the permutation group @xmath435 of @xmath328 .
following lusztig we associate a partition @xmath472 of @xmath419 with an element @xmath63 of @xmath1 as follows .
let @xmath473 be the maximal one among the cardinalities of all d - chain family sets of @xmath63 of index @xmath112
. then @xmath474 .
according to [ g , th .
1.5 ] , @xmath475 .
we call @xmath476 the * partition associated with @xmath63*. according to [ s , l3 ] , @xmath477 if and only if @xmath478 .
moreover the number of left cells in a two - sided cell corresponding to a partition @xmath479 of @xmath419 is @xmath480 , where @xmath481 is the dual partition of @xmath479 , see [ s ] .
the dual partition of @xmath479 can be defined through antichains of @xmath63 .
let @xmath482 be the maximal one among the cardinalities of all d - antichain family sets of @xmath63 of index @xmath112 .
then @xmath483 .
according to [ g , th .
1.5 ] , @xmath484 .
the partition @xmath485 is the partition dual to the partition @xmath486 , see [ g , th .
let @xmath332 be the two - sided cell of @xmath1 corresponding to a partition @xmath487 of @xmath419 .
denote by @xmath488 the longest element of the subgroup of @xmath1 generated by all @xmath489 then @xmath490 and there is a unique left cell in @xmath332 containing @xmath488 , denoted by @xmath491 .
we shall write @xmath492 for the left cell containing @xmath493 . set @xmath494 ,
this is a right cell contained in @xmath495 .
denote by @xmath496 the longest element of the subgroup of @xmath1 generated by all @xmath497 then @xmath498 and there is a unique left cell in @xmath332 containing @xmath496 , denoted by @xmath499 .
we shall write @xmath500 for the left cell containing @xmath501
. set @xmath502 .
note that the intersection of the set @xmath503 in [ s , 9.3 ] and @xmath332 is contained in the union @xmath504 .
* lemma 2.2.1 .
* _ assume that @xmath505 .
then through a succession of left star operations and of right star operations on @xmath63 we can get an element in @xmath506 for some integers @xmath507 .
_ it follows from prop .
9.3.7 , theorem 1.6.3 ( i ) and lemma 18.3.2 in [ s ] .
* corollary 2.2.2 .
* _ assume that @xmath337 is a left cell in @xmath332 and @xmath508 is a right cell in @xmath332 .
then @xmath509 can be obtained by applying a succession of left star operations and of right star operations on some @xmath506 .
_ since @xmath510 , the map @xmath511 defines a bijection between @xmath512 and @xmath506 .
thus it is very fundamental to understand the properties of @xmath512 . *
proposition 2.2.3 . *
_ let @xmath337 be a left cell in @xmath332 .
then there is a bijection @xmath513 such that @xmath514 defines an isomorphism between @xmath515 and @xmath516 _ using corollary 2.2.2 and lemma 18.3.2 in [ s ] , we can find @xmath112 such that @xmath517 is obtained from @xmath512 by applying a succession of left star operations and the corresponding right star operations . using prop .
1.5.1 we see that the assertion is true .
in this section we will show that the based ring @xmath519 of a two - sided cell @xmath520 of @xmath1 is a matrix algebra over @xmath417 for any left cell @xmath337 in @xmath380 .
this is the main result of this chapter and is also one of the key steps of our proof of lusztig conjecture for the structure of @xmath519 .
recall that for the two - sided cell @xmath520 corresponding to a partition @xmath479 of @xmath419 , we have a unique left cell @xmath491 containing @xmath488 .
we number the left cells in @xmath520 as @xmath521 , where @xmath522 and @xmath523 is the dual partition of @xmath487 .
let @xmath524 be the right cell corresponding to @xmath525 .
we would like to define a bijection @xmath526 between @xmath527 and @xmath528 .
there are several cases .
first we suppose @xmath529 .
let @xmath530 be such that @xmath531 but @xmath532 for @xmath533 .
\(a ) if @xmath534 for some @xmath535 , then obviously @xmath536 defines a bijection from @xmath537 to @xmath538 .
\(b ) now let @xmath539 be arbitrary . using [ s , prop.9.3.7 ] , we can find @xmath540 between @xmath541 and @xmath542 such that @xmath539 is obtained from @xmath543 by a sequence of left star operations .
such @xmath540 and sequence of left star operations are usually not unique .
we fix such an @xmath540 and the sequence of left star operations
. then @xmath544 is obtained from @xmath545 by applying the sequence of left star operations .
this of course establishes a bijection between @xmath544 and @xmath546 . using ( a ) we then get a bijection between @xmath544 and @xmath547 .
\(c ) now for any @xmath539 we have fixed an @xmath548 and a sequence of left star operations .
then @xmath525 can be obtained from @xmath549 by applying the corresponding sequence of right star operations . by this way and using ( a ) we get a bijection @xmath526 between @xmath550 and @xmath538 , cf .
lemma 2.2.1 .
the following are some properties of the bijection @xmath551 .
\(a ) _ let @xmath473 be the distinguished involution in @xmath552
. then @xmath553 . _
\(b ) _ note that @xmath554 . for @xmath555
we have @xmath556 . _ using prop .
1.4.6 we see that ( a ) is true .
( b ) follows from @xmath557 and @xmath558 .
we shall use @xmath559 for any square matrix whose @xmath376-entry is @xmath334 and other entries are 0 . * theorem 2.3.2 . *
_ let @xmath380 be the two - sided cell of @xmath1 corresponding to a partition @xmath479 of @xmath419 and @xmath560 the dual partition of @xmath479 .
_ ( a)_the map @xmath561 induces a ring isomorphism from @xmath562 to @xmath563 . _
\(b ) _ the based ring @xmath563 is commutative .
_ \(c ) _ the map @xmath564 defines an isomorphism from the based ring @xmath519 to @xmath565 , the @xmath566 matrix algebra over the ring @xmath567 . _
\(a ) follows from prop .
1.5.1 and its proof .
\(b ) let @xmath568 be the canonical left cell in @xmath380 , i.e. , @xmath568 is a left cell in @xmath380 and @xmath569 . by [ lx ]
we know that @xmath570 is commutative . using ( a ) we see that ( b ) is true .
\(c ) note that for any @xmath571 in @xmath12 and @xmath572 in @xmath332 we have @xmath573 let @xmath574 and @xmath575 . using theorem 1.4.5
repeatedly we get @xmath576 combining this and 1.3 ( a ) we see that ( c ) is true .
the theorem is proved . for @xmath539
if we choose different @xmath540 and/or different sequence of left star operations then we would usually get a different isomorphism in theorem 2.3.2 ( c ) .
for instance , let @xmath577 and @xmath380 the two - sided cell of @xmath1 corresponding to the partition @xmath578 .
we have @xmath579 .
let @xmath539 ( @xmath580 be the right cell of @xmath1 containing @xmath426 . then @xmath380 is the union of @xmath581 and @xmath582 .
we have @xmath583 and @xmath584 .
let @xmath585 and @xmath586 .
we also have @xmath587 and @xmath588 in this example it is easy to see that different @xmath540 and/or different sequences of left star operations usually lead to different isomorphisms in theorem 2.3.2 ( c ) .
let @xmath479 be a partition of @xmath419 and @xmath380 the two - sided cell of @xmath1 corresponding to @xmath479 .
let @xmath560 be the dual partition of @xmath479 and @xmath64 a unipotent element of @xmath3 whose jordan blocks are given by the partition @xmath560 .
denote by @xmath589 a maximal reductive subgroup of the centralizer of @xmath64 in @xmath3 . according to theorem 2.3.2 ( c ) and lemma 2.3.1 , to prove the conjecture of lusztig for @xmath519 we only need to prove the following special case of the conjecture .
there is a bijection @xmath590 such that \(a ) the map @xmath591 defines a ring isomorphism from @xmath567 to @xmath592 .
\(b ) @xmath593 for any @xmath594 .
( recall that @xmath595 is the dual of @xmath596 . )
we will prove this conjecture in chapter 8 . to define the map @xmath597 in 2.3.3 ( a ) we need some properties of antichains . in the following section we give some discussions to chains and antichains .
in chapetr 5 we will define the map @xmath597 in conjecture 2.3.3 by means of r - antichains . in this section
we prove some results about chains and antichains which will be used later . in this section
@xmath63 stands for an element of @xmath1 .
* lemma 2.4.1 . * _ if @xmath112 and @xmath598 are in a d - chain ( resp .
a d - antichain ) of @xmath63 , then any d - antichain ( resp .
d - chain ) of @xmath63 contains at most one of @xmath599 for any given integers @xmath600 .
_ suppose that @xmath601 are in a d - chain of @xmath63
. then @xmath602 .
assume that @xmath603 and @xmath604 are in a d - antichain of @xmath63 for some integers @xmath600 . if @xmath605 then we have @xmath606 since @xmath601 .
thus @xmath607 since @xmath602 .
in this case @xmath603 and @xmath604 can not be in the same d - antichain of @xmath63 . if @xmath608 then @xmath609 since @xmath601
. then we have @xmath610 since @xmath602 . in this case @xmath603 and @xmath604 are also not in the same d - antichain of @xmath63 .
suppose that @xmath601 are in a d - antichain of @xmath63 , then @xmath611 and @xmath612 .
assume that @xmath603 and @xmath604 are in a d - chain of @xmath63 for some integers @xmath600 . if @xmath613 , then @xmath614 since @xmath611 .
thus @xmath615 since @xmath616 . in this case @xmath603 and @xmath604 can not be in the same d - chain of @xmath63 . if @xmath617 , then @xmath618 since @xmath601 .
then we have @xmath619 since @xmath620 . in this case
both @xmath603 and @xmath604 are also not in the same d - chain of @xmath63 .
the lemma is proved . *
corollary 2.4.2 . *
_ if @xmath621 and @xmath622 are in an r - chain ( resp .
r - antichain ) of @xmath63 , then any d - antichain ( resp .
d - chain ) of @xmath471 contains at most one of @xmath623 for any given integers @xmath600 . _ since @xmath621 and @xmath624 are in a d - chain ( resp .
d - antichain ) of @xmath471 , the assertion follows from lemma 2.4.1 .
* lemma 2.4.3 . * _ if @xmath461 is a d - antichain of @xmath63 of length @xmath459 , then @xmath625 is also a d - antichain of length @xmath459 .
( we shall say that @xmath626 and @xmath627 are * equivalent antichains * for any integer @xmath7 and @xmath628 . ) _ since @xmath629 and @xmath630 we see that the lemma is true .
* proposition 2.4.4 .
* _ let @xmath38 and @xmath631 be the dual partition of @xmath486 .
given any @xmath632 and any consecutive @xmath419 integers @xmath633 , we can find a d - antichain family set of @xmath63 of index @xmath459 that is included in @xmath360 and has cardinality @xmath634 .
_ by definition we have a d - antichain family @xmath635 of @xmath63 of index @xmath459 that has cardinality @xmath634 . using lemma
2.4.3 we see that any d - antichain in @xmath635 is equivalent to a d - antichain of @xmath63 in @xmath360 .
noting that equivalent d - antichains have the same congruence classes modulo @xmath419 , we see that the required d - antichain family set exists .
the proposition is proved .
of course this assertion is false for chain family set .
for instance , if @xmath636 for @xmath637 , then @xmath638 since @xmath639 is a d - chain of @xmath63 of length @xmath419 .
but @xmath640 is not a d - chain of @xmath63 .
let @xmath38 and @xmath641 .
assume that the dual partition of @xmath486 is @xmath642 we call a subset @xmath360 of @xmath328 a * complete d - chain ( resp .
d - antichain ) set * of @xmath63 if \(1 ) @xmath360 contains @xmath419 numbers and any two numbers in y are noncongruent modulo @xmath419 , \(2 ) @xmath360 is a disjoint union of @xmath643 d - chains ( resp . @xmath644 d - antichains ) of @xmath63 of length @xmath645 ( resp .
@xmath646 ) respectively .
similarly we define * complete r - chain sets * and * complete r - antichain sets * of @xmath63 .
let @xmath38 and @xmath631 the dual partition of @xmath486 .
we say that @xmath647 is a * complete d - antichain family * of @xmath63 if the following four conditions are satisfied : \(1 ) all @xmath648 are d - antichains of @xmath63 , \(2 ) @xmath649 , \(3 ) the cardinality of @xmath648 is @xmath650 for all @xmath112 , \(4 ) the union of all @xmath648 is @xmath640 .
( necessarily @xmath651 if @xmath652 . )
we say that @xmath653 is a * complete r - antichain family * of @xmath63 if @xmath654 is a complete d - antichain family of @xmath63 . *
example : * let @xmath487 be a partition of @xmath419
. then @xmath488 has a complete d - antichain family .
in fact , for given @xmath655 , let @xmath656 be such that @xmath657 but @xmath658 ( we set @xmath659 . then define @xmath660 it is easy to see that @xmath661 form a complete d - antichain family of @xmath488 . in general , an element in @xmath1
does not have a complete d - antichain family .
for example , let @xmath662 and @xmath663 .
then @xmath664 , here @xmath665 and @xmath63 does not have any complete d - antichain family .
( i am grateful to the referee for providing a similar example . ) in this section we give a result ( see lemma 2.5.2 ) about the star operation .
the following result is a partial generalization of [ s , corollary 4.2.3 ] .
* lemma 2.5.1 . *
_ let @xmath32 be a simple reflection and @xmath316 .
suppose @xmath666 and @xmath667 .
then @xmath668 if @xmath202
. _ write @xmath669 where @xmath670 and @xmath671
. set @xmath672 assume that @xmath673 for some @xmath674 .
choose @xmath675 such that @xmath676 if @xmath677 , then @xmath678 and @xmath679 .
. therefore if @xmath681 and either @xmath112 or @xmath598 is not in @xmath682 then @xmath683\right|=\left|\left [ \frac{sw(j)-sw(i)}n\right]\right|.\ ] ] now @xmath684 and @xmath685 .
if @xmath686 we must have @xmath687 since @xmath202 .
if @xmath688 , we have @xmath689 for the same reason .
if neither @xmath690 nor @xmath691 is contained in @xmath692 , then we have @xmath693 if @xmath690 is either @xmath694 or @xmath695 and @xmath691 is not in @xmath696 , then @xmath697 is either equal to @xmath698 or @xmath699 and @xmath700 when @xmath701 , we must have @xmath702 or @xmath703 and @xmath704 since @xmath667 .
but @xmath705 , so we have @xmath706 similarly we see @xmath668 if @xmath707 and @xmath691 is either @xmath694 or @xmath695
. now suppose @xmath708 . if @xmath709 , then @xmath710 if @xmath711
then @xmath712 and @xmath713 we must have @xmath714 since @xmath233 . then @xmath715 since @xmath716 , but this is impossible since @xmath717 thus we proved the lemma if @xmath718 .
assume that @xmath719 .
choose @xmath720 such that @xmath721 if @xmath112 is not in @xmath722 , then @xmath678 and @xmath723 .
. therefore if @xmath725 and @xmath726 then @xmath683\right|=\left|\left [ \frac{sw(j)-sw(i)}n\right]\right|.\ ] ] note that @xmath727 and @xmath728 .
if @xmath729 we must have @xmath730 since @xmath202 .
if @xmath731 , we have @xmath732 for the same reason .
if neither @xmath690 nor @xmath691 is contained in @xmath733 , then we have @xmath693 if @xmath690 is either @xmath734 or @xmath735 and @xmath736 , then @xmath697 is either @xmath737 or @xmath738 and @xmath700 we must have @xmath739 if @xmath701 since @xmath667 .
but @xmath740 , so we have @xmath706 similarly we see @xmath668 if @xmath741 and @xmath691 is either @xmath734 or @xmath735
. now suppose @xmath742 . if @xmath743 , then @xmath744 . if @xmath745 we must have @xmath746 since @xmath667 , but this is impossible since @xmath747 for the reason of @xmath233 and @xmath748 .
the lemma is proved .
* lemma 2.5.2 . *
_ let @xmath749 and @xmath750 .
we have _
\(a ) _ suppose that @xmath621 is between @xmath751 and @xmath752 , then @xmath63 is in @xmath753 moreover , _
@xmath754 \(b ) _ suppose that @xmath752 is between @xmath621 and @xmath751 , then @xmath63 is in @xmath753 moreover , _ @xmath755 \(a ) if @xmath756 , using 2.1.3 ( f ) and lemma 2.5.1 repeatedly we see @xmath757 for some @xmath64 with @xmath758 and @xmath759 . if @xmath760 , using 2.1.3 ( f ) and lemma 2.5.1 repeatedly we see @xmath761 for some @xmath64 with @xmath758 and @xmath759 .
therefore @xmath762 in both cases we have @xmath763 so that @xmath764 if @xmath765 , and @xmath766 .
the proof of ( b ) is similar .
the lemma is proved .
in section 2.3 we proved that @xmath417 is commutative for any left cell @xmath767 of the extended affine weyl group @xmath1 associated with @xmath3 , using the fact that it is true for canonical left cells . in this chapter
we give some discussion to canonical left cells in @xmath1 .
although we do not really need the results here for our main purpose , the discussion here maybe is helpful for understanding other types . in other types
the based ring @xmath417 is not commutative in general , one may see this from [ x3 , chapter 11 ] , but it is always commutative when @xmath337 is a canonical left cell ( see [ lx ] ) .
lusztig conjectured that @xmath768 if @xmath337 is a canonical left cell in a two - sided cell @xmath380 of an extended affine weyl group ( see [ l8 ] ) . if this is true , maybe @xmath417 is a key to understand @xmath519 .
one more reason to consider canonical left cells is that the intersection @xmath769 has a good presentation if @xmath767 is a canonical left cell ( see [ lx ] ) .
this fact maybe can be used to prove that the bijection between the set of two - sided cells of an extended affine weyl group and the set of unipotent classes in the corresponding algebraic group ( see [ l8 ] ) preserves partial orders . in section 3.1
we recall some reduced expressions for fundamental weights , which can be found in [ l2 ] . in section 3.2
we determine the right cell containing a given dominant weight . in section 3.3
we discuss the shortest elements @xmath770 in the double cosets @xmath771 for all dominant weights @xmath386 .
it is not so easy to describe the partition associated with @xmath770 . in section 3.4
we describe the distinguished involution in a canonical left cell of @xmath1 .
let @xmath1 be as in chapter 2 , that is , @xmath1 is the extended affine weyl group associated with @xmath3 .
define @xmath772 . recall that @xmath421 is the subgroup of @xmath1 generated by all @xmath452 .
then the set @xmath773 of dominant weights in @xmath1 consists of the elements @xmath774 , @xmath775 and @xmath776 . for further investigation
we need some reduced expression for @xmath777 , see [ l2 ] , @xmath778 in particular , the length of @xmath777 is @xmath779 . from the reduced expressions we see \(a ) @xmath780 if @xmath386 and @xmath243 are in @xmath781 .
\(b ) @xmath782 for @xmath783 and @xmath784 if @xmath652 .
\(c ) @xmath785 for @xmath783 . in this section
we describe the partition associated with a dominant weight @xmath386 in @xmath1 .
this is equivalent to determine the right cell containing @xmath386 .
let @xmath568 be the unique left cell contained in a two - sided cell @xmath380 of @xmath1 with @xmath787 .
for convenience we set @xmath788 .
obviously we have \(a ) if @xmath386 is a dominant weight in @xmath1 then @xmath386 is contained in some @xmath789 .
let @xmath479 be a partition of @xmath419 and @xmath523 be the dual partition of @xmath479 .
let @xmath380 be the two - sided cell corresponding to @xmath479 .
* lemma 3.2.1 . *
_ keep the notation above .
let @xmath790 be a permutation of @xmath646 .
. then @xmath792 . _
set @xmath793 and @xmath794 for @xmath795 .
define @xmath796 for @xmath795 . for any @xmath797 we have @xmath798 .
thus @xmath799 is a d - antichain of @xmath800 . if @xmath797 , @xmath801 and @xmath601 , then @xmath802 . using lemma
2.4.2 we see that @xmath803 and @xmath804 can not be in the same d - antichain of @xmath805 for any integers @xmath806 .
therefore @xmath807 must form a complete d - antichain family of @xmath800 ( see 2.4.5 for definition ) and the corresponding partition is @xmath560 .
so @xmath800 is in @xmath789 . *
lemma 3.2.2 . *
_ let @xmath301 be a subset of @xmath808 . then _
\(a ) _ the elements @xmath809 and @xmath810 are contained in the same right cell of @xmath1 if all @xmath811 . _
\(b ) _ the elements @xmath812 are contained in the right cell containing @xmath813 . _ we can find a partition @xmath631 of @xmath419 and a permutation @xmath790 of @xmath646 such that @xmath814 ( see lemma 3.2.1 for definition ) .
let @xmath815 be as in the proof of lemma 3.2.1 .
then we see that @xmath815 form a complete d - antichain family for both @xmath809 and @xmath813 if all @xmath811 .
therefore ( a ) is true .
( b ) follows from ( a ) since @xmath816 is in the center of @xmath1 .
the lemma is proved .
the two lemmas show that it is easy to determine the right cell containing a given dominant weight .
\(1 ) @xmath817 is contained in the lowest two - sided cell of @xmath1 .
\(2 ) @xmath777 and @xmath818 are contained in the same two - sided cell ( we set @xmath819 , the neutral element of @xmath1 ) .
the partition corresponding to the two - sided cell is ( 2,2, ... ,2,1, ... ,1 ) , where 2 appears min@xmath820 times .
it is well known that @xmath1 is the union of the double cosets @xmath771 @xmath821 ) . for the double coset @xmath771
we have a unique element @xmath770 of minimal length .
the elements @xmath822 are interesting since we have ( see [ lx ] ) \(a ) @xmath823 where @xmath332 runs through the set of two - sided cells of @xmath1 .
\(b ) @xmath824 if @xmath825 are in @xmath781 and @xmath826 .
\(c ) let @xmath38 .
then @xmath827 if and only if @xmath828 for some @xmath786 . since @xmath829 is in the center of @xmath1 , we have \(d ) @xmath830 if @xmath786 and @xmath7 is an integer . in this section
we work out an explicit form for @xmath770 .
unfortunately the author is unable to describe the left cell containing @xmath770 .
fix a subset @xmath301 of @xmath808 . recall @xmath831 .
we set @xmath832 .
* lemma 3.3.1 . *
_ let @xmath301 be a subset of @xmath833 and @xmath834 .
then @xmath835 if @xmath811 for all @xmath112
. _ _ proof .
_ let @xmath836 , @xmath837 , and @xmath838 .
then @xmath839 for @xmath783 .
obviously we have @xmath840 for @xmath841 .
thus we only need to show that @xmath842 for @xmath783 .
let @xmath843 be such that @xmath844 . recall that @xmath845 is the root system of @xmath1 .
let @xmath846 be the set of positive roots in @xmath845 and @xmath847 .
we have ( see [ i m ] ) @xmath848 @xmath849 if @xmath850 , we can find a unique @xmath851 such that @xmath852 and @xmath853 .
then by the formulas above we have @xmath854 since @xmath855 . since @xmath856 , we necessarily have @xmath857 .
this implies that @xmath858 . by the length formulas for @xmath859 and for @xmath860 we see @xmath842 . if @xmath861 , we can find a unique @xmath851 such that @xmath862 and @xmath863 .
then by the formulas above we have @xmath864 since @xmath855 . since @xmath856 , we necessarily have @xmath865 .
this implies that @xmath866 . by the length formulas for @xmath867 and for @xmath868 we see @xmath842 . the lemma is proved . by the lemma above , to get an explicit form for @xmath770 we only need to get an explicit form for @xmath869 .
we need some notation .
for @xmath870 we define @xmath871 .
* lemma 3.3.2 . *
_ suppose @xmath872 is a subset of @xmath833 .
then we have @xmath873 where @xmath874 for @xmath875 _ it is easy to see that @xmath876 is in @xmath127 .
therefore @xmath877 is in the double coset @xmath878 . using the reduced expressions in 3.1 and using 3.1 ( a ) and 3.1 ( b ) , we see @xmath879 .
thus @xmath880 .
we also need show that @xmath881 .
we can write down explicitly the action of @xmath270 on @xmath640 , which is , @xmath882 where @xmath883 .
in other words , we have @xmath884 for @xmath628 , @xmath885 ( we set @xmath886 and @xmath887 ) .
thus @xmath888 if @xmath889 . using lemma 2.5.1 we see @xmath890 .
the lemma is proved .
\(1 ) @xmath891 .
\(2 ) @xmath892 for all @xmath112 .
thus @xmath893 if @xmath894 .
\(3 ) let @xmath895 . if @xmath896 , then @xmath897 * lemma 3.3.3 .
* _ let @xmath301 be a subset of @xmath898 .
suppose that @xmath899 and @xmath900 for all @xmath112 in @xmath301
. then @xmath901 .
_ we have @xmath835 .
now @xmath902 since @xmath900 for all @xmath112 , we necessarily have @xmath901 since @xmath903 .
the lemma is proved .
let @xmath560 be the dual partition of @xmath479 . by lemma 3.3.3 we know @xmath904
is contained in the two - sided cell @xmath380 corresponding to @xmath479 . using lemma 2.5.2 and the table in the proof of lemma 3.3.2 ,
it is easy to find the sequence of right star operations and @xmath112 for passing @xmath568 to @xmath905 .
now we determine the distinguished involution in a canonical left cell .
the involutions look complicated .
* lemma 3.4.1 . * _ suppose that @xmath822 is a distinguished involution and @xmath906 then _
@xmath907 for all @xmath783 .
_ \(b ) @xmath908 .
\(a ) according to the proof of [ lx , theorem 3.5 ] , we have @xmath909 .
the required assertion then follows from @xmath910 , @xmath911 and 3.3 ( b ) .
\(b ) since @xmath912 , we have @xmath913 . using the reduced expressions in 3.1 we see that ( b ) is true .
recall that @xmath15 is the subgroup of @xmath1 generated by all simple reflections @xmath914 see 2.1.3 ( c ) .
the lemma is proved . * theorem 3.4.2 .
* _ the following elements are all distinguished involutions contained in canonical left cells . _
( a ) ( b ) _ @xmath915 , _ where @xmath916 satisfies that @xmath917 for @xmath918 , ( we set @xmath919 and @xmath920 .
the associated partition is @xmath921 where @xmath7 appears @xmath922 times , @xmath923 appears @xmath924 times for @xmath925 , and 1 appears @xmath926 times , where @xmath927 .
( c ) _ @xmath928 , _ where @xmath929 satisfies that @xmath917 for @xmath930 , ( we set @xmath919 and @xmath931 for @xmath932 , and @xmath933 if @xmath934 , @xmath935 if @xmath936 .
the associated partition is @xmath937 where @xmath7 appears @xmath922 times , @xmath923 appears @xmath924 times for @xmath938 , @xmath939 appears @xmath940 times if @xmath934 , @xmath941 appears @xmath942 times for @xmath943 , 1 appears @xmath944 times if @xmath945 and appears @xmath946 times if @xmath936 , where @xmath947 .
\(a ) according to the proof of 3.3.2 , the action of @xmath948 on @xmath640 is given by the following table , @xmath949 set @xmath950 .
we define @xmath951 then we have @xmath952 for @xmath953 and all @xmath459 .
we now define an order among all the pairs @xmath954 for which @xmath955 are defined .
we say @xmath956 if one of the following three cases happens , ( 1 ) @xmath598 is even and @xmath957 is odd , ( 2 ) both @xmath958 have the same parity and @xmath959 , ( 3 ) both @xmath958 have the same parity and @xmath960 , @xmath961 .
it is easy to check that this defines a total order on the set of such pairs @xmath954 . arranging the pairs in the increasing order
, then we have a corresponding arrangement @xmath962 for all @xmath955 ( i.e. if @xmath963 and @xmath964 , then @xmath965 if and only if @xmath956 ) . for a fixed @xmath598
we denote by @xmath966 the cardinality of the set consisting of all @xmath955 . set @xmath967)c_j.\ ] ] we shall use @xmath968 for @xmath391 if @xmath969 and @xmath970 .
we also write @xmath971 ( @xmath972 ) for @xmath973 if the latter is defined . using lemma
2.5.2 we see that @xmath974 is well defined . moreover @xmath975 and @xmath976 . set @xmath977 .
suppose for @xmath978 ( @xmath979 ) we have defined @xmath980 with @xmath981 and @xmath982 .
then we define @xmath983 , here @xmath984 is chosen so that @xmath985 is well defined and @xmath986 and @xmath987 . by the conditions @xmath988 and @xmath989 , such @xmath990 exist .
continuing this process we finally get an element @xmath991 whose action on @xmath992 is given by @xmath993 , where @xmath994)$ ] if @xmath963 .
thus we have @xmath995)n\ ] ] if @xmath963 .
therefore we have @xmath996)n,\ ] ] where @xmath14 is defined by @xmath997 . applying the same sequence of right star operations @xmath998 ( in the same order ) to @xmath999 we get an element @xmath1000
whose action on @xmath992 is the same as the action of @xmath303 on , here @xmath301 is a suitable subset of @xmath992 , and @xmath303 is the longest element of the subgroup generated by all @xmath1001 ( @xmath1002 ) .
moreover , @xmath1003 is the partition in ( a ) . using prop .
1.4.6 we see that ( a ) is true .
\(b ) using lemma 3.3.1 and the proof of lemma 3.3.2 we can write down explicitly the action on @xmath640 of the element in ( b ) .
then as the proof of ( a ) we can see that the element in ( b ) is a distinguished involution and its associated partition is the given one in ( b ) .
the proof for part ( c ) is similar .
the theorem is proved .
the following elements are distinguished involutions , ( 1 ) @xmath1004 ( @xmath1005 ) , ( 2 ) @xmath1006 , ( 3 ) @xmath1007 , ( 4 ) @xmath1008@xmath1009 .
the associated partitions are , ( 1 ) ( 2, ... ,2,1, ... ,1 ) , where @xmath1010 appears @xmath112 times , ( 2 ) ( 3,3,1, ... ,1 ) , ( 3 ) ( 4,2, ... ,2,1, ... ,1 ) , where 2 appears @xmath1011 times , ( 4 ) ( n ) .
for later use we discuss the group @xmath589 and its representations in this chapter . in section 4.1
we give an explicit description for the group @xmath589 . in section 4.2
we give some facts about the representations of @xmath589 .
for completeness we also supply a few proofs although the facts are well known .
let @xmath1 be as in chapter 2 , that is , @xmath1 is the extended affine weyl group associated with @xmath3 .
for each two - sided cell @xmath520 of @xmath1 we have a corresponding partition @xmath479 of @xmath419 .
let @xmath523 be the dual partition of @xmath479 .
let @xmath64 be a unipotent element in @xmath3 whose jordan blocks are determined by the partition @xmath560 .
choose @xmath1014 such that \(1 ) @xmath1015 , \(2 ) for any @xmath1016 we have @xmath1017 for some @xmath459 .
let @xmath1018 let @xmath359 be the centralizer of @xmath64 in @xmath382 .
then the maximal reductive subgroup @xmath1019 of @xmath359 is isomorphic to @xmath1020 .
we shall identify @xmath589 with @xmath1021 .
\(1 ) if @xmath1022 , then @xmath1023 . in this case @xmath1024 .
\(2 ) if @xmath1025 , then @xmath1026 . in this case @xmath1027 .
\(3 ) if @xmath1028 , then @xmath1029 .
thus we have @xmath1030 if @xmath1031 .
we define @xmath1032 to be the set @xmath1033 .
the elements in @xmath1032 will be called dominant elements in @xmath1034 .
it is well known that the set irr@xmath3 of isomorphism classes of irreducible rational representations of @xmath3 is one to one corresponding to @xmath1032 .
thus the set irr@xmath589 of isomorphism classes of irreducible rational representations of @xmath589 is one to one corresponding to @xmath1035 for an element @xmath1036 in dom@xmath1037 we also call @xmath1038 the ( @xmath1039-component of @xmath1040 . for any algebraic group
@xmath121 we use @xmath1041 for its ( rational ) representation ring .
we are interested in the representation ring @xmath592 .
since @xmath589 is isomorphic to a direct product of the general linear groups @xmath1042 @xmath1043 , we see that @xmath592 is isomorphic to the tensor product ( over @xmath1044 of the representation rings @xmath1045 .
thus we are reduced to understand @xmath1046 .
the structure of @xmath845 is known . for our purpose
we list some properties of @xmath845
. we shall identify @xmath421 ( resp .
@xmath781 ) with @xmath1034 ( resp .
@xmath1032 ) by identifying @xmath452 with the element in @xmath1034 whose @xmath112th component is 1 and other components are 0 . for any @xmath786 , we denote by @xmath1047 an irreducible representation of @xmath3 with highest weight @xmath386 .
\(a ) as a ring @xmath845 is generated by @xmath1048 @xmath1049 , see 3.1 for definition of @xmath777 .
( a ) as a ring @xmath845 is generated by @xmath1050 and @xmath1051 for @xmath1052 .
\(b ) @xmath845 is a free @xmath328-module .
the elements @xmath1053 form a @xmath328-basis of @xmath845 , and the elements @xmath1054 also form a @xmath328-basis of @xmath845 .
( b ) the elements @xmath1055 form a @xmath328-basis of @xmath845 .
\(c ) for @xmath637 , the dimension of @xmath1056 is @xmath1057 .
\(d ) each weight space of @xmath1056 has dimension one and the weight set of @xmath1056 consists of all @xmath1058 @xmath1059 .
\(e ) let @xmath786 .
then @xmath1060 where @xmath1061 runs through the set of weights of @xmath1056 such that @xmath1062 .
_ let @xmath1063 . by ( d ) , @xmath1064 for any weight @xmath1061 of @xmath1056 . using the tensor product formula in [ h , ex .
24.9 ] we see that ( e ) is true .
\(f ) let @xmath2 be a ring .
assume that @xmath2 is a free @xmath328-module with a basis @xmath1065 .
if for all @xmath112 we have @xmath1066 , here @xmath1061 runs through the set of weights of @xmath1056 with @xmath1062 , then we have a ring isomorphism between @xmath845 and @xmath2 defined by @xmath1067 .
it follows from ( a ) , ( b ) and ( e ) .
\(g ) let @xmath786 and @xmath1068
. then @xmath1069 where @xmath1061 runs through the set of weights @xmath1070 of @xmath1051 with @xmath1071 for @xmath1072 .
( note that each weight space of @xmath1051 has dimension one and the weight set of @xmath1051 consists of all @xmath1073 @xmath1074 and @xmath1075 . )
it follows from littlewood - richardson rule .
when @xmath1076 , we also can see ( g ) using [ h , ex . 24.9 ] .
recently littelmann generalized the rule to arbitrary types , see [ li ] .
\(h ) let @xmath2 be a ring .
assume that @xmath2 is a free @xmath328-module with a basis @xmath1065 .
if for all positive integers @xmath7 we have @xmath1077 , here @xmath1061 runs through the set of weights @xmath1070 of @xmath1051 with @xmath1071 for @xmath1072 , then we have a ring isomorphism between @xmath845 and @xmath2 defined by @xmath1078 .
it follows from ( a ) , ( b ) and ( f ) .
in this chapter we will establish a bijection between @xmath1080 and the set of dominant weights of @xmath589 .
this bijection gives rise to a bijection between @xmath1080 and irr@xmath589 , that is in fact the bijection @xmath597 in conjecture 2.3.3 .
we use some r - antichains of elements in @xmath1080 to define the bijection . to do this
we show that each element in @xmath1080 has a complete r - antichain family , see 2.4.5 for definition .
the contents of this chapter is as follows . in section 5.1
we show that each element in @xmath1080 has a complete r - antichain family , see theorem 5.1.12 . in section 5.2
we define the map @xmath1081 from @xmath1080 to the set of dominant weights of @xmath589 by means of r - antichains of elements in @xmath1080 and give some discussions to the map .
the main result of this chapter is theorem 5.2.6 , which says that the map @xmath1081 is bijective . in section 5.3
we prove the main theorem by describing elements of @xmath1080 . in section 5.4
we give some simple properties of elements in @xmath1080 . in section 5.5
we give some elements in @xmath1080 , most of them correspond to fundamental weights . in next 3 chapters we will show that this bijection provides the isomorphism between @xmath563 and the rational representation ring @xmath592 .
let @xmath479 be a partition of @xmath419 . recall that we have defined the element @xmath488 in 2.2 and @xmath491 is the left cell of @xmath1 containing @xmath488 . in this section
we show that each element in @xmath1079 has a complete r - antichain family .
this is a rather delicate result since it does not hold even for some elements in @xmath1082 , see the example in 2.4.6 .
let @xmath487 and @xmath631 be its dual .
let @xmath1083 be as in 4.1 .
the numbers @xmath1083 can be defined in another way .
choose @xmath1084 such that @xmath1085 for @xmath1086 then @xmath1087 and @xmath1088 for all @xmath1089 ( we understand that @xmath1090 ) .
* lemma 5.1.1 . *
_ let @xmath1091 .
define @xmath1092 for @xmath1093 and @xmath1094 . then _
\(a ) _ @xmath1095 for @xmath1096 . _
\(b ) _ given @xmath1097 and @xmath1098 , we have @xmath1099 in particular , we have @xmath1100 _
\(c ) _ @xmath1101 for @xmath1102 and @xmath1103 . _ since @xmath1104 , @xmath1105 for some @xmath1106 with @xmath1107 . using lemma
2.5.1 we see that ( a ) is true .
\(b ) suppose that the assertion is not true .
then @xmath1108 thus @xmath1109 provides an r - chain family set of @xmath63 of index @xmath1110 .
the cardinality of the set is @xmath1111 since the lengths of the @xmath1110 r - chains of @xmath63 in the r - chain family set are @xmath1112 respectively .
this contradicts that @xmath1113 .
( b ) is proved .
similarly we prove ( c ) .
the lemma is proved .
* given @xmath1114 and @xmath1115 we set @xmath1116 and @xmath1117 we have @xmath1118 for all @xmath1119 . before going
further we give some discussions to complete r - antichian families .
let @xmath63 be in @xmath1079 and assume that @xmath63 has a complete d - antichain family ( see 2.4.5 for definition ) . by the argument for prop .
2.4.4 , we may decompose the set @xmath1120 into @xmath644 d - antichains @xmath1121 @xmath1122 of @xmath63 whose lengths are @xmath646 respectively . by lemma 5.1.1 and lemma 2.4.1
, any intersection @xmath1123 contains at most one element .
thus we have the d - antichain @xmath1124 of @xmath63 containing @xmath1125 has length @xmath1126 . the d - antichains
@xmath1127 form a complete d - antichain family @xmath635 of @xmath63 and @xmath1128 @xmath1129 form a complete r - antichain family of @xmath63 , see 2.4.5 for definition .
let @xmath635 be a complete r - antichain family of @xmath63 and @xmath1015 be as in section 4.1 .
then we have @xmath1130 .
thus @xmath635 contains @xmath1131 r - antichains of @xmath63 of length @xmath1132 .
let @xmath1133 be the r - antichains in @xmath635 of length @xmath1132 .
let @xmath1134 be elements in @xmath1135 , where @xmath1136 and @xmath1137 for all @xmath1138 and @xmath1139 .
* lemma 5.1.3 . *
_ let @xmath63 be in @xmath1080 and assume that @xmath63 has a complete r - antichain family .
keep the notation in 5.1.2 . for @xmath1139 ,
let @xmath1140 be the set consisting of all @xmath1141 .
then @xmath1140 contains exactly one element of @xmath1142 if @xmath1138 and contains no element of @xmath1142 if @xmath1143 .
_ using corollary 2.4.2 , any d - chain of @xmath471 contains at most one element of @xmath1140 .
but @xmath471 is in @xmath491 , so @xmath1144 is a d - chain of @xmath471 for @xmath1145 .
thus @xmath1146 contains at most one element for any @xmath1147 .
note that @xmath1140 contains @xmath1132 elements , @xmath1148 and @xmath1149 .
therefore @xmath1150 contains exactly one element of @xmath1144 for all @xmath1151 and @xmath1152 .
this forces that each @xmath1153 ( @xmath1154 contains exactly one element of @xmath1144 if @xmath1155 and contains no element if @xmath1156 .
inductively we see that @xmath1140 contains exactly one element of @xmath1142 if @xmath1138 and contains no element of @xmath1142 if @xmath1143 .
the lemma is proved .
* lemma 5.1.4 .
* _ let @xmath63 be in @xmath1080 and assume that @xmath63 has a complete r - antichain family .
keep the notation in 5.1.2 . given @xmath1157 and @xmath1139 , if @xmath1158 is in @xmath1159 , then _
\(a ) _ @xmath1160 is in @xmath1161 for all @xmath1162 , and @xmath1163 is in @xmath1164 if @xmath1165 . _
\(b ) _ @xmath1166 , and @xmath1167 . _
\(a ) let @xmath1168 be the unique number in @xmath1169 for @xmath1170 . then @xmath1171 .
we claim that @xmath1172 for some @xmath1173 if @xmath1174 . otherwise , we have @xmath1172 for some @xmath1174 and @xmath1175
. then we can find some @xmath1176 such that @xmath1177 .
since @xmath1178 and @xmath1179 we see that @xmath1180 thus @xmath1181 this is impossible since @xmath1135 is an r - antichain of @xmath63 .
so @xmath1172 for some @xmath1173 if @xmath1174 . for @xmath1182
, we then have @xmath1183 such that @xmath1184 and @xmath1185 .
we claim that @xmath1186 . otherwise , we have @xmath1187
. then @xmath1188 since @xmath1189 and @xmath1190 .
thus @xmath1191 .
this is impossible since @xmath1135 is an r - antichain of @xmath63 .
so we have @xmath1186 .
thus we have @xmath1192 if @xmath1193 .
similarly we see @xmath1194 if @xmath1195 .
\(b ) the assertion follows from the fact that @xmath1135 is an r - antichain and ( a ) .
the lemma is proved .
now we are going to show that each element in @xmath1079 has a complete r - antichain family .
the way is long .
we say that @xmath38 is positive if @xmath1196 is positive for any positive integer @xmath383 . by lemma 2.5.1
we only need to show that @xmath1197 for all @xmath1198 if the result is not true , we can find @xmath1199 such that @xmath1200 for @xmath1201 and @xmath1202
. then there exists @xmath7 in @xmath1203 such that @xmath1204 is in @xmath1142 for some @xmath1205 . note that @xmath471 is also positive since @xmath1206 whenever @xmath1207 . by lemma 5.1.1
we have @xmath1208 since @xmath7 is in @xmath1203 , by our assumption on @xmath63 , all the @xmath459 elements in the above sequences are contained in the union of @xmath1209 .
now @xmath1210 , we can find some @xmath1211 such that @xmath1212 contains at least two elements in the above sequence .
this contradicts corollary 2.4.2 since the above sequence is an r - antichain of @xmath471 and @xmath1212 is a d - chain of @xmath63 .
therefore @xmath1197 for all @xmath1119 .
the lemma is proved .
assume @xmath1213 and @xmath1214 .
we first show that @xmath1215 is in @xmath1142 .
we use descend induction on @xmath459 .
when @xmath1216 , by lemma 5.1.1 and assumption on @xmath7 we get @xmath1217 using corollary 2.4.2 we can see that @xmath1215 is in @xmath1218 and @xmath1219 is in @xmath1142 for @xmath1220
. now suppose that @xmath1221 is in @xmath1222 if @xmath1223 for some @xmath1224 . using lemma 5.1.1
we get @xmath1225 since @xmath1226 whenever @xmath1224 , using corollary 2.4.2 we see that @xmath1215 is necessarily in @xmath1142 .
now we show @xmath1227 if @xmath1213 .
if @xmath1228 for some @xmath1229 , we must have @xmath1227 since @xmath1226 and @xmath664 .
assume @xmath1213 and @xmath1230 .
we only need to show that the two sets @xmath1231 and @xmath1232 are equal .
if this is not true , then we can find @xmath1233 such that @xmath1234 and @xmath1235 .
since @xmath1226 if @xmath1236 , we have @xmath1237 for some @xmath1238 and integer @xmath1239 . since @xmath63 is positive , @xmath1239 is non - negative .
we have @xmath1240 since both @xmath1215 and @xmath1241 are in @xmath1242 and each number in @xmath1243 is greater than any number in @xmath1244 .
but @xmath1245 this contradicts that @xmath1246 .
the lemma is proved .
let @xmath1247 and @xmath1248 .
assume that @xmath1249 and @xmath1250 for all @xmath1251 . by lemma 5.1.1
we see that @xmath7 is @xmath1252 for some @xmath1132 , see the beginning of this section for the definition of @xmath1132 . since @xmath1104 and @xmath1253
, we can find a d - antichain family set @xmath1254 of @xmath63 of index @xmath1255 with cardinality @xmath1256 . by prop .
2.4.4 we may assume that @xmath1254 is included in @xmath640 .
let @xmath1257 be the d - antichains of @xmath63 in @xmath1254 .
since @xmath1142 is a d - chain of @xmath63 for any @xmath459 , using lemma 2.4.1 we see that @xmath1254 contains at most @xmath1255 elements of @xmath1142 for any @xmath459 .
since @xmath1254 has cardinality @xmath1256 , @xmath1254 contains exactly @xmath1255 elements of @xmath1142 if @xmath1138 and contains all elements in @xmath1142 if @xmath1258 .
thus each d - antichain @xmath1259 ( @xmath1260 ) contains exactly one element of @xmath1142 if @xmath1138 and we can find a d - antichain @xmath1261 ( @xmath1262 ) that contains @xmath1263 . for @xmath1264 ,
let @xmath1265 be the unique element in @xmath1266 .
then the chosen @xmath1265 s satisfy our requirement .
the lemma is proved .
let @xmath1267 be positive and @xmath1268 .
assume that @xmath1269 and @xmath1250 for all @xmath1270 .
we want to define a positive element @xmath64 in @xmath1080 with @xmath1271 .
set @xmath1272 for all @xmath459 .
choose @xmath1273 such that @xmath1274 but @xmath1275 if @xmath1276 .
then choose @xmath1277 for @xmath1278 such that @xmath1279 for @xmath1280 . according to lemma 5.1.8
, @xmath1281 ( @xmath1282 ) exists
. finally let @xmath1283 .
for simplicity we set @xmath1284 * lemma 5.1.10 . * _ keep the notation above . then _
@xmath1285 .
according to lemma 5.1.8 we can find @xmath1265 ( @xmath1286 ) in @xmath1142 such that @xmath1287 . by the definition of @xmath1288 we have @xmath1289 for @xmath1278 .
therefore @xmath1285 .
the lemma is proved .
keep the notation in 5.1.9 .
now we define @xmath64 by @xmath1290 clearly @xmath64 is positive and @xmath1271 .
* lemma 5.1.11 . *
_ let @xmath64 be as above .
then @xmath64 is in @xmath1080 .
_ we shall prove ( 1 ) @xmath1291 ( 2 ) @xmath1292 ( 3 ) @xmath1293 .
write @xmath1294 then @xmath1295 for some @xmath1296 , since by lemma 5.1.7 , @xmath1297 if @xmath1298 .
we also have @xmath1299 since @xmath1300 . by lemma 5.1.10 and the definition of @xmath1301
, we see that @xmath1302 form an r - antichain of @xmath63 .
note that @xmath1267 . by lemma 5.1.7 , @xmath1303 whenever @xmath1304 . using corollary 2.4.2
we see that each @xmath1305 contains exactly one @xmath1306 @xmath1307 . as in the proof of lemma 5.1.4
we get ( @xmath1308 if @xmath1309 is in @xmath1159 , then ( a ) @xmath1310 is in @xmath1161 for all @xmath1162 , and @xmath1311 is in @xmath1164 if @xmath1165 ; ( b ) @xmath1312 , and @xmath1313 .
\(1 ) by the definition of @xmath64 and using lemma 2.5.1 , to check @xmath1291 we only need to verify that @xmath1314 . if @xmath1309 is not in @xmath1315 , then @xmath1316 . in this case
we have @xmath1317 if @xmath1309 is in @xmath1315 , then @xmath1318 is in @xmath1252 and @xmath1319 . since @xmath1320 @xmath1246 and @xmath1321 is in @xmath1252 ( cf .
lemma 5.1.7 ) , we have @xmath1322 . therefore @xmath1323 we have proved that @xmath1324 \(2 ) now we show that @xmath1325 .
we have @xmath1326 let @xmath1327 be in the same @xmath1328 for some @xmath24 .
we need to show that @xmath1329 .
note that @xmath1330 since @xmath1246 . when both @xmath1331 are different from any @xmath1306
, we have @xmath1332 if @xmath1333 for some @xmath459 , then @xmath1334 for any @xmath1335 since @xmath1331 are in the same @xmath1328
. then we have @xmath1336 now suppose that @xmath1337 .
then @xmath1338 for any @xmath1339 since @xmath1331 are in the same @xmath1328 .
since @xmath1340 we have @xmath1341 where @xmath1248 and @xmath1342 .
thus @xmath1343 .
assume that @xmath1344 .
when @xmath1282 , we have @xmath1345 .
now suppose that @xmath1346 .
we have @xmath1347 we claim that @xmath1348 .
assume that @xmath1228 .
if @xmath1349 we of course have @xmath1348 .
now suppose that @xmath1350 . to see that @xmath1351 in this case it suffices to show that @xmath1352 since @xmath1353 are in @xmath1315 and @xmath1104 .
if @xmath1354 is in @xmath1252 , by @xmath1355 we have @xmath1356 .
in this case we have @xmath1357 since @xmath1358 . if @xmath1359 for some @xmath1360 , then @xmath1361 , so @xmath1362 . in this case
we also have @xmath1363 since @xmath1364 .
we have seen that @xmath1351 when @xmath1346 . noting that @xmath1365 ,
therefore if @xmath1346 we also have @xmath1366 assume that @xmath1367 . then @xmath1368 .
suppose @xmath1228 .
since @xmath1369 are in the same @xmath1328 , by @xmath1355 , we have @xmath1370 by the definition of @xmath1288 , we see that @xmath7 is not in @xmath1142 .
thus @xmath1371 . when @xmath1372 , we have @xmath1373 , so in this case @xmath1374 if @xmath1375 , since @xmath1376 and @xmath1104 , we have @xmath1373 .
in this case we also have @xmath1374 we have showed that @xmath1377 .
\(3 ) now we prove @xmath1378 . by ( 1 )
we see @xmath1379 .
if we can show @xmath1380 , then we are forced to have @xmath1378 .
let @xmath1381 be an r - chain family of @xmath64 ( see 2.2 for definition ) . if we can construct an r - chain family @xmath1382 of @xmath63 such that the cardinalities of @xmath1383 and @xmath1384 are the same , then we are done since it implies @xmath1385 .
note that @xmath1386 form an r - antichain of @xmath64 . according to corollary 2.4.2 , we have ( @xmath1387 ) each @xmath1388 contains at most one element of the set @xmath1389 since @xmath1390 form an r - chain family of @xmath64 , we have ( @xmath1391 ) for any @xmath1392 , at most one element of the set @xmath1393 is contained in the union @xmath1383 .
assume that @xmath1388 contains some element of @xmath25 if @xmath1394 and @xmath1388 does not contain any element of @xmath25 if @xmath1395
. then @xmath1396 are also r - chains of @xmath63 .
set @xmath1397 if @xmath1395 .
now we consider the r - chains @xmath1398 of @xmath64 .
assume that @xmath1399 ( @xmath1394 ) is contained in @xmath1388 .
it is no harm to assume @xmath1400 for @xmath1401 .
otherwise we may replace @xmath1388 by @xmath1402 .
we may further assume @xmath1403 .
let @xmath1404 be the elements in @xmath1388 .
we have two cases .
case 1 . for any @xmath1394 ,
if @xmath1405 exists , ( that is , @xmath1406 is not the greatest number in @xmath1388 , ) then @xmath1405 is in @xmath1407 .
we can find some @xmath1394 such that @xmath1405 exists and @xmath1405 is not in @xmath1407 .
set @xmath1408 .
assume that we are in case 1 . by ( @xmath1409 , the definition of @xmath1410 and lemma 5.1.10 ,
we see @xmath1411 if @xmath1412 exists .
let @xmath1413 ( @xmath1394 ) be the set consisting of the elements @xmath1414 note that @xmath1407 is a d - chain of @xmath64 . by the definition of @xmath64 we see that @xmath1415 are r - chains of @xmath63 .
clearly @xmath1416 form an r - chain family of @xmath63 and the cardinalities of @xmath1383 and @xmath1417 are the same .
now assume that we are in case 2 .
let @xmath459 ( @xmath1394 ) be the maximal such that either @xmath1405 does not exist or @xmath1405 exists and is not in @xmath1407 .
we claim that @xmath1388 is an r - chain of @xmath63 .
when @xmath1418 , this is clear since @xmath1419 is a d - chain of @xmath63 .
when @xmath1420 , we necessarily have @xmath1421 since @xmath1422 .
noting that @xmath1423 is positive and @xmath1424 if @xmath1270 ( cf .
lemma 5.1.7 ) , we see @xmath1425 since @xmath1426 is not in @xmath1315 when @xmath1426 exists . therefore @xmath1388 is also an r - chain of @xmath63 when @xmath1421 .
if @xmath1427 , we would like to construct two r - chains @xmath1428 of @xmath63 from @xmath1429 such that @xmath1430 and @xmath1431 have the same cardinality . if @xmath1432 does not exist or @xmath1432 is not in @xmath1433 when it exists , we set @xmath1434 and @xmath1435
now assume that @xmath1432 exists and is in @xmath1433 .
if @xmath1436 , we set @xmath1437 and @xmath1435 .
suppose that @xmath1438 and @xmath1432 is in @xmath1433 . if @xmath1405 exists and is in @xmath1433 , we choose @xmath1439 and @xmath1440 such that both @xmath1441 and @xmath1442 are in @xmath1443 but neither @xmath1444 nor @xmath1445 is in @xmath1433 .
we can move all elements in the intersection of @xmath1446 and @xmath1431 to @xmath1388 and form two sets @xmath1447 as follows .
if @xmath1448 , we set @xmath1449 @xmath1450 if @xmath1451 , let @xmath1452 be the set consisting of @xmath1453 @xmath1454 and let @xmath1455 be the set consisting of @xmath1456 suppose that @xmath1438 and @xmath1432 is in @xmath1433 .
if @xmath1405 does not exist or @xmath1405 exists but @xmath1405 is not in @xmath1433 , we choose @xmath1457 such that @xmath1441 is in @xmath1443 but @xmath1444 is not in @xmath1433 .
we can move all elements in the intersection of @xmath1446 and @xmath1458 to @xmath1388 and form two sets @xmath1447 as follows .
if @xmath1405 does not exist or @xmath1405 exists and @xmath1459 , we set @xmath1449 @xmath1450 if @xmath1405 exists and @xmath1460 , let @xmath1452 be the set consisting of @xmath1461 and let @xmath1455 be the set consisting of @xmath1462 it is easy to see that in any case @xmath1455 and @xmath1452 are r - chains of @xmath63
. moreover @xmath1430 and @xmath1463 have the same cardinality .
if @xmath1464 , for the pair @xmath1465 , we apply the same process , then we get two r - chains @xmath1466 of @xmath63 such that @xmath1467 and @xmath1468 have the same cardinality .
continuing this procedure , we get @xmath459 r - chains @xmath1469 of @xmath63 . from the construction we see that each of the r - chains @xmath1469 of @xmath63 contains some @xmath1470 with @xmath1471 . when @xmath1472 , we set @xmath1473 . when @xmath1474 , we set @xmath1475 .
now assume @xmath1476 . recall that we are in case 2 and @xmath1388 is also an r - chain of @xmath63 . when @xmath1427 , we have constructed @xmath459 r - chains @xmath1469 of @xmath63 .
if @xmath1421 we set @xmath1477 . by our assumption on @xmath459 , for any @xmath1478 , @xmath1479 exists and @xmath1479 is in @xmath1222 . if @xmath1480 or @xmath1423 is not in @xmath1481 , then let @xmath1482 ( @xmath1478 ) be the set consisting of the elements @xmath1483 and we set @xmath1473 .
assume that @xmath1484 and @xmath1423 is in @xmath1481 .
we construct two r - chains @xmath1485 of @xmath63 as follows .
note that @xmath1486 exists and is in @xmath1487 .
let @xmath1488 be the elements in @xmath1489 and @xmath1490 be the elements in @xmath1491 we have @xmath1492 .
note that @xmath1423 is positive .
we have @xmath1493 if @xmath1494 exists , since @xmath1481 is an r - chain of @xmath63 and @xmath1424 if @xmath1270 ( cf .
lemma 5.1.7 ) . if @xmath1495 exists and @xmath1496 is in @xmath1252 , we choose @xmath1497 and @xmath1498 such that both @xmath1499 and @xmath1500 are in @xmath1501 but neither @xmath1502 nor @xmath1503 is in @xmath1252 . if @xmath1504 , we set @xmath1505 @xmath1506 if @xmath1507 , let @xmath1508 be the set consisting of @xmath1509 and let @xmath1510 be the set consisting of @xmath1511 if @xmath1495 does not exist or @xmath1495 exists but @xmath1496 is not in @xmath1252 , we choose @xmath1512 such that @xmath1500 is in @xmath1501 but @xmath1503 is not in @xmath1252 . if @xmath1495 does not exists or @xmath1495 exists and @xmath1513 , we set @xmath1514 @xmath1506 if @xmath1495 exists and @xmath1515 , let @xmath1508 be the set consisting of @xmath1516 and let @xmath1510 be the set consisting of @xmath1517 it is easy to see that in any case @xmath1508 and @xmath1518 are r - chains of @xmath63 . moreover @xmath1519 and @xmath1520 have the same cardinality .
if @xmath1521 , for the pair @xmath1522 , we apply the same process as that for @xmath1523 ( in the case @xmath1427 ) , then we get two r - chains @xmath1524 of @xmath63 such that @xmath1525 and @xmath1526 have the same cardinality .
continuing this procedure , we get @xmath1527 r - chains @xmath1528 of @xmath63 . from the construction we see that each of the @xmath1527 r - chains of @xmath63 contains some @xmath1470 with @xmath1529 .
we set @xmath1530 and @xmath1531 . from the construction
it is clear that in any case @xmath1416 form an r - chain family of @xmath63 and the cardinalities of @xmath1383 and @xmath1384 are the same .
the lemma is proved .
now we are in the position to state the main result of this section .
let @xmath63 be an element in @xmath1080 .
we need show that @xmath63 has a complete r - antichain family .
we show the result first in the case when @xmath63 is positive ( see 5.1.5 for definition ) , and then in general .
suppose that @xmath63 is positive .
we use induction on the sum @xmath1532 .
when @xmath1533 , by lemma 5.1.6 , @xmath1534 . according to 2.4.6 , the result is true now suppose that @xmath1535 and the result is true if @xmath1536 is positive and @xmath1537
. we can find @xmath1132 and @xmath598 such that @xmath1269 and @xmath1538 whenever @xmath1539 .
let @xmath64 be as in lemma 5.1.11 .
then @xmath1536 is positive and @xmath1540 . by induction hypothesis
, @xmath64 has a complete r - antichain family .
keep the notation in 5.1.9 .
let @xmath635 be a complete r - antichain family of @xmath64 .
we may require that the r - antichain in @xmath635 containing @xmath1541 has length @xmath1132 for the reason explained below .
if @xmath1542 , then the r - antichain in @xmath635 containing @xmath1541 has length @xmath1132 , since @xmath1543 when @xmath1298 . if @xmath1544 , by lemma 5.1.7 , we have @xmath1545 if @xmath1546 . since @xmath1300 , by lemma 5.1.1 , we see @xmath1547
thus at least one r - antichain @xmath1254 in @xmath635 that contains @xmath1548 for some @xmath1549 and has length @xmath1132 .
let @xmath1244 be the r - antichain in @xmath635 containing @xmath1541 . according to lemma 5.1.4 , @xmath1550 and @xmath1551 are r - antichains of @xmath64 .
thus it is harmless to assume that the r - antichain @xmath1244 in @xmath635 containing @xmath1541 has length @xmath1132 .
assume that @xmath1552 and @xmath1553 is in @xmath1244 for all @xmath459 .
note that @xmath1554 form an r - antichain of @xmath64 and @xmath1555 if @xmath1298 .
let @xmath1413 be the r - antichain in @xmath635 containing @xmath1556 . according to lemma 5.1.4 and the assertion ( @xmath1308 in the proof of lemma 5.1.1 , the following sets are also r - antichains : @xmath1557 , @xmath1558 ( @xmath1282 ) . replacing @xmath1559 in @xmath635 by @xmath1560 , respectively
, we get a complete r - antichain family @xmath360 of @xmath64 such that the r - antichain @xmath1557 of @xmath64 is in @xmath360 . replacing @xmath1561 in @xmath360 by the r - antichain @xmath1562 of @xmath63
, we then get a complete r - antichain family of @xmath63 .
we have showed the result for positive elements in @xmath1080 . for any element @xmath63 in @xmath1080 ,
we can find a positive integer @xmath7 such that @xmath1563 is positive . noting that any d - antichain of @xmath1564 is also a d - antichain of @xmath63 , thus @xmath63 has a complete r - antichain family since @xmath1564 has one . the theorem is proved .
in next section we use this result to establish a bijection between @xmath1080 and dom@xmath1037 . in this section
we establish the bijection between @xmath1079 and dom@xmath1565 by means of complete r - antichain family .
let @xmath63 be in @xmath1079 . by theorem 5.1.12
, @xmath63 has a complete r - antichain family @xmath635 . in 5.1.2
we have seen that @xmath635 contains @xmath1566 r - antichains of @xmath63 of length @xmath1132 .
let @xmath1133 be the r - antichains in @xmath635 of length @xmath1132 .
let @xmath1134 be elements in @xmath1135 , where @xmath1136 and @xmath1137 for all @xmath1138 and @xmath1139 .
it is no harm to assume that @xmath1567 define @xmath1568 . from lemma 5.1.4 and our assumption on the arrangement @xmath1569
,@xmath1570 , we see @xmath1571 if @xmath1572 .
thus we have defined an element @xmath1573 we will show that @xmath1574 is independent of the choice of the complete r - antichain family @xmath635 . to do this ,
we construct a particular complete r - antichain family of @xmath63 in next subsection .
assume that @xmath1575 is in @xmath1034 .
we call that @xmath1254 is @xmath479-*admissible * if \(1 ) any two components of @xmath1254 are different , \(2 ) @xmath1576 for @xmath1577 , \(3 ) given @xmath1578 , @xmath1579 , we have @xmath1580 , in particular , we have @xmath1581 .
let @xmath1582 be a @xmath479-admissible element of @xmath1034 .
we define @xmath1583 inductively for all @xmath1584 and @xmath1585 , ( see the beginning of section 5.1 for the definition of @xmath1586 ) .
first we define @xmath1587 for all @xmath1588 .
let @xmath386 be the unique greatest number among all components of @xmath1254 .
then @xmath1589 for some @xmath1590 and @xmath1591 .
let @xmath1592 .
assume that we have defined @xmath1593 for some @xmath1594 .
we then define @xmath1595 to be the maximal number in @xmath1596 .
suppose that we have defined @xmath1597 for all @xmath1151 in @xmath1598 and @xmath1599 , @xmath1600 . here
@xmath1601 are some nonnegative integers , ( note that we have defined @xmath1602 for all @xmath1603 ) .
let @xmath1604 be obtained from @xmath1254 by removing all the components @xmath1597 ( @xmath1605 ) for some @xmath1606 .
let @xmath1607 be the greatest number among all components of @xmath1244 .
then it is easy to see @xmath1608 for some @xmath1609 .
suppose that we have defined @xmath1610 for @xmath1611 .
then we define @xmath1612 to be the maximal number in @xmath1613 then we define @xmath1614 for any @xmath1615 . continuing this way
we define all @xmath1583 for @xmath1616 and @xmath1617 .
. then @xmath1619 . and @xmath1620 @xmath1621 is @xmath479-admissible .
we have @xmath1622 @xmath1623 @xmath1624 let @xmath1267 .
set @xmath1625 for any @xmath1093 and @xmath1626 .
according to lemma 5.1.1 we see that @xmath1627 is @xmath479-admissible .
thus we can define @xmath1583 . to indicate its relation with @xmath63
, we shall denote it by @xmath1628 .
we would like to understand the properties of @xmath1628 .
* lemma 5.2.3 . *
_ let @xmath63 be in @xmath1080 . for given integers @xmath1590 and @xmath1139 , the numbers @xmath1629 form an r - antichain of @xmath63 of length @xmath1132 .
thus the r - antichains @xmath1630 of @xmath63 form a complete r - antichain family , we denote it by @xmath1631 .
_ for simplicity we write @xmath1583 instead of @xmath1628 in this proof .
let @xmath635 be a complete r - antichain of @xmath63 .
if @xmath1632 is maximal among all @xmath1633 , using 5.1.2 ( a ) , we see that the r - antichain @xmath1244 in @xmath635 containing @xmath1632 has length @xmath1132 .
moreover by the definition of @xmath1583 we see that the smallest number in @xmath1244 is not greater than @xmath1634 .
thus @xmath1635 so all @xmath1583 ( @xmath1138 ) form an r - antichain of @xmath63 of length @xmath1132 . in general , we consider the r - antichains of @xmath63 in @xmath635 .
let @xmath1636 @xmath1637 be two r - antichains of @xmath63 in @xmath635
. then @xmath1638 and @xmath1639 are in @xmath1144 for all @xmath1151 .
suppose @xmath1640 .
if @xmath1641 for some @xmath1642 , then @xmath1643 are two r - antichains of @xmath63 . replacing @xmath1644 by @xmath1645 in @xmath635 respectively , we then get a new complete r - antichain family @xmath1646 of @xmath63 . continuing this process ,
finally we get a complete r - antichain family @xmath1647 of @xmath63 with the following property . if @xmath1648 and @xmath1649 are two r - antichains of @xmath63 in @xmath1647 and @xmath1640 , then @xmath1650 for all @xmath1651 whenever @xmath1652 . by the definition of @xmath1583 we see that @xmath1653 is exactly an r - antichain in @xmath1647 .
the lemma is proved .
* lemma 5.2.4 . *
let @xmath1267 and @xmath635 be a complete r - antichain family of @xmath63 .
then the dominant weight @xmath1574 is independent of the choice of the complete r - antichain family @xmath635 of @xmath63 and only depends on @xmath63 .
thus we can denote the dominant weight by @xmath1654 and denote @xmath1655 by @xmath1656 .
_ let @xmath1657 be as in the proof of lemma 5.2.3 .
we claim that @xmath1658 and @xmath1659 . by lemma 5.1.4 ,
@xmath1660 and @xmath1661 are completely determined by @xmath1662 ) .
so we always have @xmath1658 . if @xmath1663 , then @xmath1664 ) is the maximal number in @xmath327 and @xmath1665 as well . by lemma 5.1.4
we see in this case @xmath1659 . if @xmath1666 , then @xmath1667 .
write @xmath1668 and @xmath1669 , where @xmath1670 and @xmath1671 . using lemma 5.1.4 for the r - antichains @xmath301 and @xmath1672 we can find @xmath459 such that both @xmath1673 are in @xmath1142 and then @xmath1674 . using lemma
5.1.4 we see @xmath1659 .
thus we have @xmath1675 .
the lemma is proved .
it seems that the number @xmath1656 has a strange property which now we are going to state . define @xmath1676 if @xmath1677 . *
lemma 5.2.5 . *
@xmath1678 _ if _
let @xmath1680 and @xmath1681
. then @xmath1682 therefore @xmath1683 .
the lemma is proved .
it is likely that @xmath1684 the following is the main result of this chapter .
* theorem 5.2.6 . *
_ let @xmath63 be in @xmath1080 . we have _
( a)@xmath1685 .
@xmath1686 .
\(b ) _ the map @xmath1687 defines a bijection between @xmath1080 and _ dom@xmath1037 .
\(a ) let @xmath1688 be an r - antichain of @xmath63 of length @xmath1132 consisting of @xmath1689 where all @xmath1306 are in @xmath1690 $ ] and @xmath1691 are in @xmath328 . using lemma 5.1.4 it is easy to see that all @xmath1692 form an r - antichain @xmath1561 of @xmath471 that is equivalent to @xmath1693 .
( see lemma 2.4.3 for the definition of equivalent r - antichains ) .
all such r - antichains @xmath1561 of course form a complete r - antichain family of @xmath471 . by definition
we have @xmath1694 by the definition of @xmath1654 and of @xmath1695 we see that ( a ) is true
. we will prove ( b ) in next section . before going further , we give two examples .
\(1 ) assume that @xmath1022 .
then the dual @xmath560 of @xmath479 is ( 1, ... ,1 ) and @xmath1696 is the longest element of @xmath1697 . in this case
we have @xmath1698 and @xmath1024 .
therefore dom@xmath1699 .
let @xmath63 be in @xmath1080 .
then @xmath63 has a unique complete r - antichain family , which consists of @xmath1700 .
assume that @xmath1701 , where @xmath1702 and @xmath776 .
then we have @xmath1703 which is in @xmath1704 \(2 ) assume that @xmath1031 and @xmath1028 .
then the dual @xmath560 of @xmath479 is ( @xmath453,1 ) and @xmath1705 . in this case
we have @xmath1706 and @xmath589 is isomorphic to @xmath1707
. therefore dom@xmath1708 .
we have @xmath1709 and @xmath1710 in this section we will give a proof of theorem 5.2.6 ( b ) . to do this we need construct elements of @xmath1712 .
more precisely for a given dominant weight @xmath1713 @xmath1714 in dom@xmath1037 we will construct an element @xmath1715 in @xmath1712 such that @xmath1716 .
then we show that the map @xmath1040 is a bijection from @xmath1712 to dom@xmath1037 .
first we assume that all @xmath1038 are nonnegative .
if @xmath1717dom@xmath1037 , then @xmath1715 is defined by @xmath1718 it is easy to see that @xmath1719 and @xmath1720 ( see 2.2 for the definition of @xmath1721 and @xmath1722 ) .
this forces that @xmath1723 .
it is easy to check that @xmath1715 is in @xmath1080 and @xmath1724 .
now suppose that @xmath1725 and we have defined @xmath1726 such that \(1 ) @xmath1716 , here @xmath1727 , \(2 ) @xmath1227 if @xmath1539 , and @xmath1728 if @xmath1729 .
\(3 ) all @xmath1730 whenever @xmath1731 .
then we define @xmath1732 for @xmath1733 as follows .
set @xmath1283 and choose @xmath1734 for @xmath1735 such that @xmath1736 for @xmath1735 , ( we set @xmath1737 ) .
then define @xmath1732 by @xmath1738 * lemma 5.3.1 . *
_ @xmath1732 is in @xmath1080 and @xmath1739 .
moreover , @xmath1545 if @xmath1539 and all @xmath1740 whenever @xmath1731 .
_ we shall prove ( 1 ) @xmath1291 ( 2 ) @xmath1292 ( 3 ) @xmath1741 and ( 4 ) @xmath1739 .
set @xmath1742 write @xmath1743 then @xmath1295 for some @xmath1296 , since by the assumption we have @xmath1303 if @xmath1298 .
we have @xmath1744 since @xmath1745 , where @xmath1746 and @xmath1747 .
it is easy to see that @xmath1302 form an r - antichain of @xmath63 .
note that @xmath63 is in @xmath1080 . since @xmath1303 whenever @xmath1748 , using corollary 2.4.2 we see that each @xmath1305 contains exactly one @xmath1306 @xmath1307 .
as in the proof of lemma 5.1.4 we get ( @xmath1308 if @xmath1309 is in @xmath1159 , then ( a ) @xmath1310 is in @xmath1161 for all @xmath1162 , and @xmath1311 is in @xmath1164 if @xmath1165 ; ( b ) @xmath1312 , and @xmath1313 .
\(1 ) by the definition of @xmath64 and using 2.1.3 ( f ) , to check @xmath1291 we only need to verify that @xmath1749 . when @xmath529 we need do nothing .
now assume that @xmath1750 , @xmath1751 is in @xmath1142 and @xmath1309 is in @xmath1752 . note that @xmath1753 . if @xmath1754 , then @xmath1755 .
otherwise , @xmath1756 .
this contradicts that @xmath1757 since @xmath1040 is in dom@xmath1037 . by @xmath1355
, we have @xmath1758 , where @xmath1759 if @xmath1760 and @xmath1761 if @xmath1762 . if @xmath1760 , then @xmath1763 . so @xmath1764 .
thus we have @xmath1765 in this case . if @xmath1762 and @xmath1427 , then @xmath1761 and @xmath1356 . in this case ,
@xmath1766 and @xmath1767 , we also have @xmath1765 . when @xmath1768 we have @xmath1761 and @xmath1769 .
we then have @xmath1770 since @xmath1246 , we see @xmath1766
. therefore @xmath1765 .
we have showed that if @xmath1754 then @xmath1765 .
now suppose that @xmath1771 .
then @xmath1772 and @xmath1773 .
we have @xmath1759 and @xmath1763 . if @xmath1774 , we have @xmath1766 , so @xmath1765 . if @xmath1775 , we have @xmath1770 since @xmath471 is in @xmath491 , thus @xmath1766 , so @xmath1765 .
so if @xmath1771 we also have @xmath1765 . we have showed that @xmath1765 is true .
so @xmath1776 .
\(2 ) now we show that @xmath1325 .
we have @xmath1777 let @xmath1327 be in the same @xmath1328 for some @xmath24 .
we need to show that @xmath1329 .
note that @xmath1330 since @xmath471 is in @xmath491 . when both @xmath1331 are different from any @xmath1306
, we have @xmath1332 if @xmath1778 for some @xmath459 , then @xmath1338 for any @xmath1335 since @xmath1331 are in the same @xmath1328 and each @xmath1328 contains at most one element of @xmath1779 .
then we have @xmath1780 now suppose that @xmath1781 .
then @xmath1334 for any @xmath1339 since @xmath1331 are in the same @xmath1328 and @xmath1328 contains at most one element of @xmath1779 .
since @xmath1340 we have @xmath1782 where @xmath1248 and @xmath1342 .
thus @xmath1783 .
if @xmath1784 , we have @xmath1785 for @xmath1786 . for @xmath1421 , we have @xmath1787 we claim that @xmath1788 .
otherwise , @xmath1789 , then we must have @xmath1790 since @xmath1227 for @xmath1791 and then @xmath1792 since @xmath1793 . by assumptions on @xmath63 , @xmath1794 , a contradiction , so @xmath1788 .
thus in this case we also have @xmath1795 we have showed that @xmath1796 if @xmath1784 . if @xmath1367 , then @xmath1797 . note that @xmath1798 . if @xmath1799 , we have @xmath1800 since @xmath1801 are contained in the same @xmath1328 and @xmath1802 is not in @xmath1328 and @xmath1803 . by the definition of @xmath1288
, we see that @xmath7 is not in @xmath1142 in this case .
if @xmath1346 , we also have that @xmath7 is not in @xmath1804 .
otherwise we have @xmath1805 .
then @xmath1806 , since @xmath1807 are in the same @xmath1328 , @xmath1797 and @xmath1808 .
this contradicts the assumption @xmath1809 in this case ) .
therefore in any case @xmath7 is not in @xmath1804 . since @xmath1797
, we see that @xmath7 is in @xmath1144 for some @xmath1810 .
if @xmath1811 , then @xmath1812 if @xmath1813 , then @xmath1814 are contained in @xmath1144 . since @xmath1815 and @xmath1816 , we have @xmath1817 thus @xmath1818 since @xmath1104 .
therefore @xmath1819 if @xmath1775 .
we have showed that @xmath1377 .
\(3 ) now we prove that @xmath1378 . by ( 1 )
we see @xmath1379 . if we can show that @xmath1820 , then we are forced to have @xmath1378 .
let @xmath635 be a complete r - antichain family of @xmath63 .
if @xmath1300 , by the assumptions on @xmath63 , we see that the r - antichain in @xmath635 that contains @xmath1821 has length @xmath1132 . if @xmath1822 , then @xmath1823 . by the assumptions on @xmath63 we see that @xmath1824
let @xmath1244 be the r - antichain in @xmath635 that provides @xmath1656 ( i.e. @xmath1825 and @xmath1244 has length @xmath1132 ) .
let @xmath1826 be the r - antichain in @xmath635 that contains @xmath1821 .
assume that @xmath1827 is in @xmath1244 . then by the assumptions on @xmath63
, we have @xmath1828 and @xmath1829 . replacing @xmath1827 in @xmath1244 by @xmath1821 and replacing @xmath1821 in @xmath1826 by @xmath1827
, we get a complete r - antichain family @xmath1646 of @xmath63 .
the r - antichain in @xmath1646 that contains @xmath1821 has length @xmath1132 .
thus it is harmless to assume that the r - antichain in @xmath635 that contains @xmath1821 has length @xmath1132 , let @xmath1830 @xmath1637 be two r - antichains in @xmath635 . note that @xmath1831 are contained in @xmath1142 .
assume that @xmath301 contains @xmath1821 .
then @xmath1832 and @xmath1833 .
write @xmath1834 where @xmath1835 and @xmath1836 .
suppose that for some @xmath1837 we have @xmath1838 we claim that ( 3a ) @xmath1839 are contained in the same @xmath1328 for some @xmath24 and @xmath1840 .
assume that @xmath1841 is in @xmath1328 .
since @xmath1842 , we have @xmath1843 . if @xmath1844 , using lemma 5.1.4 , then @xmath1845 and @xmath1846 .
thus @xmath1847 and @xmath1848 is in @xmath1328 or @xmath1849 . in this case
we have @xmath1850 since @xmath1246 , we must have @xmath1851 .
thus ( 3a ) is true if @xmath1852 .
now suppose that @xmath1853 . by lemma 5.1.4 ,
@xmath1854 and @xmath1855 .
thus @xmath1856 or @xmath1857 .
if @xmath1858 , then we must have @xmath1859 and @xmath1847 , since @xmath1842 .
but in this case we have @xmath1860 this contradicts that @xmath1246 .
therefore @xmath1856 . now @xmath1282 and by assumptions on @xmath63 we have @xmath1861 if @xmath1862 .
thus @xmath1848 is not in @xmath1222 for any @xmath1862 .
this forces that @xmath1863 since @xmath1864 and @xmath1856 . in conclusion ,
( 3a ) is true if @xmath1853 .
we have seen that ( 3a ) is always true .
thus the following two sequences @xmath1865 & \qquad\qquad>\cdots > w(\xi_{h-1})>\cdots > w(\xi_1),\\[4 mm ] & j ' : w(\e_t)>\cdots > w(\e_{k+1})>w(\xi_k)>\cdots > w(\xi_h)\\[4 mm ] & \qquad\qquad>\cdots > w(\e_{h-1})>\cdots > w(\e_1)\end{array}\ ] ] are two r - antichains of @xmath63 . replacing @xmath1644 in @xmath635 by @xmath1645 respectively
, we get a new complete r - antichain family of @xmath63 . continuing this process ,
finally we can find a complete r - antichain family @xmath360 of @xmath63 with the following two properties ( 3b ) the r - antichain @xmath301 in @xmath360 that contains @xmath1821 has length @xmath1132 , ( 3c ) let @xmath1866 , @xmath1867 , be two r - antichains in @xmath360 with @xmath1832 and @xmath1833 .
for any @xmath1868 , if @xmath1869 , then @xmath1870 . by the definition of @xmath1288 , we see @xmath1871 . replacing the r - antichain @xmath1872 in @xmath360 by @xmath1873
, we then get a set @xmath1874 consisting of @xmath644 r - antichains of @xmath64 .
the lengths of the r - antichains of @xmath64 in @xmath1874 are @xmath646 respectively .
moreover the union of all elements in the r - antichains of @xmath64 in @xmath1874 is @xmath1875 . by definition
we have @xmath1876 .
therefore we have @xmath1877 and @xmath1378 .
\(4 ) from the proof above we see clearly @xmath1739 . from the definition of @xmath64
we see @xmath1545 if @xmath1539 and all @xmath1740 whenever @xmath1731 .
the lemma is proved .
* lemma 5.3.2 . *
_ let @xmath63 be in @xmath1080 .
if all @xmath1878 are non - negative , then @xmath1879 for all @xmath1880 . _
otherwise , we can find some @xmath1880 such that @xmath1881 , where @xmath1248 and @xmath1882 is a positive integer .
let @xmath301 be an r - antichain of @xmath63 in a complete r - antichain family of @xmath63 that contains @xmath1883 .
let @xmath1884 ( @xmath1885 ) be the largest number in @xmath301 . since @xmath1886 for all @xmath507 , we have @xmath1887
. then @xmath1888 .
this contradicts that @xmath301 is an r - antichain of @xmath63 containing @xmath1883 .
the lemma is proved .
* lemma 5.3.3 . *
_ let @xmath1267 be such that all components of @xmath1656 are non - negative .
let @xmath1590 and @xmath1139 . if all @xmath1889 for @xmath1890 , or @xmath1891 and @xmath1892 , then @xmath1227 whenever @xmath1539 .
_ by lemma 5.3.2 , @xmath63 is positive . by the assumption on @xmath63
we see that @xmath1538 if @xmath1539 . using lemma
5.1.7 we see that @xmath1227 if @xmath1539 .
the lemma is proved .
let @xmath63 be in @xmath1893 .
suppose that @xmath1725 and all @xmath1894 are non - negative and we have @xmath1895 then we define @xmath64 for @xmath1896 as follows .
choose @xmath1273 such that @xmath1274 but @xmath1275 if @xmath1276 .
then choose @xmath1277 for @xmath1278 such that @xmath1279 for @xmath1280 .
( we set @xmath1272 for all @xmath459 ) . by lemma 5.1.8 ,
such @xmath1897 exist .
finally let @xmath1283 . for simplicity we set @xmath1284
now we define @xmath1898 by @xmath1899 actually we have defined this element in section 5.1 ( see lemma 5.1.11 ) . according to lemma 5.1.11 we get
@xmath64 is in @xmath1893 . from the proofs of lemma 5.1.11 and of theorem 5.1.2 we see clearly @xmath1900 .
let @xmath1901 be in dom@xmath1037 .
choose @xmath1902 such that @xmath1903 for all @xmath1904 .
by lemma 5.3.1 we can find @xmath1267 such that @xmath1905 then we have @xmath1906 and @xmath1907 therefore the map @xmath1040 is surjective .
suppose that @xmath1908dom@xmath1037 . by lemma 5.3.2
, @xmath63 is positive . by the definition of @xmath1654
, we see @xmath1909 if @xmath1248 .
thus @xmath1910 . using lemma
5.1.7 we get @xmath1534 .
now suppose that @xmath1911 and @xmath1912 suppose all @xmath1894 are nonnegative and @xmath1725 . by subsection 5.3.4
we can construct @xmath1913 from @xmath1914 respectively such that @xmath1915 we use induction on the sum of all components of @xmath1654 . by induction hypothesis
we see @xmath1916 .
now we can recover @xmath1914 from @xmath1913 using the construction at the beginning of this section , see lemma 5.3.1 .
therefore @xmath1917 if all components of @xmath1918 are nonnegative .
in general we can find @xmath459 such that all components of @xmath1919 ) are nonnegative .
thus @xmath1920 .
hence @xmath1917 if @xmath1918 .
we proved that the map @xmath1040 is injective .
theorem 5.2.6 ( b ) is proved . in the following section we give some simple properties of elements in @xmath1080 .
recall that we defined @xmath1921 for @xmath1922 , @xmath1923 , and @xmath1924 .
fix @xmath63 in @xmath1925 .
we write @xmath1926 , where @xmath1927 and @xmath1928 . *
lemma 5.4.1 . *
assume that @xmath1929 .
if both @xmath1930 and @xmath1931 are contained in the same @xmath1328 for some @xmath24 , then @xmath1932 . _ note that @xmath1933 since @xmath1934 .
thus we have @xmath1935since @xmath1246 and both @xmath1930 and @xmath1931 are contained in @xmath1328 , we see @xmath1932 .
* lemma 5.4.2 . * _ assume that @xmath1936 and @xmath1937 and @xmath1938 , where @xmath1939 . then @xmath1940
. moreover , if @xmath1941 then @xmath1942 .
since @xmath1943 we see @xmath1940 . if @xmath1941 , from @xmath1943 we get @xmath1942 .
* lemma 5.4.3 . * ( a ) @xmath1944 _ for some _ @xmath112 .
\(b ) @xmath1945 _ for some _ @xmath112 .
\(c ) @xmath1946 _ for some _ @xmath112 .
\(d ) @xmath1947 _ for some _ @xmath112 .
\(e ) _ if @xmath1948 is maximal among all @xmath1949 , then _ @xmath1950 _ for some @xmath598 .
_ \(a ) assume that @xmath1951 where @xmath1952 and @xmath1953 .
if @xmath1954 , we can find @xmath1880 such that @xmath1955 .
then @xmath1956 since @xmath1957 , see lemma 5.1.1 ( b ) .
thus @xmath1958 .
this contradicts @xmath1959 .
so we must have @xmath529 .
\(b ) applying ( a ) to @xmath471 we see that ( b ) is true .
\(c ) assume that @xmath1960 where @xmath1961 and @xmath1962 .
if @xmath1963 , we can find @xmath1880 such that @xmath1964 . since @xmath1246 , we have @xmath1965
. then @xmath1966 .
this contradicts @xmath1113 .
so @xmath1967 .
\(d ) it follows from @xmath1968 and ( c ) .
the proof of ( e ) is similar to that of ( a ) .
it follows from the construction of elements in @xmath1080 , see section 5.3 . to see it more clearly we use induction on the sum of all components of @xmath1654 as in section 5.3 .
when the sum of all components of @xmath1654 is 0 , we have @xmath1534 . in this case
the lemma is true .
now assume that @xmath1969 where @xmath972 , @xmath1970 and @xmath1971 .
let @xmath64 be in @xmath1080 such that @xmath1972 whenever @xmath1973 and @xmath1974 .
by induction hypothesis , we have ( @xmath1308@xmath1975 whenever @xmath1976 and @xmath1977 . according to lemma 5.3.1 and theorem 5.2.6
, we have @xmath1978 where @xmath1979 and @xmath1980 are inductively defined by @xmath1981 for @xmath1982 , ( we set @xmath1983 ) . using @xmath1355 we see that @xmath1984 @xmath1985 let @xmath1986 and @xmath1977 .
if both @xmath1987 and @xmath1988 are different from any ( @xmath1989 for @xmath1990 , we have @xmath1991 if @xmath1992 for some @xmath1993 and @xmath1988 is different from any ( @xmath1989 for @xmath1990 , noting that @xmath1994, ... ,@xmath1995 form an r - antichain of @xmath64 ( see the proof of lemma 5.3.1 ) , we have @xmath1996 suppose that @xmath1987 is different from any ( @xmath1989 for @xmath1997 and @xmath1998 for some @xmath1170 .
by @xmath1984 we have @xmath1999 .
now @xmath2000 , so @xmath2001 .
when @xmath2002 , we then have @xmath2003 when @xmath2004 , we have @xmath2005 finally suppose that @xmath2006 and @xmath2007 . by the relation between @xmath63 and @xmath64 we have
@xmath2008 by induction we see that the lemma is true . in this section
we write explicitly the elements in @xmath1080 corresponding to fundamental weights of @xmath589 .
let @xmath2009 ( @xmath1904 ) be in dom@xmath1037 such that its @xmath2010-component is 0 whenever @xmath2011 or @xmath1892 and its @xmath2012-component is 1 for @xmath2013 .
then @xmath2009 is the @xmath376th fundamental weight of @xmath589 . given @xmath2014 ,
let @xmath2015 be the element of @xmath1 defined by @xmath2016 for @xmath2017 , @xmath2018 , and @xmath2019 if @xmath2020(mod@xmath419 ) for all @xmath14 .
let @xmath1590 and @xmath1139 . for simplicity
we write @xmath1151 for @xmath2021 .
define @xmath2022 then @xmath2023 where @xmath1248 .
note that @xmath2024 for all @xmath2025 and @xmath2026 .
thus we have @xmath2027 where @xmath1248 .
recall that @xmath2028 for @xmath2029 and @xmath2030 . according to lemma 5.3.1
we see easily \(a ) @xmath2031 is in @xmath1080 .
\(b ) @xmath2032 . using 2.1.3 ( e ) and noting that @xmath2033 if @xmath2034 for some @xmath2035 ( cf .
2.1.3 ( f ) ) , we get \(c ) @xmath2036 and @xmath2037 .
let @xmath2038dom@xmath1037 be such that its @xmath2039-component is @xmath598 and other components are 0 . using lemma
5.3.1 we see \(d ) @xmath2040 is in @xmath1080 .
\(e ) @xmath2041 . using 2.1.3 ( e )
we see that @xmath2042 obviously we have @xmath2043 if @xmath2034 for some @xmath2035 ( cf .
2.1.3 ( f ) ) .
therefore we have \(f ) @xmath2044 and @xmath2045 . using the reduced expressions in 2.1.3 ( d ) we get \(g ) a reduced expression of @xmath2046 is @xmath2047 where @xmath2048 means that @xmath2049 is omitted .
in this chapter we will establish a factorization formula in @xmath563 , see theorem 6.4.1 .
this formula is important for our proof of lusztig conjecture on based ring for type @xmath0 and is obviously motivated by the corresponding formula in @xmath592 , which says that each irreducible rational @xmath589-module is a tensor product of some irreducible modules of reductive components of @xmath589 . in section 6.1
we note that in some cases the integers @xmath2051 can be computed through the basis @xmath2052 instead of the basis @xmath2053 . in section 6.2
we compute the product @xmath2054 for some special @xmath2055 .
this computation is a key to our factorization formula . in section 6.3
we give some consequences of the computation in section 2 . in section 6.4
we prove the factorization formula .
in this section we show that in some cases the integer @xmath2051 can be computed through the product @xmath2056 instead of @xmath2057 , the latter is usually much more difficult to compute .
let @xmath479 and @xmath488 be as in section 2.2 . using induction on @xmath2058 we see \(a ) if @xmath2059 and @xmath2060 , then @xmath2061 for some @xmath2062 with @xmath2063 .
\(b ) if @xmath2064 and @xmath2060 , then @xmath2065 for some @xmath2062 with @xmath2063 . noting that if @xmath2055 are in @xmath353 , we can find @xmath2066 in @xmath1 such that @xmath2067 , @xmath2068 and @xmath2069 , @xmath2070 .
thus we can find bar invariant elements @xmath2071 such that @xmath2072 note that @xmath2073 let @xmath2074 then @xmath2075 is bar invariant , i.e. @xmath2076 .
we have @xmath2077 and then the degree of @xmath2078 is either bigger than or equal to @xmath2079 whenever @xmath2080 . if @xmath63 is in @xmath353 , we must have @xmath2081 and @xmath2082 .
write @xmath2083 note that @xmath2084 c_y\quad\text { for any $ x\in w$.}\ ] ] we see deg@xmath2085 for any @xmath63 since deg@xmath2086 .
moreover deg@xmath2087 if and only if deg@xmath2088 , and in this case the leading coefficients of @xmath2089 and @xmath2090 coincide .
for @xmath349 we have @xmath2091 therefore @xmath2092 for any @xmath2093 .
the lemma is proved .
it follows from the following fact @xmath2094 obviously when @xmath2095 are pairwise different , then @xmath2096 if @xmath2097 , and @xmath2098 is either 0 or a power of @xmath2099 .
_ we use induction on @xmath2100 .
when @xmath2101 , we have @xmath2102 the lemma is true in this case .
now suppose that the lemma is true when @xmath243 is replaced by @xmath2103 .
then we have @xmath2104 note that @xmath2105 for any sequence @xmath2106 .
thus we have @xmath2107 & = \sum_{k\ge k_1>k_2>\cdots > k_m\ge j+1}\xi^{k - j - m } \tt_{xk_1\cdots k_m}\tt_{s_j } \\[4 mm ] & = \sum_{k\ge k_1>k_2>\cdots > k_m\ge j+1}\xi^{k - j - m}(\xi \tt_{xk_1\cdots k_m}+\tt_{xk_1\cdots k_m j } ) \\[4 mm ] & = \sum_{k\ge k_1>k_2>\cdots > k_m\ge j}\xi^{k - j+1-m}\tt_{xk_1 \cdots k_m}. \end{array}\ ] ] the lemma is proved .
let @xmath2108 be the subgroup of @xmath1 generated by all simple reflections that appear in a reduced expression of @xmath488 .
we have @xmath2109 , @xmath2110 and @xmath2111 . using induction on @xmath2112
we can prove the lemma .
in next section we compute some @xmath2056 . in this section
we compute @xmath2056 for some special @xmath2055 in @xmath353 .
suppose that @xmath349 and @xmath2113 .
denote by @xmath2114 the element in dom(@xmath2115 whose @xmath376-component is @xmath1656 for @xmath2116 and other components are 0 . then @xmath2117 .
the main result of this section is when @xmath2118 , the proposition is proved in [ l1 ] .
now we suppose that @xmath2119 .
recall that @xmath2120 , @xmath2028 for @xmath2121 , @xmath2122 .
set @xmath1762 , see section 5.1 for definition of @xmath1132 .
let @xmath2123 where @xmath2048 means @xmath2049 is omitted . according to 5.5 ( a)-(c ) and 5.5 ( g ) we have @xmath2124 and @xmath2070 .
now we compute @xmath2125 . to avoid complicated subscripts we also use @xmath2126 for @xmath2127 for any @xmath38 .
set @xmath2128 for @xmath2129 , note that @xmath2130 ; @xmath2131 for @xmath2132 . for simplicity
we often use @xmath2133 for @xmath2134 .
since @xmath2135 for all @xmath2136 , according to lemma 6.1.3 we have @xmath2137 where @xmath1751 stands for @xmath2099 .
let @xmath2138 recall that @xmath2139 for all @xmath1114 and @xmath2140 .
we have \(a ) @xmath2141 for some @xmath2142 .
it is easy to see \(b ) @xmath2143 .
moreover @xmath2144 if and only if @xmath2145 for @xmath2146 .
\(c ) @xmath2147 for all @xmath2148 .
according to lemma 5.1.1 ( c ) we have @xmath2149 for @xmath2150 . for convenience we set @xmath2151 for all @xmath383 .
choose @xmath2152 such that @xmath2153 then @xmath2154 . set @xmath2155 then @xmath2156 .
moreover we have @xmath2157 for @xmath2158 .
thus , using lemma 6.1.3 we get @xmath2159 & = \displaystyle{\sum\xi^{m_r - k_r}\tt(u'_r j_{r , k_r } j_{r , k_r-1}\cdots j_{r,1 } ) \tt ( v_{r-1 } ) } , \end{array}\ ] ] where the sum is for all sequences @xmath2160 @xmath2161 .
set @xmath2162 we have \(d ) if @xmath2163 then @xmath2164 , in this case we set @xmath2165 . if @xmath2166 then @xmath2167 for some @xmath2168 .
note that in any case the @xmath2169 elements @xmath2170 are bigger than @xmath2171 , and the @xmath2169 elements @xmath2172 are less than @xmath2173 .
\(e ) @xmath2174 for all @xmath2175 suppose that we have defined @xmath2176 for all @xmath2177 such that \(f ) @xmath2178 for some @xmath2179 if @xmath2180 ; and @xmath2181 if @xmath2182 , in this case we set @xmath2183 , ( we understand that @xmath2184 ; ) \(g ) @xmath2185 for all @xmath2186 .
now we define @xmath2187 as follows .
choose @xmath2188 such that @xmath2189 set @xmath2190 then @xmath2191 and @xmath2192 for @xmath2193 . using lemma
6.1.3 we see @xmath2194 & = \displaystyle{\sum\xi^{m_c - k_c}\tt(u'_{c } j_{c , k_c } j_{c , k_c-1}\cdots j_{c,1})\tt ( v_{c-1 } ) , } \end{array}\ ] ] where the sum is for all sequences @xmath2195 @xmath2196 set @xmath2197 if @xmath2198 we can find @xmath2199 such that @xmath2200 ; if @xmath2201 we set @xmath2202 , and in this case we have @xmath2203 .
obviously we have \(h ) the @xmath2204 elements @xmath2205 are bigger than @xmath2206 , and the @xmath2204 elements @xmath2207 are less than @xmath2208 .
\(i ) @xmath2209 for all @xmath2210 . in this way we defined @xmath2211 for all @xmath2212 . from ( i ) we see \(j ) @xmath2213 if @xmath2214 and @xmath2215 . by lemma 6.1.3
we have @xmath2216 & \qquad \displaystyle{\times \tt(u_{h+1 } \prod_{h\ge c\ge 2}(j_{c , k_c } j_{c , k_c-1}\cdots j_{c,1 } ) ) . }
\end{array}\ ] ] write @xmath2217 we are concerned with the degree of @xmath2218 . from the construction above we see \(k ) the degree of @xmath2218 is @xmath2219 we have @xmath2220 now we consider the degree of @xmath2221 .
note that @xmath2222 for some @xmath2223 whenever @xmath2224 . obviously we have \(l ) @xmath2225 for @xmath2224 . according to lemma 5.4.4
we have \(m ) @xmath2226 if @xmath2224 .
note that if @xmath2227 then @xmath2228 are in @xmath2229 .
thus , \(n ) among @xmath2228 at least @xmath2230 of them are less than @xmath2231 .
\(o ) suppose that among @xmath2232 we have that @xmath2233 of them are less than @xmath2234 .
since @xmath2235 are in @xmath2236 we see at least @xmath2237 of @xmath2235 are less than @xmath2238 . using ( h ) and
( o ) we get an r - chain of @xmath64 consisting of @xmath2239 & \ \ ... , \ u_r(e_{r-1}),\ u(a_{r , l_r+1}),\ ... , \
u(a_{rm_r}),\ u(a_{1l_1}+n),\ ... , \ u(a_{1\l_1}+n ) , \end{array}\ ] ] whose length is @xmath2240 if @xmath2241
then we have @xmath2242 this is impossible since the partition of @xmath64 is @xmath479 .
so we have ( p ) @xmath2243 write @xmath2244 such that @xmath2245 is the shortest element in @xmath2246 and @xmath2247 is in @xmath2108 .
according to ( n ) and ( o ) , we have \(q ) @xmath2248 using lemma 6.1.4 we see that the degree of @xmath2221 is less than or equal to @xmath2249 \(r ) since the degree of @xmath2090 is the maximal number in @xmath2250 we see that the degree of @xmath2090 is less than or equal to @xmath2251 the proposition is proved .
\(a ) _ @xmath2252 is equal to @xmath2253 for @xmath2254 . _
\(b ) @xmath2255 _ for @xmath2256 .
( note that @xmath1762 . ) _
\(c ) @xmath2257 _ for _ @xmath2258 .
( d ) @xmath2259 .
( e ) @xmath2260 and @xmath2261 for @xmath2262 .
\(a ) if @xmath2263 for some @xmath1239 , then @xmath2264 for some @xmath2265 thus @xmath2266 would be less than @xmath2267 from the proof in 6.2.2 we see that the degree of @xmath2090 then would be less than @xmath2079 .
( when @xmath2268 we can see that @xmath2269 using 6.2.2 ( l ) and 6.2.2 ( r ) . )
\(b ) by lemma 5.4.4 we have @xmath2270 .
suppose that @xmath2271 for some @xmath1239 with @xmath2272 .
then among @xmath2273 at least @xmath2274 of them are less than @xmath2275
. then @xmath2266 would be less than @xmath2276 from 6.2.2 ( l ) and 6.2.2 ( p - r ) we see that the degree of @xmath2090 then would be less than @xmath2079 .
therefore ( b ) is true .
\(c ) the proof is similar . if @xmath2277 for some @xmath2268 , then among @xmath2278 ... ,@xmath2279 at least @xmath2274 of them are less than @xmath2275 . as the proof of ( b ) , this is impossible . therefore ( c ) is true .
\(d ) it is clear from the proof of prop .
\(e ) when @xmath2262 , using ( c ) and lemma 5.4.4 , we see that @xmath2261 .
if @xmath2280 , then @xmath2281 of @xmath2235 are less than @xmath2238 .
thus @xmath2266 would be less than @xmath2276 by 6.2.2 ( q - r ) , this contradicts that @xmath2090 has degree @xmath2079 .
therefore @xmath2260 .
( e ) is proved .
the proof is completed .
\(a ) _ we have @xmath2282 for any @xmath2283 and @xmath2284 .
_ \(b ) _ suppose that @xmath2285
. then @xmath2286 _
\(c ) _ @xmath2287 _ \(d ) _ suppose that @xmath2288 .
we have @xmath2289 _
\(a ) from the proof of prop .
6.2.1 we see @xmath2290 . using lemma
2.5.1 we get ( a ) .
\(b ) using corollary 6.2.3 ( b)-(c ) we see that ( b ) is true .
\(c ) and ( d ) are clear from the proof of prop .
the corollary is proved .
let @xmath2055 be as in section 6.1 .
in this section we will figure out the @xmath63 in @xmath353 with @xmath2090 having degree @xmath2079 .
we keep the notation in 6.2.2 . in 6.2.2
we have showed that @xmath2291 when @xmath2292 , we have @xmath2293 therefore the degree of @xmath2090 is either less than or equal to @xmath2294 \ & + \l_{h}-1-k_{h}+\cdots+\l_2 - 1-k_2\\[3 mm ] \ & + l(w_\l)-(\l_{h-1}-l_{h})-\cdots-(\l_1-l_2 ) \\[3 mm ] < & l(w_\l)=a(w_\l ) .
\end{array}\ ] ] the lemma is proved .
\(a ) _ if @xmath2295 then @xmath63 is not in @xmath353 . _
\(b ) if @xmath63 is in @xmath353 , then @xmath2296 is equal to @xmath2297 if @xmath2298 .
\(a ) according to corollary 6.2.3 ( d ) we must have @xmath2299 thus the r - chain of @xmath63 , ( here we need corollary 6.2.4 and 6.2.2 ( h ) ) @xmath2300 & w(a_{h+1 , l_{h+1}}),\ \ ... , \ \
w(a_{h+1,m_{h+1}}),\\[2 mm ] & \qquad\quad ... , \\[2 mm ] & w(a_{r , l_r}),\ \ ... , \ \ w(a_{r , m_r } ) , \ \
\end{array}\ ] ] has length @xmath2301 we have @xmath2302 by corollary 6.2.4 ( a ) , the sequence @xmath2303 is an r - chain of @xmath63 of length @xmath2304 if @xmath2227 . thus @xmath63 has an r - chain family set of index @xmath1151 and the cardinality of the r - chain family set is bigger than @xmath2305 .
therefore @xmath2306 .
\(b ) since @xmath2307 for @xmath2308 and all components of @xmath2309 are non - negative , using lemma 5.3.2 , theorem 5.1.12 and lemma 5.1.4 we see that @xmath2310 by ( a ) and corollary 6.2.3 ( c ) we get @xmath2311 using corollary 6.2.4 we then get @xmath2312 using lemma 5.4.4 for @xmath64 and corollary 6.2.4 we see clearly that @xmath2313 now we show that @xmath2314 we must have @xmath2315 since deg@xmath2316 . using lemma 5.4.4 and corollary 6.2.4 we see that ( 5 ) is true . using ( 1 ) , ( 3 - 5 ) , lemma 5.1.4 and lemma 5.2.3 , we see that @xmath2317 if @xmath2298 .
( b ) is proved .
the lemma is proved .
let @xmath2318 .
write @xmath2319 .
the * level * of @xmath7 is defined to be the pair @xmath2320 .
we say that @xmath2321 if @xmath2322 or @xmath2323 and @xmath961 .
let @xmath38 .
an r - antichain of @xmath63 is called * saturated * if the r - antichain is contained in some complete r - antichain family of @xmath63 .
analogously we define * saturated d - antichains * of @xmath63 .
suppose that the conclusion is not true .
then there exists some @xmath1239 with @xmath2288 and @xmath2324 such that some pair of neighboring terms in the sequence @xmath2325 have distinct levels but all @xmath2326 have the same level for all @xmath2327 with @xmath2328 .
let @xmath635 be a complete r - antichain family of @xmath64 .
for @xmath2329 we choose @xmath2330 such that @xmath2331 and @xmath2332 are in an r - antichain of @xmath64 that is contained in @xmath635 .
suppose that all @xmath2333 have the same level but @xmath2334 and @xmath2332 have distinct levels for some @xmath2335
. obviously we have \(a ) @xmath2336 suppose that @xmath2337 .
we claim that \(b ) @xmath2338 .
write @xmath2339 then @xmath2340 for some @xmath2341 . by assumption ,
@xmath2342 and @xmath2343 have the same level for all @xmath2344 and @xmath2026 . by lemma 5.1.4
, then we may assume that the r - antichain @xmath1826 in @xmath635 containing @xmath2345 has length @xmath1239 and there is a saturated r - antichain of @xmath63 that contains @xmath2346 and has length @xmath1239 .
note that @xmath2347 is in @xmath1826 and @xmath2348 ( since @xmath2337 ) .
using lemma 5.1.4 , then we can find a saturated r - antichain of @xmath63 that contains @xmath2349 and has length @xmath2350 since @xmath2345 and @xmath2346 have distinct levels .
we claim that @xmath2351
. otherwise , @xmath2352 .
write @xmath2353 by lemma 5.1.4 , we have @xmath2354 .
but there is a saturated r - antichain of @xmath63 that contains @xmath2355 and has length @xmath2350 , by lemma 5.1.4 , this is impossible .
therefore @xmath2351 and @xmath2356 and @xmath2357 .
using lemma 5.1.4 we see that @xmath2355 and @xmath2358 have the same level .
write @xmath2359 and @xmath2360 then @xmath2361 and @xmath2362 .
since @xmath2363 , @xmath2364 , and @xmath2345 , @xmath2334 have the same level , we have @xmath2365 and @xmath2366 for some @xmath2367 if @xmath2368 . when @xmath2369 , clearly ( b ) is true . if @xmath2368 and @xmath2370 , ( b ) is also obviously true .
now suppose that @xmath2368 and @xmath2371 .
using lemma 5.4.2 we see that ( b ) is true . thus ( b ) is always true .
similarly if @xmath2372 have the same level and @xmath2373 have distinct levels and @xmath2374 for some @xmath2344 and @xmath2026 , we have @xmath2375 . if @xmath2372 have the same level but @xmath2376 and @xmath2346 have distinct levels and @xmath2377 , using @xmath2378 we can see that @xmath2375 .
thus in this case ( b ) is true .
note that @xmath2379 for all @xmath2330 .
thus @xmath2380 for all @xmath2330 since @xmath2381 .
using corollary 6.2.4 we see that the following elements @xmath2382 & w(a_{h+1,l_{h+1}}),\ ... , \ w(a_{h+1,m_{h+1}}),\ \ ... , \ \ w(a_{c-2,l_{c-2}}),\ ... , \ w(a_{c-2,m_{c-2}}),\\[3 mm ] & w(a_{c-1,l_{c-1 } } ) , \ ... , \ w(a_{c-1,m_{c-1 } } ) , \ w(a_{c-1,\beta_{l_c}}),\
w(a_{c-1,\beta_{l_c+1}}),\ ... , \ w(a_{c-1,\beta_{\theta-1}}),\\[3 mm ] & w(a_{c,\theta-1}),\ ... , \ w(a_{cm_c } ) , \\[3 mm ] & w(a_{c+1,l_{c+1}}),\ ... , \ w(a_{c+1,m_{c+1}}),\ \ ... , \ \ w(a_{r-1,l_{r-1}},\ ... , \ w(a_{r-1,m_{r-1}}),\\[3 mm ] & w(a_{rl_r}),\ w(a_{r , l_r+1}),\ ... , \ w(a_{r , m_r}),\ w(a_{1,l_1 + 1}+n),\ ... , \ w(a_{1\l_1}+n ) , \end{array}\ ] ] form an r - chain of @xmath63 . since @xmath2383 the above r - chain of @xmath63 has length @xmath2384 we have @xmath2385 this is impossible since @xmath349 , so the lemma is true in this case .
suppose that @xmath2386 . by assumption
, we have \(c ) @xmath2387 and @xmath2388 have the same level for all @xmath2389 with @xmath2328 . using corollary 6.2.4 we get ( note that @xmath2390 in our case ) \(d ) @xmath2391 , and @xmath2342 and @xmath2343 have the same level for all @xmath2392 and @xmath2393 . and
@xmath2394 using lemma 5.1.4 , thus we have \(e ) there is a complete r - antichain family @xmath2395 of @xmath64 such that the r - antichain @xmath1244 in @xmath2395 containing @xmath2396 has length @xmath1110 ; \(f ) there is a complete r - antichain family @xmath2397 of @xmath63 such that the r - antichain @xmath1826 in @xmath2397 containing @xmath2398 has length @xmath1151 ; and the r - antichain @xmath2399 in @xmath2397 containing @xmath2400 has length @xmath1110 . by corollary 6.2.3 ( e ) ,
@xmath2401 for @xmath2402 .
using lemma 5.4.2 we see \(g ) @xmath2403 and @xmath2404 have distinct levels whenever @xmath2405 .
let @xmath1561 be an r - antichain in @xmath2395 .
suppose that @xmath1561 has length greater than @xmath1110 or of length @xmath1110 but does not contain @xmath2396 .
assume that @xmath2404 is in @xmath1561 for some @xmath1394 . by ( g ) and
( d ) , we can find some @xmath2406 that has the same level as @xmath2404 and in an r - antichain @xmath2407 in @xmath2395 of length less than the length of @xmath1561 .
replacing @xmath2404 by @xmath2406 in @xmath1561 and replacing @xmath2406 by @xmath2404 in @xmath2407 , and continuing this process , finally we get a complete r - antichain family @xmath2408 of @xmath64 with the following property .
\(h ) any r - antichain in @xmath2408 of length greater than @xmath1110 or of length @xmath1110 but not containing @xmath2396 , does not contain any @xmath2404 for @xmath2402 .
moreover the r - antichain in @xmath2408 containing @xmath2396 has length @xmath1110 .
\(i ) it is harmless to require that @xmath2409
. then , for any r - antichain @xmath1561 in @xmath2395 that has length greater than @xmath1110 or has length @xmath1110 but does not contain @xmath2396 , replacing @xmath2410 by @xmath2411 if @xmath2387 is @xmath1561 , we get an r - antichain @xmath2412 of @xmath63 .
it is harmless to assume all such @xmath2412 are in @xmath2397 .
suppose that @xmath2413 are in @xmath1244 for @xmath2402 .
write @xmath2414 @xmath2415 by ( e ) , ( f ) and lemma 5.1.4 we get \(j ) @xmath2416 is in @xmath2417 for some @xmath2418 and @xmath2419 is in @xmath2420 . since @xmath2421 and they have distinct levels , by corollary 6.2.4 , if @xmath2422 then @xmath2423 is in @xmath2424 , if @xmath2425 then @xmath2423 is in @xmath2424 or in @xmath2426 . then in both cases , by ( i ) and ( f ) , @xmath2423 is not in any r - antichain in @xmath2397 of length greater than @xmath1151 . by lemma 5.1.4 , this is impossible . hence @xmath2427 and @xmath2428 .
but we have @xmath2429 , this contradicts that @xmath1244 is an r - antichain of @xmath64 .
therefore @xmath2396 and @xmath2430 have the same level .
the lemma is proved when @xmath2431 , this is lemma 6.3.2 ( b ) . using lemma 6.3.4 and corollary 6.2.4 we see that @xmath2432 and @xmath2433 have the same level if @xmath2434 and @xmath2435 .
therefore @xmath2436 if @xmath1890 .
the corollary is proved .
\(a ) _ @xmath2437 and @xmath2438 have distinct levels and @xmath2439 .
_ \(b ) _ @xmath2345 and @xmath2440 have distinct levels for @xmath2441 .
_ \(c ) _ there is a complete r - antichain family @xmath635 of @xmath64 such that the r - antichain @xmath1244 in @xmath635 containing @xmath2437 has length @xmath1151 .
( recall that @xmath2130 . ) _
\(a ) by lemma 6.3.2 ( a ) we have @xmath2439 .
if @xmath2438 and @xmath2442 have the same level then @xmath2443 for some @xmath2444 , @xmath2445 and some @xmath2446 .
then @xmath2447 for some @xmath2448 , and @xmath2449 for some @xmath2450 .
thus @xmath2451 .
this contradicts that @xmath349 .
therefore @xmath2438 and @xmath2442 have distinct levels .
\(b ) similarly we see that ( b ) is true .
\(c ) let @xmath2452 ( resp .
@xmath2453 ) be the set of elements in @xmath2454 ( resp .
@xmath2455 ) that have the same level as @xmath2442 . let @xmath635 be a complete r - antichain family of @xmath64 .
assume that ( c ) is not true .
by lemma 5.1.4 , then any r - antichain in @xmath635 that contains one element in @xmath2452 has length different from @xmath1151 . by corollary 6.2.3
( e ) and lemma 5.4.2 , @xmath2456 and @xmath2442 have distinct levels .
now @xmath2455 is the union of @xmath2457 and @xmath2456 .
thus @xmath2458 . by lemma 5.1.4 , this forces that @xmath2459 for some @xmath2011 .
this is impossible by corollary 6.3.5 .
so there must be some r - antichain in a complete r - antichain family of @xmath64 that hash length @xmath1151 and contains @xmath2442 .
suppose that @xmath2460 and @xmath2461
. then @xmath2462 .
( we understand that @xmath2463 ) let @xmath635 be a complete r - antichain family of @xmath64 and @xmath1244 an r - antichain in @xmath635 that has length @xmath1132 and provides @xmath2464 .
let @xmath2465 ) be in @xmath1244 for @xmath1401 .
we may choose @xmath2466 as big as possible .
by lemma 5.1.4 , then @xmath2467 and @xmath2404 have distinct levels . according to lemmas 5.4.4 and 5.2.3
, we see that @xmath2468 .
now suppose that @xmath2469 satisfies ( a ) and ( c ) in lemma 6.3.6 .
assume that @xmath1244 is an r - antichain in a complete r - antichain family @xmath635 that has length @xmath1132 and provides @xmath2464 .
then obviously we have @xmath2470 .
we need show that there exist unique @xmath2471 for @xmath2441 such that @xmath2472 for all @xmath2283 and the corresponding @xmath63 in 6.2.2 satisfies that ( 1 ) @xmath349 and ( 2 ) deg@xmath2316 .
keep the notation in 6.2.2 . if @xmath2473 , we choose @xmath2474 inductively so that ( 1 ) @xmath2475 ( 2 ) @xmath2345 and @xmath2346 have the same level but @xmath2440 and @xmath2345 have distinct levels , noting that if no such @xmath2252 we set @xmath2202 .
then set @xmath2476 .
let @xmath2477 be the r - antichain @xmath1244 in lemma 6.3.6 ( c ) .
we may require that @xmath2478 and @xmath2479 have distinct levels for @xmath2480 .
then we set @xmath2481 for @xmath2480 . according to the arguments in 6.2.2
, it suffices to prove the following three assertions .
\(a ) @xmath2482 .
\(b ) @xmath2483 for @xmath2262 .
\(c ) let @xmath2484 be such that @xmath2485 .
then @xmath2486 ( note that @xmath2487 .
so @xmath2488 . )
for all @xmath2254 we write @xmath2489 let @xmath635 be a complete r - antichain family of @xmath64 containing @xmath1244 .
now we argue for ( a ) . if @xmath2277 for some @xmath2268 , then there is an r - antichain @xmath1826 in @xmath635 such that @xmath1826 contains @xmath2490 and @xmath2491 for some @xmath2492 and @xmath2493 .
write @xmath2494 where @xmath2495 and @xmath2496 .
assume that @xmath2497 if @xmath2498 , by lemma 5.1.4 , then @xmath2499 and @xmath2500 .
since @xmath2501 , @xmath2440 and @xmath2345 have distinct levels , and @xmath2502 ( see lemma 5.4.4 ) , we see \(d ) either @xmath2503 and @xmath2504 , or @xmath2505 ; and either @xmath2506 and @xmath2507 , or @xmath2508 .
if @xmath1750 , using lemma 5.1.4 we see that @xmath2509 and @xmath2362 .
this contradicts ( d ) . if @xmath529 , then @xmath2505 . by lemma 5.3.2 , @xmath2510 ,
thus @xmath2511 , see section 5.2 for the definition of @xmath2512 . since @xmath2513 for @xmath2514 ,
the length of @xmath1826 is not less than @xmath1151 . by lemma 5.1.4 , then @xmath2515 .
but by lemma 5.1.4 , @xmath2516 .
hence by ( d ) we see @xmath2508 .
therefore @xmath2517 .
this contradicts that @xmath1826 is an r - antichain of @xmath64 . if @xmath2352 , then @xmath2518 and @xmath2519 .
thus @xmath2520 and @xmath2521 . since @xmath1826 is an r - antichain in @xmath635 , by lemma 5.1.4 we must have @xmath2522 , @xmath2523 and @xmath529 .
this contradicts that @xmath2345 and @xmath2490 have distinct levels .
thus if @xmath2277 for some @xmath2268 we would be led to a contradiction .
therefore ( a ) is true .
now we show ( b ) .
if @xmath2524 for some @xmath2256 , using lemma 5.4.2 , we see that @xmath2525 and @xmath2345 have distinct levels .
write @xmath2526 recall that @xmath2527 .
then \(e ) either @xmath2528 and @xmath2529 , or @xmath2530 . since @xmath2257 and @xmath2531 ,
@xmath2532 have distinct levels , if @xmath2533 , by lemma 5.1.4 , we have that @xmath2534 and @xmath2535 , or @xmath2536 .
this contradicts ( e ) .
therefore @xmath2352 and @xmath2530 .
then @xmath2537 implies that @xmath2538 .
moreover we have @xmath2539 since @xmath2531 , @xmath2532 have distinct levels and @xmath2540 .
then it is easy to see that the r - antichain @xmath2399 in @xmath635 containing @xmath2525 does not contain any of @xmath2541, ... ,@xmath2440 since none of them is in @xmath2542 or @xmath2543 . thus there is an r - antichain @xmath2544 in @xmath635 such that @xmath2544 contains @xmath2490 and @xmath2491 for some @xmath2492 and @xmath2545 . as before ,
write @xmath2546 , @xmath2547 , where @xmath2548 and @xmath2496 .
assume that @xmath2549 .
from the arguments above we see that @xmath2550 and @xmath2551 .
thus by lemma 5.1.4 we must have @xmath2522 and @xmath529 .
this is impossible since @xmath2440 and @xmath2345 have distinct levels .
we proved ( b ) .
now we prove ( c ) .
let @xmath384 be the set consisting of @xmath2552 @xmath2438 , and let @xmath377 be the union of the two sets @xmath2553 @xmath2554 and @xmath2555 } .
let @xmath1826 be an r - antichain of @xmath64 .
we shall show that if @xmath1826 contains some element in @xmath384 , then @xmath1826 contains exactly one element of @xmath377 , and vice versa .
let @xmath2556 .
then @xmath2557 . let @xmath1826 be the r - antichain in @xmath635 containing @xmath2558 .
assume that @xmath2559 .
we claim that @xmath2560 .
write @xmath2561 where @xmath2562 .
assume that @xmath2563 and @xmath2564 .
then either @xmath2565 and @xmath2566 , or @xmath2567 assume that @xmath2568 . by lemma 5.1.4 ,
then @xmath2569 and @xmath2570 for some @xmath2571 whenever @xmath2572 .
if @xmath2573 , then @xmath2574 . in any case
we have @xmath2575 since @xmath2576 .
so we have @xmath2560 . if @xmath2577 , then @xmath2578 , so we have @xmath2575 since @xmath2576 .
thus we also have @xmath2560 in this case .
now suppose that @xmath2579 and @xmath2580 has the same level as @xmath2171 and @xmath2399 is the r - antichain in @xmath635 containing @xmath2580 .
assume that @xmath2581 is in @xmath2399 .
we claim that @xmath2560 .
we may write @xmath2582 where @xmath2583 .
assume that @xmath2584 .
then either @xmath2585 , @xmath2586 and @xmath2587 ; or @xmath2588 , @xmath2589 , and @xmath2570 for some @xmath2571 . in any case
we have @xmath2575 .
hence @xmath2560 . obviously if an r - antichain in @xmath635 contains some @xmath2580 for some @xmath2590 and @xmath2580
has the same level as @xmath2171 , then @xmath1826 does not contains any @xmath2591 with @xmath2592 , since @xmath2593 .
now suppose that @xmath2560 and @xmath1826 is the r - antichain in @xmath635 containing @xmath2581 . we claim that either @xmath1826 contains some @xmath2558 for some @xmath2556 or @xmath1826 contains some @xmath2594 for some @xmath2590 and @xmath2594 has the same level as @xmath2171 . as before write @xmath2595 .
assume that @xmath2570 and @xmath2596 .
then either @xmath2597 and @xmath2598 , or @xmath2599 , since @xmath2581 and @xmath2442 have distinct levels .
since @xmath2600 for @xmath2601 , by lemma 5.1.4 we see that the length of @xmath1826 is either @xmath1151 or greater than @xmath1151 .
let @xmath2602 be in @xmath1826 , where @xmath2340 for some @xmath2341 and @xmath1747 .
assume that @xmath2599 .
suppose that @xmath2603 .
then @xmath2604 since @xmath2605 and @xmath2606 . by lemma 5.1.4 ,
@xmath2607 and @xmath2608 . a contradiction .
therefore we have @xmath2557 and we are done in this case .
assume that @xmath2597 and @xmath2589 . if @xmath2557 then we are done .
now suppose that @xmath2603 .
then we have two cases , ( 1 ) @xmath2604 and @xmath2606 , ( 2 ) @xmath2609 and @xmath2610 .
when @xmath2604 , since @xmath2611 , by lemma 5.1.4 we see that there is @xmath2612 such that @xmath2613 for some @xmath2614 with @xmath2615 and @xmath2616 .
thus @xmath2617 has the same level as @xmath2171 . when @xmath2609 and @xmath2610 , using lemma 5.1.4 we see there is some @xmath2612 such that @xmath2613 for some @xmath2618 with @xmath2615 and @xmath2619 .
therefore ( c ) is true .
the proposition is proved .
apply prop .
6.3.7 and use 1.3 ( b ) and note that @xmath2620 is commutative ( see theorem 2.3.2 ( b ) ) .
now we can prove the factorization formula .
we need a notation .
for @xmath2621 we define @xmath2622 to be the unique element @xmath270 in @xmath353 determined by @xmath2623 .
we show the result first in the case when @xmath63 satisfies @xmath2624 for all @xmath806 , and then in general .
_ assume that all @xmath2624 .
it is sufficient to prove that if @xmath2625 for @xmath2626 and @xmath2627 , then @xmath2628 let @xmath2629 be such that @xmath2630 for all pairs @xmath2010 except @xmath2631 .
let @xmath2632 . according to prop .
6.3.7 , we have ( a)@xmath2633 where @xmath64 is in @xmath353 such that ( 1 ) all components of @xmath2309 are nonnegative , ( 2 ) @xmath2634 , ( 3 ) @xmath2635 , and ( 4 ) @xmath2636 is given by the following formula @xmath2637 recall that @xmath2638 is an irreducible @xmath589-module of highest weight @xmath2309 . by 1.3 ( d ) , the positivity of the structural coefficients @xmath2639 s ( see 1.3 ( f ) ) and ( a ) we get \(b ) if @xmath2640 , then @xmath2641 for some @xmath2642 with ( 1 ) @xmath2643 for all @xmath806 , ( 2 ) @xmath2644 and ( 3 ) @xmath2645 .
moreover if @xmath2646 and @xmath2647 , then @xmath2648 .
clearly @xmath2649 is not zero if and only if ( 1 ) all components of @xmath2309 are nonnegative , ( 2 ) @xmath2634 , ( 3 ) @xmath2635 .
let @xmath2650 be such that @xmath2651 for all @xmath598 and other components of @xmath2652 are 0 .
then @xmath2653 for some @xmath2654 with @xmath2655 and with @xmath2656 . by prop . 6.3.7 , we have @xmath2657 .
hence @xmath2658 for some @xmath2659 with @xmath2660 and with @xmath2661 .
thus @xmath2662 by prop.6.3.7 , this forces that @xmath2663
. thus we have \(c ) @xmath2664 .
now we define a total order on the subset @xmath2665 of @xmath353 consisting of all elements @xmath64 in @xmath353 with ( 1 ) all components of @xmath2309 are nonnegative and are less than or equal to @xmath459 and ( 2 ) @xmath2634 .
let @xmath2666 be in @xmath2665 .
if @xmath2667 for @xmath2668 but @xmath2669 then we define @xmath2670 .
this of course introduces a total order on @xmath2665 .
we have \(d ) if @xmath2671 and @xmath2672 , then @xmath2673 .
otherwise , by ( b ) we see that @xmath2674 .
let @xmath2675 .
write @xmath2676 by prop .
6.3.7 , @xmath2677 . by ( c ) and
6.3.7 , @xmath2678 appears in @xmath2679 with nonzero coefficient @xmath2680 but @xmath2678 appears in @xmath2681 with zero coefficient .
thus @xmath2682 .
this contradiction shows that @xmath386 must be in @xmath2665 .
( for a given element @xmath2683 , we say that @xmath2684 appears in @xmath159 with coefficient @xmath2685 . )
now we claim that \(e ) @xmath2628 for all @xmath64 in @xmath2665 . by ( c )
we see that ( e ) is true for the maximal element in @xmath2665 .
assume that for any @xmath2686 with @xmath2687 we have @xmath2688 .
by ( c ) we may assume that @xmath64 is not maximal and we can find @xmath2689 such that @xmath2690 .
we have @xmath2672 for some @xmath2673 .
suppose that @xmath2691 .
since for @xmath2692 we have @xmath2693 , using ( b ) we get @xmath2694 . denote by @xmath2695 the element in @xmath2696 whose @xmath2012-component is 1 and other components are 0 .
consider the product @xmath2697 let @xmath2698 be determined by @xmath2699 , then @xmath2700 appears in @xmath2701 with coefficient 1 . by induction hypothesis
, @xmath2702 appears in @xmath2703 with nonzero coefficient . according to prop .
6.3.7 , this forces that @xmath2704 since @xmath2694 . if @xmath2705 or there is @xmath2706 such that @xmath2707 , then @xmath2708 appears in @xmath2701 with coefficient 1 , where @xmath2709 with @xmath2710 or @xmath2711 .
obviously @xmath2712 and @xmath2713 , by induction hypothesis , then @xmath2714 appears in @xmath2703 with coefficient 1 .
but this is impossible since by prop . 6.3.7 , if @xmath2715 and @xmath2716 appears in @xmath2703 with non zero coefficient , then @xmath2717 .
thus we must have @xmath2718 define @xmath2719 . if @xmath2720 or @xmath2721 but @xmath2722 , then @xmath2723 but @xmath2724 .
this is impossible . if @xmath2725 we have @xmath2726 since @xmath2727 . if @xmath2728 and @xmath2729 , we can prove directly that the assertion ( e ) is true in this case .
in fact , from @xmath2664 and by considering @xmath2730 we see that @xmath2731 if @xmath2732 .
then by considering @xmath2733 we see that @xmath2734 if @xmath2735 and @xmath979 .
thus by considering @xmath2736 we see that @xmath2628 in this case .
the theorem is proved when all @xmath2624 .
_ now @xmath349 is arbitrary .
we can find @xmath2737 such that @xmath2738 for all @xmath806 .
let @xmath2739 and @xmath2740 with @xmath2741 .
then @xmath2742 for all @xmath806 .
by step 1 we have \(f ) @xmath2743 .
obviously we have \(g ) @xmath2744 since @xmath2745 and @xmath2746 , by step 1 , we have \(h ) @xmath2747 , where @xmath2748 is determined by @xmath2749 . using 1.3 ( b ) and theorem 2.3.2 ( b ) we get \(i ) @xmath2750 .
obviously we have @xmath2751 . using 1.3 ( c ) and 1.3 ( f )
we see that @xmath2752 for all @xmath459 .
thus we have \(j ) @xmath2753 . according to step 1 we have
\(k ) @xmath2754 for all @xmath459 , where @xmath2755 .
note that @xmath2756 for all @xmath459 . using theorem 5.2.6 ( b ) , we get @xmath2757 for all @xmath459 . then obviously we have @xmath2758 . using ( f)-(k )
we get @xmath2759 & = \prod_{k=1}^p(t_{\th_1}t_{\th_2}\cdots t_{\th_{k-1}}t_{u_k}t_{\th_{k+1}}\cdots t_{\th_p}t_{\th_1^{-1}}t_{\th_2^{-1}}\cdots t_{\th_p^{-1}})\\[3 mm ] & = \prod_{k=1}^pt_{v_k}t_{\om^{-qn}w_\l}\\[3 mm ] & = t_{w_1}t_{w_2}\cdots t_{w_p}. \end{array}\ ] ] the theorem is proved .
let @xmath479 be as in section 2.2 and @xmath1131 be as in section 4.1 .
then @xmath1042 is the @xmath112th reductive component of @xmath589 .
let @xmath2760 and @xmath2761 be two elements in dom@xmath1037 such that @xmath2762 whenever @xmath2011 and @xmath2763 .
let @xmath2764 and @xmath2765 be two irreducible @xmath589-modules of highest weight @xmath1040 and @xmath2766 respectively .
then we have a simple formula for the product @xmath2767 in @xmath592 ( see 4.2 ( g ) ) . in this chapter
we will establish a multiplication formula in @xmath563 that is corresponding to the formula for @xmath2767 , see theorem 7.2.2 . to do this
we compute some product @xmath2768 in section 7.1 . then
in section 7.2 we prove our formula . in chapter 8 , using this formula and the factorization formula ( theorem 6.4.1 ) we show that conjecture 2.3.3 is true .
let @xmath2055 be in @xmath353 such that @xmath2627 and @xmath2769 and @xmath2630 whenever @xmath2770 .
in this section we will compute the product @xmath2056 provided that all components of @xmath2309 are non - negative .
we show that @xmath2090 have degree less than or equal to @xmath2079 for the @xmath2055 and all @xmath38 .
keep the notation in chapter 5 .
let @xmath2771 be the @xmath2046 in section 5.5 .
we have * lemma 7.1.1 . * _ let @xmath2029 .
suppose that @xmath2772 . then _
\(a ) @xmath2773 , \(b ) @xmath2774 .
\(a ) according to section 5.5 we have @xmath2775
using this we see easily that @xmath2776 and @xmath2777 . thus @xmath2778 we proved ( a ) .
( b ) follows from ( a ) .
* 7.1.2 . *
let @xmath64 be in @xmath353 such that ( 1 ) @xmath2634 and ( 2 ) all components of @xmath2779 are nonnegative .
let @xmath2780 . by 5.5 ( d - e )
, we have @xmath2629 , @xmath2781 if @xmath2770 and @xmath2769 .
now we compute @xmath2056 . recall that we have set @xmath2782 , @xmath1094 and @xmath2130 . by the construction in section 5.3 we see ( @xmath2783 ) @xmath2784 if either @xmath2785 or @xmath2786 for some @xmath459 with @xmath2787 . according to 5.5 ( g )
we get @xmath2788 set @xmath2789 let @xmath2790 .
suppose that @xmath2791 for @xmath2792 .
using 2.1.3 ( f ) we see \(a ) @xmath2793 for all @xmath2794 . according to @xmath1355 and lemma 6.1.3 ,
we have @xmath2795 using lemma 5.3.2 and 2.1.3 ( f ) we see that @xmath2796 thus we have @xmath2797 let @xmath2798 then for @xmath2799 we have @xmath2800 for some @xmath2801 and @xmath2802 for some @xmath2803 .
thus @xmath2804 for some @xmath2805 .
we have @xmath2806 let @xmath2807 . then @xmath2808 note that @xmath2809 .
thus we have @xmath2810 moreover we have @xmath2811 in particular we have \(e ) @xmath2812 if @xmath2813 for some @xmath2814 . for any @xmath38 we set @xmath2815 and @xmath1272 for all @xmath459 . for @xmath1394
choose @xmath2816 such that @xmath2817 using 2.1.3 ( f ) we get @xmath2818 let @xmath2819 .
then we have @xmath2820 as a consequence we have \(h ) @xmath2821 if @xmath2822 and @xmath2823 if @xmath2824 for some @xmath1394 .
moreover , @xmath2825 if @xmath2826 for some @xmath2814 .
thus we have @xmath2827 and ( here we need lemma 6.1.3 and recall that @xmath2126 is set to be @xmath2127 ) @xmath2828 we may check that @xmath2829 therefore we have @xmath2830 assume that @xmath2831 for @xmath2832 and @xmath2833 then @xmath2834 for some @xmath2835 .
we have @xmath2836 let @xmath2837 . then @xmath2838 note that @xmath2839 .
thus we have @xmath2840 moreover we have \(k ) @xmath2841 for @xmath2832 and @xmath2842 .
\(l ) @xmath2843 for all @xmath2844 , and @xmath2845 if @xmath2846 for some @xmath2847 . from the above discussion and
using lemma 7.1.1 we see that @xmath2848 where @xmath2849 has degree @xmath2850 using lemma 5.4.4 , ( d ) and lemma 5.1.1 ( c ) we see that \(n ) @xmath2851 for @xmath2852 , and @xmath2853 .
since @xmath2854 for @xmath2832 and @xmath2855 , we have \(o ) let @xmath2856 .
we have @xmath2857 if @xmath2858 , then @xmath2859 in this case we have @xmath2860 and @xmath2861 or @xmath2862 . using ( d ) , ( g ) and lemma 5.4.4 we see that ( o1 ) among @xmath2863 , at least @xmath2864 of them are smaller than @xmath2865 . similarly , if @xmath2866 , then we have ( o2 ) among @xmath2863 , at least @xmath2867 of them are smaller than @xmath2865 , and @xmath2868
. similarly for @xmath2869 we have ( o3 ) either among @xmath2863 , at least @xmath2864 of them are smaller than @xmath2865 , or among @xmath2863 , at least @xmath2867 of them are smaller than @xmath2865 and @xmath2868 , where we set @xmath2870 and @xmath2871 .
let @xmath386 be the element in @xmath2872 of minimal length ( see 6.1 for the definition of @xmath2108 ) .
then @xmath2873 for some @xmath2874 .
using ( n ) and ( o1-o3 ) we get \(p ) @xmath2875 .
we also have \(q ) @xmath2876 and @xmath2877 for all @xmath2878 .
\(r ) for any @xmath2879 we have @xmath2880 thus we see in the expression @xmath2881 , the degree @xmath2882 of @xmath2090 is at most @xmath2883 . using ( p ) we see that @xmath2884 using ( q ) and ( r ) we see that \(s ) if @xmath2885 then @xmath2886 are the neutral element of @xmath1 for all @xmath2878 , here we set @xmath2887 . in this section
we give a multiplication formula in @xmath338 , based on the computation in section 7.1 .
keep the notation in 7.1.2 .
assume that @xmath2888 and @xmath2889 .
for all @xmath1394 , set @xmath2890 and @xmath2891 from the construction in section 5.3 and the arguments in 7.1.2 we see that @xmath2090 has degree @xmath2079 and its leading coefficient is 1 .
now suppose that @xmath349 and @xmath2090 has degree @xmath2892 ) .
let @xmath825 be in @xmath353 such that @xmath2893 , @xmath2894 and all other components of @xmath2895 and @xmath2652 are 0 . according to prop .
6.3.7 we have @xmath2896 using prop .
6.3.7 and the positivity 1.3 ( f ) we see that @xmath2897 moreover , all @xmath2898 ( @xmath2899 are not greater than @xmath2900 . by lemma 5.4.4 , 7.1.2 ( n - p ) , @xmath2901 for @xmath2852 and @xmath2902 .
we claim that @xmath2903 .
otherwise , @xmath2904 for some @xmath1394 ( recall that @xmath2905 ) .
by @xmath1984 and the construction in section 5.3 , @xmath2906 can not contain both @xmath2907 and @xmath2908 ( we set @xmath2909 ) . by the arguments in 7.1.2 ( o ) we see that ( @xmath1308 @xmath2908 is in @xmath2910 ( we set @xmath2911 ) . using the construction in section 5.3
, we must have @xmath2912 .
a contradiction , so the assumption @xmath2904 is not true . in a complete similar way
we see that @xmath2913 .
since @xmath349 and @xmath2914 , we must have @xmath2915 . from the arguments in 7.1.2 we have @xmath2916 .
thus @xmath2917 for @xmath2918 .
the theorem is proved .
let @xmath825 be in @xmath353 such that @xmath2893 , @xmath2919 and all other components of @xmath2920 are 0 . according to prop .
6.3.7 we have @xmath2896 now considering @xmath2921 and using prop.6.3.7 and theorem 7.2.1 we see that the required result is true .
in this chapter we will prove conjecture 2.3.3 using the formulas in chapters 6 and 7 .
this completes our proof of lusztig conjecture for type @xmath0 . in section 8.1
we give some lemmas about multiplication in @xmath2620 . in section 8.2
we give a proof for conjecture 2.3.3 and give a summary of our proof of lusztig conjecture on based ring for type @xmath0 .
also we give a few comments about a possible geometric realization of @xmath2620 . in section 8.3
we consider the affine weyl group associated with the projective linear group @xmath6 . for the affine weyl group we show that lusztig conjecture on based ring should have a weak form since the algebraic group @xmath6 is not simply connected . in section 8.4 we show that lusztig conjecture on based ring is true for the extended affine weyl group associated with the special linear group @xmath4 . in this section
we establish some lemmas about multiplication in @xmath2620 .
let @xmath833 be the subset of @xmath353 consisting of all elements @xmath64 in @xmath353 with @xmath2914 .
let @xmath2924 ( 1@xmath2925 ) be in @xmath2926 such that ( 1)@xmath2927 \(2 ) other components of @xmath2928 are 0 .
then @xmath2924 is corresponding to the @xmath376-th fundamental weight of @xmath589 . using the bijection @xmath2929dom@xmath1037 we shall identify @xmath833 with the set @xmath2930 ( see 4.1 for the definition of @xmath2930 ) . under the identification
, we have @xmath2931 if @xmath2932 .
* lemma 8.1.1 . *
_ let @xmath2932 .
then we have @xmath2933 see section 6.4 for the definition of @xmath2934 .
_ when all components of @xmath1654 are nonnegative , it is entirely similar to the proof for theorem 6.4.1 .
here we need consider @xmath2935 first .
thus the lemma is true if all components of @xmath1654 are nonnegative .
note that @xmath2936 is in @xmath353 .
in general we can find a positive integer @xmath459 such that all components of @xmath2937 are nonnegative .
let @xmath2938 be in @xmath353 such that @xmath2939 .
note that all components of @xmath2940 are nonnegative and @xmath2941 . according to theorem 6.4.1 and its proof , and prop .
6.3.7 , we have @xmath2942 the lemma is proved .
* lemma 8.1.2 . *
_ let @xmath2932 .
recall that @xmath2943 .
we have @xmath2944 and @xmath2945 for all @xmath459 if @xmath2946 .
_ first we show that @xmath2947 . using lemma
8.1.1 we see that it is harmless to assume that all components of @xmath1654 are non - negative . using prop .
6.3.7 we see that @xmath64 is in @xmath833 if @xmath2946 .
since @xmath2948 appears in @xmath2949 with non - zero coefficient , considering @xmath2950 , by the positivity ( see 1.3 ( f ) ) and prop .
6.3.7 we see that @xmath2951 if @xmath2946 .
now we show that @xmath2952 .
let @xmath2953 .
using prop .
6.3.8 we see that @xmath2954 appears in @xmath2955 with non - zero coefficient .
now consider @xmath2956 . by lemma 8.1.1
we may assume that all components of @xmath1654 are non - positive . by the positivity and prop .
6.3.8 we see that @xmath2957 and @xmath64 is in @xmath833 if @xmath2958 .
multiplying both sides of @xmath2959 by @xmath2960 and using lemma 8.1.1 we see that @xmath2961 for all @xmath459 if @xmath2946 .
the lemma is proved .
* proposition 8.1.3 . *
_ let @xmath2932 .
then @xmath2962 , where @xmath64 runs through the set consisting of all @xmath2963 such that @xmath2964 for some sequence @xmath2965 .
see the proof of theorem 6.4.1 for the definition of @xmath2966 .
_ using lemma 8.1.1 we may and will assume that all components of @xmath1654 are non - negative .
we use induction on the sum @xmath2967 when @xmath2968 , the lemma is trivial . when @xmath2969 , the lemma follows from prop .
now suppose that the lemma is true when @xmath2970 .
we need show that the lemma then is true for @xmath2971 .
we define two bilinear forms , one for @xmath2972 , the subring of @xmath2620 spanned by all @xmath2973 , the other for @xmath2974 .
we define @xmath2975 and @xmath2976 where @xmath2977 stands for an irreducible module of @xmath1042 of highest weight @xmath2944 , recall that we identify @xmath833 with @xmath2930 .
let @xmath2978 .
we say that @xmath2979 ( @xmath2980 ) has the right @xmath243-component if @xmath2981 we say that @xmath2979 has the right form if it has the right @xmath243-component for all @xmath2982 . by prop .
6.3.7 and theorem 7.2.2 , we have \(a ) if @xmath2979 has the right @xmath243-component ( resp .
right form ) , then both @xmath2983 and @xmath2984 have the right @xmath243-component ( resp .
right form ) .
obviously we have \(b ) if @xmath2979 and @xmath2985 have the right @xmath243-component ( resp .
right form ) , then @xmath2986 has the right @xmath243-component ( resp .
right form ) .
we need to show that @xmath2987 has the right from .
let @xmath2988 be such that its @xmath459th component is 1 and other components are 0 . by induction hypothesis , we know that @xmath2989 has the right form , recall that we identify @xmath833 with @xmath2930 .
thus @xmath2990 ( convention : if @xmath2991 , then @xmath2992 for any @xmath2993 ; and if @xmath386 is in @xmath2994 but not in @xmath2995 , then @xmath2996 . )
multiplying @xmath2997 to both sides of ( c ) and using prop .
6.3.7 , we get @xmath2998 by ( d ) and lemma 8.1.2 , we see
\(e ) @xmath2999 appears in @xmath3000 with coefficient 1 ; and all the three elements @xmath3001 @xmath3002 , @xmath3003 appear in @xmath3004 with coefficient 1 .
we wish to show that @xmath3000 has the right form . by induction hypothesis
, @xmath3005 has the right form , so @xmath3006 has the right form .
we may use prop .
6.3.7 to expand the expression @xmath3006 .
using lemma 8.1.2 we see that @xmath3007 appears in @xmath3004 with coefficient 1 .
thus @xmath3008 has the right form . by ( d )
, @xmath3000 has the right form .
we shall use the lexicographical order on @xmath2994 with @xmath3009 .
now suppose that if all components of @xmath3010 are non - negative and @xmath3011 , then @xmath3012 has the right form .
we need show that @xmath2987 has the right form .
choose @xmath459 such that @xmath3013 but @xmath3014 for all @xmath3015 .
if @xmath1421 , we have showed that @xmath2987 has the right form .
so we may assume that @xmath1427 .
by induction hypothesis and ( a ) , @xmath3016 has the right form .
by induction hypothesis again , we see \(f ) @xmath3017 has the right form .
since @xmath3018 has the right form , by induction hypothesis , we get \(g ) @xmath3019 has the right form . using ( b ) and ( f - g )
, we see that @xmath2987 has the right form if all components of @xmath1654 are nonnegative .
the lemma is proved .
now we can prove conjecture 2.3.3 .
use theorem 6.4.1 , lemma 8.1.3 and 4.2 ( f ) . combining theorem 2.3.2 , theorem 5.2.6 and theorem 8.2.1
, we see that lusztig conjecture on based ring for type @xmath0 is true . here
we give a summary of our proof of lusztig conjecture on based ring for type @xmath0 .
let * c * be a two - sided cell of the extended affine weyl group @xmath1 associated with @xmath3 .
according to shi [ s ] and lusztig [ l3 ] , we have a corresponding partition @xmath479 of @xmath419 .
let @xmath560 be the dual of @xmath479 and @xmath3020 a unipotent element whose jordan form has partition @xmath560 .
then the centralizer of @xmath64 in @xmath3 is connected and so is a maximal reductive subgroup @xmath589 of the centralizer .
thus lusztig conjecture on the based ring @xmath357 says that @xmath333 is isomorphic to the @xmath3021 matrix algebra over the representation ring @xmath592 of @xmath589 , where @xmath3022 is the number of left cells of @xmath1 in @xmath495 . to prove the required result
we first show that @xmath333 is isomorphic to the @xmath566 matrix algebra over the based ring of the intersection of a left cell in * c * and its inverse , see theorem 2.3.2
. then we establish a bijection between the intersection and the set irr@xmath589 of isomorphism classes of rational irreducible modules of @xmath589 in chapter 5 , see theorem 5.2.6 . in chapters 6 - 8
we show that the bijection leads to an isomorphism between the based ring of the intersection and @xmath592 .
this completes our proof of lusztig conjecture for type @xmath3023 .
let @xmath64 be a unipotent element in @xmath3 whose jordan blocks are determined by the dual partition of a partition @xmath479 of @xmath419 .
let @xmath3024 be the variety consisting of all borel subgroups of @xmath3 that contain u. in [ x4 ] we show that there is an equivariant @xmath589-partition of @xmath3024 . using the partition we determine the equivariant @xmath2-group @xmath3025 . according to 3.2 ( b ) in [ x4 ] and theorems 2.3.2 , 5.2.6 , 8.2.1
, we know that as @xmath592-modules , the equivariant @xmath2-group @xmath3026 is isomorphic to @xmath333 , where @xmath332 is the two - sided cell of @xmath1 corresponding to @xmath479 .
one can define a convolution on @xmath3027 . in [ l10 ]
lusztig found some canonical bases for some equivariant @xmath2-groups .
it is likely that \(1 ) under the convolution the equivariant @xmath2-group @xmath3026 becomes an associative ring with 1 , \(2 ) there exists a canonical @xmath328-basis of @xmath3025 whose elements are one to one corresponding to the elements of the two - sided cell @xmath332 , \(3 ) the bijection between the canonical @xmath328-basis in ( 2 ) and @xmath332 leads to the ring isomorphism between @xmath3026 and @xmath333 .
\(4 ) the map from an affine hecke algebras to the based ring of a two - sided cell of the corresponding extended affine weyl group defined in [ l5 ] has a natural geometric explanation . in next section
we explain that lusztig conjecture on based ring can not be generalized to arbitrary extended affine weyl groups . in this section
we consider the projective linear group @xmath6 of degree n. the extended affine weyl group associated with @xmath6 is just an affine weyl group of type @xmath3023 .
we shall identify it with the affine weyl group @xmath15 in 2.1.3 ( c ) .
for each two - sided cell @xmath380 ( resp . left cell @xmath337 , right cell @xmath508 ) of the extended affine weyl group @xmath1 associated with @xmath3 , the intersection @xmath3028 ( resp .
@xmath3029 , @xmath3030 ) is a two - sided cell ( resp .
left cell , right cell ) of @xmath15 .
obviously , the based ring @xmath3031 of @xmath3032 is a subring of @xmath518 .
now assume that @xmath3033 and @xmath3032 is the lowest two - sided cell of @xmath15 , i.e. , the two - sided cell containing @xmath3034 .
the two - sided cell @xmath3032 contains two left cells .
let @xmath3035 be the left cell in @xmath3032 such that @xmath3036 .
then @xmath3037 @xmath3038 the reductive group corresponding to @xmath3032 is @xmath3039 .
suppose that there existed a finite @xmath3040-set @xmath360 and a bijection @xmath3041 the set of isomorphism classes of irreducible @xmath3040 vector bundles on @xmath363 such that the map @xmath364 defines a ring isomorphism ( preserving the unit element ) between @xmath3031 and @xmath3042 .
since @xmath3040 is connected , @xmath360 must be a trivial @xmath3040-set .
moreover , since @xmath3032 contains two left cells , this forces that @xmath360 contains two elements .
thus @xmath3042 is isomorphic to the @xmath3043 matrix algebra @xmath3044 over the representation ring @xmath3045 of @xmath3040 . as a consequence there should exist an element @xmath63 in @xmath3046 and an element @xmath64 in @xmath3047 such that @xmath3048 .
but this is impossible since we have @xmath3049 @xmath3050 and @xmath3051 therefore we can not find the required @xmath3040-set @xmath360 and map @xmath597 for the two - sided cell @xmath3032 of @xmath15 , so that lusztig conjecture on based ring can not be generalized to arbitrary extended affine weyl group .
however a weak form of the conjecture might be true which now we are going to state .
let @xmath3052 be the extended affine weyl group associated with a connected reductive algebraic group @xmath121 over @xmath122 and let @xmath380 be a two - sided cell of @xmath3052 .
denote by @xmath372 the reductive group corresponding to @xmath380 .
it is likely that for any left cell @xmath337 in @xmath380 we can find a finite transitive @xmath372-set @xmath3053 such that there exists a bijection @xmath3054 the set of isomorphism classes of irreducible @xmath3055 vector bundles on @xmath3056 such that the map @xmath364 defines a ring isomorphism ( preserving the unit element ) between @xmath417 and @xmath3057 .
now we show that the weak form of lusztig conjecture on based ring is true for @xmath15 .
let @xmath380 , @xmath525 , @xmath491 , @xmath550 be as in section 2.3 .
then @xmath3058 is a left cell of @xmath15 .
clearly we have \(a ) the map @xmath3059 in section 2.3 induces a bijection @xmath3060 . using theorem 2.3.2 ( a - b )
we get \(b ) the map @xmath3061 induces a ring isomorphism from @xmath3062 to @xmath3063 , where @xmath3064 is the left cell of @xmath15 containing @xmath488 .
\(c ) the based ring @xmath3063 is commutative .
recall that we also regard @xmath1 as a permutation group of @xmath328 ( see 2.1 ) .
then @xmath15 is a permutation group of @xmath328 consisting of all the permutations @xmath3065 that satisfy ( 1 ) @xmath3066 for all @xmath749 and ( 2 ) @xmath3067 thus we have \(d ) let @xmath63 be in @xmath1080 .
then @xmath63 is in @xmath3068 if and only if the sum of all @xmath1656 ( @xmath3069 is 0 .
let @xmath3070 be the quotient group of @xmath589 moduloing the center of @xmath3 .
then @xmath3070 is isomorphic to the reductive group corresponding to the two - sided cell of @xmath15 containing @xmath488 .
let dom@xmath3071 be the subset of dom@xmath1037 consisting of all elements @xmath3072 in dom@xmath1037 that satisfy @xmath3073 .
then we have \(e ) the map @xmath3074dom@xmath3071 is a well defined bijection .
\(f ) the set of isomorphism classes of rational representations of @xmath3070 is one to one corresponding to the set dom@xmath3071 . for @xmath3075dom@xmath3071 we shall use @xmath2764 to denote a rational representation of @xmath3070 with highest weight @xmath1040 .
according to the above discussion and using theorem 8.2.1 we get in next section we will show that lusztig conjecture on based ring is true for the extended affine weyl group associated with the special linear group @xmath4 . * 8.4.1 . * in this section we show that lusztig conjecture on based ring is true for the extended affine weyl group
@xmath3076 associated with the special linear group @xmath4 of degree @xmath419 .
it is known that @xmath3076 is a quotient group of the extended affine weyl group @xmath1 associated with @xmath3 .
the kernel is generated by @xmath439 .
for an element @xmath63 or a subset @xmath2 in @xmath1 we shall use @xmath3077 or @xmath3078 to denote its image in @xmath3076 .
the following property is obvious .
\(a ) let @xmath63 and @xmath64 be in @xmath1 .
then @xmath63 and @xmath64 are contained in the same left cell ( resp .
right cell , two - sided cell ) of @xmath1 if and only if @xmath3077 and @xmath3079 are contained in the same left cell ( resp .
right cell , two - sided cell ) of @xmath3076 .
let @xmath3080 be the hecke algebra of @xmath3076 over @xmath27 $ ] with parameter @xmath28 .
then @xmath3080 is a quotient algebra of the hecke algebra @xmath25 of @xmath1 .
the kernel is generated by all @xmath3081 .
let @xmath3082 be in @xmath3076 .
as for @xmath1 , we can define the laurant polynomial @xmath3083 by means of the kazhdan - lusztig basis of @xmath3080 and define the integer @xmath3084 using the laurant polynomial @xmath3085 .
let @xmath156 be in @xmath1 .
then we have \(b ) @xmath3086 .
\(c ) @xmath3087 .
let @xmath3088 be the based ring of @xmath3076 with basis elements @xmath3089 and structure constants @xmath3090 .
then @xmath3088 is a quotient ring of the based ring @xmath327 of @xmath1 .
the kernel is generated by all @xmath3091 .
let * c * be a two - sided cell of @xmath1 and @xmath3092 the two - sided cell of @xmath3076 corresponding to @xmath520 .
then @xmath3093 is a quotient ring of the based ring @xmath2923 of @xmath520 , the kernel is generated by all @xmath3091 , @xmath3094 .
similar conclusion is true for the based ring @xmath3095 , where @xmath3096 is a left cell in @xmath3097 .
let @xmath3098 and @xmath526 be as in section 2.3 .
the image in @xmath3076 of @xmath550 is denoted by @xmath3099 .
then @xmath526 induces a bijection from @xmath3100 to @xmath3101 .
we denote the induced map also by @xmath526 . according to the above discussion and theorem 2.3.2 we have * theorem 8.4.2 . * _
let @xmath3102 be the two - sided cell of @xmath3076 corresponding to a partition @xmath479 of @xmath419 and @xmath560 the dual partition of @xmath479 and @xmath380 the corresponding two - sided cell in @xmath1 .
_ \(a ) _ the map @xmath3103 induces a ring isomorphism from @xmath3104 to @xmath3105 . _
\(b ) _ the based ring @xmath3105 is commutative .
_ \(c ) _ the map @xmath3106 defines an isomorphism from the based ring @xmath3107 to @xmath3108 , the @xmath566 matrix algebra over the ring @xmath3109 . _ * 8.4.3 .
* let @xmath64 and @xmath589 be as in chapter 4 .
then @xmath64 is in @xmath4 .
let @xmath3110 .
then @xmath3111 is a maximal reductive subgroup of the centralizer in @xmath3112 of @xmath64 .
since @xmath3111 contains the derived group of @xmath589 , we have \(a ) the restriction to @xmath3111 of an irreducible rational representation of @xmath589 is irreducible .
any irreducible representation of @xmath3111 can be obtained in this way .
let dom@xmath3113 ) be as in section 4.1 and @xmath1132 be as in section 5.1 .
the following result is obvious from the definition of @xmath589 and @xmath3111 .
\(b ) let @xmath2760 be a dominant weight in dom@xmath1037 and @xmath2764 an irreducible representation of @xmath589 of highest weight @xmath1040 .
then the restriction to @xmath3111 of @xmath2764 is trivial if and only if there exists an integer @xmath459 such that @xmath3114 for all @xmath1089 and @xmath3115 .
we introduce an equivalence relation on dom@xmath1037 .
let @xmath3116 and @xmath3117 be in dom@xmath1037 .
we say that @xmath1040 and @xmath1751 are equivalent if there exists an integer @xmath459 such that @xmath3118 for all @xmath507 .
the equivalence class containing @xmath1040 will be denoted by @xmath3119 and the set of all equivalence classes in dom@xmath1037 is denoted by dom@xmath3120 . by ( b )
we get \(c ) for each element @xmath3119 in dom@xmath3120 we have an irreducible rational representation @xmath3121 of @xmath3111 with highest weight @xmath3119 .
the map @xmath3122 is a bijection from dom@xmath3120 to the set of isomorphism classes of irreducible rational representations of @xmath3111 .
let @xmath491 be as in section 2.3 and let @xmath63 and @xmath64 be elements in @xmath1712 .
it is easy to see that @xmath3123 if and only if @xmath1654 and @xmath2309 are equivalent .
thus for each @xmath3077 in @xmath3124 we have a well defined element @xmath3125 in dom@xmath3120 , where @xmath63 is a preimage in @xmath1 of @xmath3077 . using theorem 5.2.6 ( b )
we get using 8.4.3 ( a ) , 8.4.1 ( c ) , theorem 8.2.1 and theorem 8.4.4 we get combining theorems 8.4.2 , 8.4.4 and 8.4.5 we see that lusztig conjecture on based ring is true for the extended affine weyl group associated with @xmath4 . it is expected that the explicit knowledge on the based rings will have applications to understand the representations of hecke algebras of @xmath1 . also we can compute some @xmath3126 ( the coefficient of @xmath3127 in the kazhdan - lusztig polynomial @xmath42 ) using the explicit knowledge on the based rings , the details will appear elsewhere . c. greene , _ some partitions associated with a partially ordered set , _ j. of combinatorics theory ( a ) 20 ( 1976 ) , 69 - 79 .
humphreys , _ introduction to lie algebras and representation theory , _ gtm 9 , springer - verlag , 1972 .
d. kazhdan and g. lusztig , _ representations of coxeter groups and hecke algebras , _ invent .
53 ( 1979 ) , 165 - 184 .
d. kazhdan and g. lusztig , _ proof of the deligne - langlands conjecture for hecke algebras , _ invent .
87 ( 1987 ) , 153 - 215 .
n. iwahori and h. matsumoto , _ on some bruhat decomposition and the structure of hecke rings of p - adic chevalley groups , _ publ .
i.h.e.s . 25 ( 1965 ) 5 - 48 .
p. littelmann , _ a littlewood - richardson formula for symmetrizable kac - moody algebras , _ invent .
( 1994 ) , 329 - 346 .
g. lusztig , _ singularities , character formulas , and a @xmath24-analog of weight multiplicities , _ astrisque 101 - 102 ( 1983 ) , 208 - 227 .
g. lusztig , _ some examples of square integrable representations of semisimple @xmath3128-adic groups , _ trans .
( 1983 ) , 623 - 653 .
g. lusztig , _ the two - sided cells of the affine weyl group of type @xmath3129 , _ in infinite dimensional groups with applications " , springer - verlag , new york , 1985 , pp . 275 - 287 .
g. lusztig , _ cells in affine weyl groups , _ in algebraic groups and related topics " , advanced studies in pure math .
, vol . 6 , kinokunia and north holland , 1985 , pp .
255 - 287 .
g. lusztig , _ cells in affine weyl groups , ii , _ j. alg . 109 ( 1987 ) , 536 - 548 .
g. lusztig , _ cells in affine weyl groups , iii ,
tokyo sect .
34 ( 1987 ) , 223 - 243 .
g. lusztig , _ leading coefficients of character values of hecke algebras , _ proc . sympos .
pure math .
47 , part 2 , amer .
math.soc . , providence , r. i. , 1987 , 235 - 262 .
g. lusztig , _ cells in affine weyl groups , iv , _ j. fac .
tokyo sect .
36 ( 1989 ) no . 2 , 297 - 328 .
g. lusztig , _
periodic @xmath1-graphs , _ represent .
theory 1 ( 1997 ) , 207 - 279 .
g. lusztig , _ bases in equivariant @xmath2-theory , _ preprint , 1998 .
g. lusztig and n. xi , _ canonical left cells in affine weyl groups , _ adv . in math .
72 ( 1988 ) , 284 - 288 .
a. ram , _ calibrated representations of affine hecke algebras , _ preprint , 1998 .
a. ram , _ representations of rank two affine hecke algebras _ , preprint , 1998 .
shi , _ kazhdan - lusztig cells of certain affine weyl groups , _ lnm 1179 , springer - verlag , berlin , 1986 .
n. xi , _ the based ring of the lowest two - sided cell of an affine weyl group , _ j. alg . 134 ( 1990 ) , 356 - 368 .
n. xi , _ the based ring of the lowest two - sided cell of an affine weyl group , ii , _ ann .
ec . norm .
27 ( 1994 ) , 47 - 61 .
n. xi , _ representations of affine hecke algebras , _ lnm 1587 , springer - verlag , berlin , 1994 .
n. xi , _ a partition of the springer fibers @xmath3130 for type @xmath3131 and some applications , _ indag .
, n.s . , 10 ( 2 ) ( 1994 ) , 307 - 320 .
@xmath7-function1.2 affine weyl group , extended affine weyl group1.3 based ring1.5 cell ( left , right , two - sided)1.2 complete d - chain set , complete d - antichain set2.4 complete r - chain set , complete r - antichain set2.4 complete d - antichain family , complete r - antichain family2.4.5 d - chain , d - antichain2.2 d - chain family , d - antichain family2.2 d - chain family set , d - antichain family set 2.2 distinguished involution1.3 equivalent antichain2.4.3 hecke algebra1.1 kazhdan - lusztig basis1.1 kazhdan - lusztig polynomial1.1 level6.3.3 lusztig conjecture on based ring1.5 r - chain , r - antichain2.2 r - chain family , r - antichain family2.2 r - chain family set , r - antichain family set2.2 saturated d - antichain , saturated r - antichain6.3 star operation ( left , right)1.4 @xmath479-admissible5.2
|
in this paper we prove lusztig s conjecture on based ring for an affine weyl group of type @xmath0 .
= -2truecm
|
compressed sensing is a recently emerged technique for signal sampling and data acquisition which enables to recover sparse signals from much fewer linear measurements @xmath0 where @xmath1 is the sampling matrix with @xmath2 , @xmath3 denotes an @xmath4-dimensional sparse signal , and @xmath5 denotes the additive noise . such a problem has been extensively studied and a variety of algorithms , e.g. the orthogonal matching pursuit ( omp ) algorithm @xcite , the basis pursuit ( bp ) method @xcite , and the iterative reweighted @xmath6 and @xmath7 algorithms @xcite , were proposed . besides these methods , another important class of compressed sensing techniques that have received significant attention are bayesian methods , among which sparse bayesian learning ( also referred to as bayesian compressed sensing ) is considered as one of the most popular compressed sensing methods . sparse bayesian learning ( sbl )
was originally proposed by tipping in his pioneering work @xcite to address the regression and classification problems . later on in @xcite , sparse bayesian learning was adapted for sparse signal recovery , and demonstrated superiority over the greedy methods and the basis pursuit method in a series of experiments . despite its superior performance , a major drawback of the sparse bayesian learning method is that it requires to compute an inverse of an @xmath8 matrix at each iteration , and thus has computational complexity that grows exponentially with the dimension of the signal .
this high computational cost prohibits its application to many practical problems with even moderately large data sets . in this paper
, we develop a computationally efficient generalized approximate message passing ( gamp ) algorithm for sparse bayesian learning .
gamp , introduced by donoho _ et .
_ @xcite and generalized by rangan @xcite , is a newly emerged bayesian iterative technique developed in a message passing - based framework for efficiently computing an approximation of the posterior distribution of @xmath3 , given a pre - specified prior distribution for @xmath3 and a distribution for @xmath5 . in many expectation - maximization
( em)-based bayesian methods ( including sbl ) , the major computational task is to compute the posterior distribution of the hidden variable @xmath3 .
gamp can therefore be embedded in the em framework to provide an approximation of the true posterior distribution of @xmath3 , thus resulting in a computationally efficient algorithm .
for example , in @xcite , gamp was used to derive efficient sparse signal recovery algorithms , with a markov - tree prior or a gaussian - mixture prior placed on the sparse signal . in this work , by resorting to gamp , we develop an efficient sparse bayesian learning method for sparse signal recovery .
simulation results show that the proposed method performs similarly as the em - based sparse bayesian learning method , meanwhile achieving a significant computational complexity reduction .
we note that an efficient sparse bayesian learning algorithm was developed in @xcite via belief propagation .
the work , however , requires a sparse dictionary @xmath9 to facilitate the algorithm design , which may not be satisfied in practical applications .
we first provide a brief review of the sparse bayesian learning method . in the sparse bayesian learning framework ,
a two - layer hierarchical prior model was proposed to promote the sparsity of the solution . in the first layer , @xmath3 is assigned a gaussian prior distribution @xmath10 where @xmath11 is a non - negative hyperparameter controlling the sparsity of the coefficient @xmath12 .
the second layer specifies gamma distributions as hyperpriors over the hyperparameters @xmath13 , i.e. @xmath14 where @xmath15 is the gamma function . besides
, @xmath5 is assumed gaussian noise with zero mean and covariance matrix @xmath16 .
we place a gamma hyperprior over @xmath17 : @xmath18 .
an expectation - maximization ( em ) algorithm can be developed for learning the sparse signal @xmath3 as well as the hyperparameters @xmath19 . in the em formulation ,
the signal @xmath3 is treated as hidden variables , and we iteratively maximize a lower bound on the posterior probability @xmath20 ( this lower bound is also referred to as the q - function ) . briefly speaking , the algorithm alternates between an e - step and a m - step . in the e - step ,
we need to compute the posterior distribution of @xmath3 conditioned on the observed data and the estimated hyperparameters , i.e. @xmath21 it can be readily verified that the posterior @xmath22 follows a gaussian distribution with its mean and covariance matrix given respectively by @xmath23 where @xmath24 .
the q - function , i.e. @xmath25 $ ] , can then be computed , where the operator @xmath26 $ ] denotes the expectation with respect to the posterior distribution @xmath22 . in the m - step ,
we maximize the q - function with respect to the hyperparameters @xmath19 , which leads to the following update rules @xmath27 where @xmath28 denotes the expectation with respect to the posterior distribution @xmath22 .
it can be seen that the em algorithm , at each iteration , requires to update the posterior distribution @xmath22 , which involves computing an @xmath8 matrix inverse .
thus the em - based sparse bayesian learning algorithm has a computational complexity of order @xmath29 flops , and therefore is not suitable for many real - world applications with increasingly large data sets and unprecedented dimensions .
we , in the following , will develop a computationally efficient sparse bayesian learning algorithm via gamp .
generalized approximate message passing ( gamp ) is a very - low - complexity bayesian iterative technique recently developed @xcite for providing an approximation of the posterior distribution @xmath22 , conditioned on that the prior distribution for @xmath3 the distribution for the additive noise @xmath5 are factorizable .
it therefore can be naturally embedded within the em framework to provide an approximate posterior distribution of @xmath3 to replace the true posterior distribution . from the gamp s point of view ,
the hyperparameters @xmath19 are considered as known and fixed .
the hyperparameters can be updated in the m - step based on the approximate posterior distribution of @xmath3 .
gamp was developed in a message passing - based framework . by using central - limit - theorem approximations , the message passing between variable nodes and factor nodes
can be greatly simplified , and the loopy belief - propagation on the underlying factor graph can be efficiently performed . in general , in the gamp algorithm development , the following two important approximations are adopted .
let @xmath30 denote the hyperparameters .
firstly , gamp assumes posterior independence among hidden variables @xmath31 and approximates the true posterior distribution @xmath32 by @xmath33 where @xmath34 and @xmath35 are quantities iteratively updated during the iterative process of the gamp algorithm , here we have dropped their explicit dependence on the iteration number @xmath36 for simplicity . substituting ( [ hm-1 ] ) into ( [ eqn-1 ] ) , it can be easily verified that the approximate posterior @xmath37 follows a gaussian distribution with its mean and variance given respectively as @xmath38 the other approximation is made to the noiseless output @xmath39 , where @xmath40 denotes the @xmath41th row of @xmath9 .
gamp approximates the true marginal posterior @xmath42 by @xmath43 where @xmath44 and @xmath45 are quantities iteratively updated during the iterative process of the gamp algorithm , again here we dropped their explicit dependence on the iteration number @xmath36 . under the additive white gaussian noise assumption
, we have @xmath46 .
thus @xmath47 also follows a gaussian distribution with its mean and variance given by @xmath48 with the above approximations , we can now define the following two important scalar functions : @xmath49 and @xmath50 that will be used in the gamp algorithm . in the minimum mean - squared error ( mmse )
mode , the input scalar function @xmath49 is simply defined as the posterior mean @xmath51 @xcite , i.e. @xmath52 the scaled partial derivative of @xmath53 with respect to @xmath34 is the posterior variance @xmath54 , i.e. @xmath55 the output scalar function @xmath50 is related to the posterior mean @xmath56 as follows @xmath57 the partial derivative of @xmath58 is related to the posterior variance @xmath59 in the following way @xmath60 given definitions of @xmath49 and @xmath50 , the gamp algorithm can now be summarized as follows ( details of the derivation of the gamp algorithm can be found in @xcite ) , in which @xmath61 denotes the @xmath62th entry of @xmath9 , @xmath63 and @xmath64 denote the posterior mean and variance of @xmath12 at iteration @xmath36 , respectively .
* gamp algorithm * [ cols= " < " , ] .,width=264 ] we first examine the phase transition behavior of respective algorithms . the phase transition is used to illustrate how sparsity level ( @xmath65 ) and the oversampling ratio ( @xmath66 ) affect the success rate of each algorithm in exactly recovering sparse signals in noiseless scenarios . in particular , each point on the phase transition curve corresponds to a success rate equal to @xmath67 .
the success rate is computed as the ratio of the number of successful trials to the total number of independent runs .
a trial is considered successful if the normalized squared error @xmath68 is no greater than @xmath69 .
[ fig1](a ) plots the phase transitions of respective algorithms , where we set @xmath70 , and the oversampling ratio @xmath66 varies from @xmath71 to @xmath72 . from fig .
[ fig1](a ) , we see that , when @xmath73 , the proposed sbl - gamp algorithm achieves performance similar to sbl - em , and is superior to bp - amp .
the proposed method is surpassed by bp - amp as the oversampling ratio increases . nevertheless , sbl - gamp is still more appealing since we usually prefer compressed sensing algorithms work under high compression rate regions .
the average run times of respective algorithms as a function of the signal dimension @xmath4 is plotted in fig .
[ fig1](b ) , where we set @xmath74 and @xmath75 .
results are averaged over 10 independent runs .
we see that the sbl - gamp consumes much less time than the sbl - em due to its easy computation of the posterior distribution of @xmath3 , particularly for a large signal dimension @xmath4 .
also , it can be observed that the average run time of the sbl - em grows exponentially with @xmath4 , whereas the average run time of the sbl - gamp grows very slowly with an increasing @xmath4 .
this observation coincides with our computational complexity analysis very well .
lastly , we examine the recovery performance in a noisy scenario , where we set @xmath76 , @xmath77 , and the signal to noise ratio ( snr ) is set to 20db . fig .
[ fig2 ] depicts the normalized mean square errors ( nmse ) of respective algorithms vs. @xmath66 .
results are averaged over 1000 independent runs .
we see that the sbl - gamp algorithm achieves a similar recovery accuracy as the sbl - em algorithm even with a much lower computational complexity .
we developed a computationally efficient sparse bayesian learning ( sbl ) algorithm via the gamp technique .
the proposed method has a much lower computational complexity ( of order @xmath78 ) than the conventional sbl method .
simulation results show that the proposed method achieves recovery performance similar to the conventional sbl method in the low oversampling ratio regime .
d. wipf and s. nagarajan , `` iterative reweighted @xmath6 and @xmath7 methods for finding sparse solutions , '' _ ieee journals of selected topics in signal processing _ , vol . 4 , no . 2 , pp . 317329 , apr .
|
the sparse beyesian learning ( also referred to as bayesian compressed sensing ) algorithm is one of the most popular approaches for sparse signal recovery , and has demonstrated superior performance in a series of experiments .
nevertheless , the sparse bayesian learning algorithm has computational complexity that grows exponentially with the dimension of the signal , which hinders its application to many practical problems even with moderately large data sets . to address this issue ,
in this paper , we propose a computationally efficient sparse bayesian learning method via the generalized approximate message passing ( gamp ) technique .
specifically , the algorithm is developed within an expectation - maximization ( em ) framework , using gamp to efficiently compute an approximation of the posterior distribution of hidden variables .
the hyperparameters associated with the hierarchical gaussian prior are learned by iteratively maximizing the q - function which is calculated based on the posterior approximation obtained from the gamp .
numerical results are provided to illustrate the computational efficacy and the effectiveness of the proposed algorithm .
sparse bayesian learning , generalized approximate message passing , expectation - maximization .
|
in two companion papers ( mclure et al . 1999 ; dunlop et al . 2000 ) we present initial and final results from a deep hubble space telescope ( hst ) imaging survey of radio - quiet quasars ( rqqs ) , radio - loud quasars ( rlqs ) and radio galaxies ( rgs ) .
the results presented in these papers were derived from the hst images using a two - dimensional modelling technique developed to cope with such complications as central image saturation , the undersampled nature of wide field ( wf ) camera images , accurate image centering , and the precise form of the hst point spread function ( psf ) .
the primary purpose of this paper is to provide a description of this image analysis method , and to present the results of extensive testing on simulated active - nucleus+host - galaxy images constructed to span the full range of parameter space , and to mimic as closely as possible the real hst data .
these tests on simulated data were central to the development of our modelling technique , and also provide a means of estimating the typical errors in the derived host galaxy parameters as a function of redshift .
we also present new data in the form of infrared tip - tilt images of 4 of the most luminous quasars in our hst - imaging sample , and give the results of applying our two - dimensional modelling technique to these new @xmath3-band data .
the results provide further confidence in our analysis technique , and allow us to determine accurate @xmath1 colours for the hosts of these 4 quasars .
the layout of the paper is as follows . in section 2
we briefly review the main modelling algorithm ( for details see taylor et al .
1996 ) , and explain how the various problems which are specific to deep hst imaging were tackled . in section 3 we explain how synthetic hst quasar+host images were constructed to allow tests of the ability of the modelling technique to reclaim reliable host - galaxy parameters as a function of redshift and nuclear : host ratio .
we then present the results of these tests before proceeding in sections 4 and 5 to describe extensions to the modelling algorithm which were introduced to avoid having to make the assumption that the host galaxies were either perfect de - vaucouleurs spheroids , or freeman discs .
the new ukirt tip - tilt ircam3 images of a subset of the quasar sample are presented in section 6 along with the results of modelling these data in a manner basically identical to that used to model the hst data .
we then briefly describe the outcome of comparing the ukirt- and hst - derived galaxy parameters for these objects , before summarizing our main conclusions in section 7 .
given the difficulties associated with one - dimensional analysis techniques it was decided that a fully two - dimensional approach was needed to fully exploit the depth and resolution of the new hst and ukirt host - galaxy data .
the two - dimensional modelling code used in the analysis described in this paper , mclure et al .
( 1999 ) and dunlop et al .
( 2000 ) , is a development of that which was originally designed for the analysis of the ircam 1 @xmath3band imaging of this sample ( dunlop et al .
1993 , taylor et al .
consequently , only a brief outline of the general algorithm is supplied here , with those interested in a detailed description being referred to taylor et al .
( 1996 ) .
the model host galaxies are initially constructed on a two - dimensional array with the surface - brightness of each pixel described by either a de vaucouleurs @xmath4 law ( de vaucouleurs & capaccioli 1979 ) : @xmath5\ ] ] or a freeman disc law ( freeman 1970 ) : @xmath6 the model galaxies are constructed at much higher resolution than the data plate - scale in order to accurately simulate the rapidly varying surface brightness at small radii ( particularly important with the @xmath4 profile ) . following re - sampling of the model host galaxies , an arbitrarily large amount of flux
is added to the central pixel to represent the unresolved nuclear contribution .
in order to produce a simulated observation this ` zero - seeing ' model is then convolved with a high signal - to - noise observation of the instrumental point spread function ( psf ) .
the goodness of fit between this convolved model and the actual data is then tested on a pixel - by - pixel basis via the @xmath7 statistic , with the parameters determining the form of the model being iterated to find the minimum @xmath8 solution .
the model parameters which are left free during this minimization process are : * the luminosity of the nucleus * the central brightness of the host galaxy * the scalelength of the host galaxy * the position angle of the host galaxy * the axial ratio of the host galaxy many authors who have investigated host galaxies using the hst ( hutchings et al .
1994 , boyce et al .
1998 , hooper et al .
1997 ) have made use of the synthetic psfs produced by the tinytim software package ( krist 1998 ) .
this software is capable of reproducing the wfpc2 psf through any of the on - board filters , seemingly removing the need to sacrifice valuable orbits observing the empirical psf .
however , the inability of the tinytim software to reproduce the scattered light halo associated with the wfpc2 psf ( krist 1998 ) make these model psfs unsuitable for the convolution of model quasars .
this fact is illustrated in fig [ psfcom ] which shows a comparison between the high dynamic range empirical psf used in our analysis and the equivalent tinytim model .
it is clear from this that if the model quasars do not include the scattered - light halo , the scalelengths of the underlying host galaxies could be seriously overestimated .
a further issue to be considered when analysing wfpc2 images is the severe undersampling of the wf ccds . given that the central @xmath9 of the @xmath10band quasar images presented in mclure et al .
( 1999 ) and dunlop et al .
( 2000 ) are dominated by the unresolved nuclear component , it was imperative that the undersampling issue was dealt with properly . in order to deal with this problem
a solution was arrived at which , makes use of the sub - sampling capabilities of the tinytim software . although inadequate as a substitute for the empirical psf at large radii , the synthetic models produced by tinytim are an excellent match to the empirical psf in the very central regions where the optical effects are well understood .
the ability of tinytim to produce psf models at fifty times higher resolution than the wf2 plate - scale was used to predict the appearance of the central nine pixels of the wf2 psf with any possible sub - pixel centring .
these predictions are then @xmath8 matched to the core ( central nine pixels ) of the quasar image to give centroiding with theoretical accuracy of @xmath11 pixels . assuming that the tinytim model provides an accurate representation of the true psf in the central @xmath12 , it is then possible to correct the sampling of the central regions of the empirical psf to the same sub - pixel centring as that of the quasar image .
regardless of the care with which model quasars are constructed , two - dimensional modelling will fail to produce accurate results if incorrect allocation of error weighting in the @xmath8 test causes one area of the image to dominate the fit at the expense of others .
the minimum possible error that can be associated with any pixel is a combination of the poisson error due to photon shot noise , plus the read - noise and dark current contributions .
errors introduced during flat - fielding have proven to be negligible in both the hst data and the ircam 3 images considered in here .
figure [ errprof ] shows a comparison between poisson noise predicted by the wf2 noise model ( biretta et al .
1996 ) , and the actual statistical sampling errors as measured from a typical image . as can be seen from fig [ errprof ] ,
the errors calculated using the noise model are in near - perfect agreement with what is actually seen from the data for all pixels outside a radius of @xmath13 from the quasar core .
consequently , all pixels outside a radius of @xmath9 are simply allocated their expected poisson error during the model fitting process .
however , fig [ errprof ] also demonstrates that this system would grossly underestimate the actual error introduced inside @xmath14 by the severe undersampling of the wfpc2 psf .
therefore , inside a radius of @xmath9 a different system is used to allocate the error weighting .
a series of ten pixel - wide circular annuli centred on the quasar are constructed , and the variance ( @xmath15 ) of the distribution of the pixels falling within each annuli calculated .
all of the pixels falling within a particular annulus are then allocated the annulus variance as their error weighting .
this procedure is still justified even when the underlying host galaxy has a clear ellipticity since , even at hst resolution , the central @xmath9 of the quasar images are dominated by the circular symmetry of the psf . due to the success of the re - sampling technique
discussed above , it was possible to assign a poisson error weighting to the central pixel without biasing the fitting procedure .
this complicated process has proven to be successful , with the vast majority of the 33 hst objects modelled having minimum @xmath8 solutions lying in the @xmath16 region expected for correct weighting ( dunlop et al .
confidence in the weighting system is further strengthened by examination of the pixel maps of reduced @xmath8 values produced for the best - fitting models . shown in fig [ chimap ] a typical @xmath8 map from the best - fit to the radio - loud quasar 1217 + 023 .
it is clear from this figure that the desired uniform spread of @xmath8 values has been achieved , with no one area of the image dominating the fit .
the main advantage of the two - dimensional modelling technique being adopted here is that it allows each individual pixel in the image frame to be included in the model fitting as a degree of freedom .
however , due to the high angular resolution of our data , a typical model fit to a radius of @xmath17 will include some 45000 wfpc2 pixels ( @xmath18 ircam3 pixels ) in each @xmath8 evaluation .
the adopted solution was to utilise the downhill simplex method ( press et al .
1989 ) to locate the global minimum .
although significantly slower than the conjugate gradient method used in taylor et al .
( 1996 ) , downhill simplex minimization was chosen for its robustness in finding the true global minimum , and its insensitivity to the selection of initial conditions . to determine the prevalence of false minima in the hyper - surface , a high - resolution grid search was undertaken for two objects , with the extent of the search grid being @xmath19 from the parameter values found by the minimization routine .
the results of the grid search for the radio galaxy 3c234.0 , which contains a @xmath20 nuclear component , are shown in fig [ conts ] with the position of the global minimum marked with @xmath21 .
two features of this diagram are worthy of comment .
firstly it can be seen that in both parameter slices the minimum @xmath8 value lies at the bottom of a narrow ` valley ' , suggesting that false minima are not ubiquitous . secondly , despite the expected correlation between galaxy scalelength and central surface - brightness ( abraham et al .
1992 ) , the minimization routine has still successfully located the global minimum .
for the purposes of the full hst host galaxy study it was essential to determine the typical accuracy with which host - galaxy parameters could be recovered . due to the nature of host galaxy work , a data - set for which the host galaxy parameters have been accurately determine previously does not exist , and consequently , an extensive programme of tests was undertaken using synthetic data . to allow a realistic representation of the range of redshifts displayed by the wfpc2 data
the synthetic agn were constructed at three redshifts , z=0.1 , z=0.2 & z=0.3 . at each redshift 28 elliptical and
28 disc host galaxies models were constructed .
each group of 28 host galaxies consisted of a range of four possible half - light radii ( 5,10,15 and 20 kpc for @xmath22 ) , all of which had the same integrated luminosity .
to each of these four different galaxy scalelengths a range of seven possible nuclear components were added , giving @xmath23 ratios in the range @xmath24 .
the synthetic agn with no nuclear component were included to check that the modelling code did not have a bias towards preferring a nuclear component where none existed .
the final ensemble of models totalled 168 and covered the full range of parameters likely to occur in reality .
when constructing the suite of synthetic quasars several steps were taken to ensure that these simulations reflected the characteristics of actual wfpc2 observations as closely as possible .
the first step was to construct the final synthetic quasar images from a stack of separate frames in an identical fashion to the real data .
as described in mclure et al .
( 1999 ) , the actual quasar observations consist of three deep 600-second exposures which were complimented by three shorter snap - shot exposures of 5 , 26 and 40 seconds , designed to ensure a unsaturated measure of the nuclear component .
the final reduced quasar images consist of a stack of the three long exposures , with the central regions replaced by a scaled snap - shot exposure to recover the quasar s full dynamic range .
the construction of the synthetic quasars proceeded in exactly the same fashion , with individual simulations of each of the six separate exposures .
an appropriate level of background counts were then added to each of the simulated exposures before they were processed by the iraf routine mknoise which simulated the effects of both shot and read - out noise .
the fudicial luminosity of the simulated host galaxies was chosen to match the absolute magnitude of the best - fit to pks 2247 + 14 ( m@xmath25 ) , which was typical of the 12 hosts which had been observed at the time .
the calculation of the absolute luminosity of the hosts at the three redshifts was then determined assuming a typical spectral index of @xmath26 for the f675w filter ( @xmath27 ) , and a cosmology with @xmath28kms@xmath29mpc@xmath29 and @xmath30 .
the cosmological dimming of the agn point source assumed a spectral index of @xmath31 ( neugebauer et al .
figure [ synthgals ] shows the apparent magnitudes for all 33 host galaxies from the hst programme ( see dunlop et al .
2000 ) plotted against redshift .
also shown are the apparent magnitudes of the synthetic host galaxies .
there is a suggestion from fig [ synthgals ] that the synthetic galaxies are fractionally too bright compared to the data , making it arguably slightly too easy for the code to recover the galaxy parameters from the simulated images .
this is confirmed by the results for the full sample which show the mean absolute luminosity of the 33 host galaxies to be m@xmath32 , 0.27 magnitudes fainter than the synthetic hosts used in the testing programme . to illustrate this point the dashed line in fig [ synthgals ]
shows the apparent magnitude of the synthetic host galaxies dimmed by a further 0.27 magnitudes , which clearly provides a better match to the data .
however , the overestimate of the typical host luminosity should be more that off - set by the much larger range of @xmath23 tackled by the modelling code during testing . while the average @xmath23 of the quasars from the hst programme is only @xmath33 , the code has been tested with values of @xmath23 in the range @xmath24 .
one final measure was taken to improve the realism of the synthetic quasar images .
the empirical psf which was used during the production of the synthetic data had an artificial centroiding shift of @xmath34 applied to it , significantly greater than the estimated centroiding error .
this precaution was taken in light of the fact that using the same psf to convolve both the synthetic images and the models used in the fitting process is obviously an idealised situation .
the results from the modelling of the synthetic quasars are listed in tables [ ellips ] , [ discs ] and [ moreres ] .
two features of these results are worthy of individual comment .
firstly , it can be seen that for both host morphologies the errors associated with the determination of all the parameters steadily increase with redshift , as is expected due to inevitable drop in signal - to - noise .
secondly , with regards to the host scalelength it can be seen to be significantly easier to accurately determine this parameter for disc hosts than for ellipticals .
this is again as is expected considering the different behaviour of the freeman and de vaucouleurs surface - brightness laws in the central @xmath14 .
the cusp - like nature of the @xmath4 law at small radii inevitably leads to greater difficulty in de - coupling the relative contributions of the host and nuclear components .
the results of the synthetic data testing can be summarized as follows : * @xmath35 success in host morphology discrimination .
* host flux determination @xmath36 accurate in all cases . *
error in @xmath37 out to @xmath38 for ellipticals . * error in @xmath39 out to @xmath38 for discs . where successful morphological determination refers to a @xmath40 between the best - fitting model of the correct morphology , and the best - fitting alternative model , a difference equivalent to the @xmath41 confidence level for a 5-parameter fit ( press et al .
although the high degree of accuracy achieved in these simulations is impressive , it should be noted that these error estimates are only valid for the high resolution data provided by hst . as will be seen in section [ tip - tilt ] , the errors associated with ground - based data can be significantly larger .
.results of the two - dimensional modelling tests using synthetic quasars with elliptical host galaxies .
column 1 gives the redshift of the quasar .
column 2 gives the actual scalelength of the simulated host in kpc .
columns 3 - 8 give the mean percentage error in the reclaimed value of the relevant parameter .
each value is the mean for the seven synthetic quasars produced at a particular redshift , with a particular scalelength ( for each redshift and scalelength combination there where seven different synthetic quasars produced with @xmath23=0 , 0.5 , 1.0 , 2.0 , 4.0 , 8.0 & 16 ) . [
cols="^,^,<,^,^,^,^,^ " , ] the integrated absolute @xmath3band and @xmath10band magnitudes of the best - fitting host galaxy models are presented in table [ kmags ] .
the absolute magnitudes have been calculated using @xmath42-corrections assuming a spectral index @xmath43 and @xmath26 for the @xmath3- and @xmath10band respectively ( @xmath27 ) .
the calculation of the host luminosities in terms of @xmath44 have been performed taking @xmath45 ( gardner et al .
1997 ) and @xmath46 ( calculated by converting the value of @xmath47 ( lin et al . 1996 ) to an integrated magnitude ) .
again it can be seen that these results confirm the findings of the hst @xmath10band imaging , and the previous @xmath3band imaging ( taylor et al .
1996 ) , that the quasar hosts are all luminous galaxies with @xmath48 .
the fact that in this small group of objects the rlqs are substantially more luminous than the rqqs should not be taken as significant since , 1004 + 130 and 1217 + 023 are found to be brighter than average for rlqs from the @xmath10band modelling , while 0953 + 415 has the largest @xmath23 ratio in the entire 33-object sample .
the excellent agreement of the model results from the @xmath3- and @xmath10band imaging in terms of host morphology and scalelength allows an accurate measurement of the host galaxy optical - infrared colour to be made . the integrated , rest - frame ,
@xmath1 colours listed in table [ kmags ] can be seen to be perfectly consistent with that expected from an old passively evolving stellar population formed at @xmath49 .
this conclusion is confirmed by the @xmath1 colours of the full 33-object sample , where comparison with spectral synthesis models indicates that none of the agn host galaxies have stellar populations younger that @xmath50 gyr at z=0.2 ( dunlop et al .
a further conclusion which can be drawn from the red colours of these host galaxies is that any star formation associated with the agn activity must be confined to the central few kpc , and not widely distributed throughout the body of the host galaxy .
the results of applying our two - dimensional modelling technique to the full range of simulated quasar images provides confidence that , at least at @xmath51 all derived host galaxy parameters can be trusted to a typical accuracy of better than 10% .
this is perhaps not surprising for parameters such as host luminosity , position angle , and axial ratio , but accurate scalelength determination has proved particularly difficult in virtually all previous studies of quasar hosts .
the reliability of our derived scalelengths is further re - inforced by the virtually identical values yielded by the analysis of the infrared tip - tilt images discussed in the previous section .
indeed the fact that such similar scalelengths have resulted from analysis of images taken at different wavelengths , with different plate - scales , and using very different telescopes with very different psf complications is undeniably impressive .
it is this accuracy of scalelength determination that has enabled us , for the first time , to demonstrate that the hosts of rqqs , rlqs and rgs follow a kormendy relation with slope and normalization identical to that displayed by ` normal ' low - redshift massive ellipticals ( mclure et al . 1999 ; dunlop et al .
2000 ) .
the analysis of the new , higher - resolution ( cf dunlop e al . 1993 ) infrared images of the 4 quasars imaged with ircam3 indicates that galaxy scalelength is ( perhaps unsurprisingly ) essentially unchanged between the @xmath10 and @xmath3 wavebands , and that the hst - derived scalelengths should be trusted in preference to those derived by taylor et al .
( 1996 ) from the ircam1 images of dunlop et al .
( 1993 ) . in at least some cases the original infrared - based host - galaxy scalelengths appear to have been
over - estimated from the ircam1 images , although not sufficiently to alter the basic conclusions of taylor et al .
this has been taken into account in a re - analysis of the ircam1 data undertaken by dunlop et al .
( 2000 ) in order to derive reliable @xmath1 colours for _ all _ the host galaxies in the hst sample .
we thank the referee for comments which improved a number of aspects of this paper .
the united kingdom infrared telescope is operated by the joint astronomy centre on behalf of the u.k . particle physics and astronomy research council . based on observations with the nasa / esa hubble space telescope , obtained at the space telescope science institute , which is operated by the association of universities for research in astronomy , inc . under nasa contract no .
nas5 - 26555 .
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 .
mjk acknowledges the award of a pparc pdra , and also acknowledges support for this work provided by nasa through grant numbers o0548 and o0573 from the space telescope science institute , which is operated by aura , inc .
, under nasa contract nas5 - 26555 .
rjm acknowledges a pparc studentship .
abraham r.g . , crawford c.s . ,
mchardy i.m . , 1992 ,
apj , 401 , 474 + bahcall j.n . , kirhakos s. , schneider d.p . , 1994 ,
apj , 435 , l11 + bahcall j.n . , kirhakos s. , schneider d.p .
, 1995a , apj , 447 , l1 + biretta j.a . , et al . , 1996 ,
wfpc2 instrument handbook version 4.0 + boyce p.j . ,
et al . , 1998 , mnras , 298 , 121 + de vaucouleurs g. , capaccioli m. , 1978 , apjs , 40 , 699 + donofrio m. , capaccioli m , caon n. , 1994 , mnras , 271 , 523 + dunlop j.s . , taylor g.l . , hughes d.h . , robson e.i . , 1993 , mnras , 264 , 455 + dunlop j.s . , mclure r.j .
, kukula m.j . ,
baum s.a . ,
odea c.p . , 2000 , mnras , submitted + freeman k.c , 1970 , apj , 160 , 811 + gardner j.p .
, sharples r.m . ,
frenk c.s .
, carrasco b.e . , 1997 ,
a&as , 190 , 4303 + graham a. , lauer t.r .
, colless m. , postman m. , 1996 , apj , 465 , 534 + hooper e.j .
, impey c.d .
, foltz c.b . , 1997 ,
apj , 480 , l95 + hutchings j.b . ,
morris s.c . ,
gower a.c .
, lester m.l . , 1994 , apj , 429 , l1 + krist j. , 1998 , tinytim user manual + lauer t.r .
, et al . , 1995 , aj , 110 , 2622 + lin h. , krishner r.p . , shectman s.a . , landy s.d . , oemler a. , tucker d.l . , schechter p.l . , 1996 , apj , 464 , 60 + mclure r.j . , kukula m.j . , dunlop j.s . , baum s.a . , odea c.p . , hughes d.h . , 1999 ,
mnras , 308 , 377 + neugebauer g. , green g.f . , matthews k. , mcgill j. , scoville n. , soifer b.t . , 1987 , apjs , 93 , 1057 + percival w. , miller l. , mclure r.j . , dunlop j.s
, 2000 , mnras , astro - ph/0002199 + press w.h .
, 1989 , `` numerical recipes '' , cambridge university press + schombert j.m . , 1987 , apjs , 64 , 643 + sersic j.l . , 1968 , atlas de galaxies australes .
observatorio astronomico cordoba + taylor g.t . , dunlop j.s .
, hughes d.h . ,
robson e.i . , 1996 , mnras , 283 , 930 +
|
a description is given of the method used to extract quasar host - galaxy parameters from the deep hubble space telescope ( hst ) quasar images presented by mclure et al .
( 1999 ) and dunlop et al .
( 2000 ) .
we then give the results of extensive testing of this technique on a wide range of simulated quasar+host combinations spanning the redshift range of our hst study ( @xmath0 ) .
these simulations demonstrate that , when applied to our deep hst images , our method of analysis can easily distinguish the morphological type of a given host galaxy , as well as determining its scalelength , luminosity , axial ratio and position angle to within an accuracy of a few percent .
we also present new infrared tip - tilt images of 4 of the most luminous quasars in our hst sample , along with the outcome of modelling these data in a similar manner .
the results provide further confidence in the accuracy of the derived host - galaxy scalelengths , and allow accurate determination of @xmath1 colours for this subset of sources .
all 4 of these quasar host galaxies have very similar red colours , @xmath2 , indicative of a well - evolved stellar population .
galaxies : active galaxies : photometry infrared : galaxies quasars : general
|
chemical vapor deposition ( cvd ) of graphene on metal surfaces@xcite is currently viewed as one of the most promising scalable methods for economically producing large and abundant high - quality monolayer graphene sheets .
it is thus greatly important to fully understand and control the behavior of electrons on this form of graphene .
cvd graphene , as any other solid grown by chemical vapor deposition , is generally a polycrystal composed by several grains with distinct crystallographic orientations .
these grains are separated by grain boundaries ( gbs),@xcite which due to the @xmath0 bonding structure of carbon atoms in graphene , are typically made of pentagonal , heptagonal and octagonal rings of carbon atoms.@xcite grain boundaries generally intercept each other at random angles , being neither periodic nor perfect straight lines . the properties of cvd graphene flakes are strongly influenced by the quantity , distribution and microscopic character of its grain boundaries.@xcite each type of grain boundary exhibits distinctive chemical,@xcite mechanical@xcite and electronic@xcite properties .
this is particularly evident in what concerns the electronic transport in cvd graphene .
for instance , there is abundant experimental evidence that the details of the cvd - growth recipes used to synthesize cvd graphene flakes greatly constrain its transport properties.@xcite furthermore , application of strain and chemical decoration are also expected to strongly influence cvd - graphene transport properties.@xcite in particular , as shown by several recent experiments probing the transport properties of single grain boundaries,@xcite the electron - scattering off a grain boundary is determined by the its microscopic details and the relative orientation of the grains it separates.@xcite observation and probing of graphene grain boundaries has been constantly refined in recent years , as shown by a quick survey of the recent literature in the field.@xcite more interestingly , several promising new methods of controlling and manipulating the position , orientation and microscopic configuration of grain boundaries have been recently unveiled.@xcite some of these methods allow for the creation of periodic and straight grain boundaries,@xcite whose transport properties have been extensively investigated theoretically.@xcite this widens the prospects for the engineering of graphene - based electronic devices that take advantage of the scattering properties of these grain boundaries , to manipulate graphene electrons various degrees of freedom , such as its valley quantum number .
following these recent advances , in this manuscript , we will focus our attention on the electronic properties of a particular class of periodic and linear grain boundaries that is often disregarded in the literature .
namely , we will investigate grain boundaries with periodicities such that both dirac points ( on each side of the gb ) are mapped into the @xmath1 point of the projected brillouin zone see ref .
for a brief discussion of their properties . due to this mapping of the dirac points ,
such grain boundaries allow for intervalley scattering of low - energy charge carriers . in what follows
we will show how can we work out the low - energy electronic scattering off such gbs , and will see how the intervalley scattering depends on the system s microscopic details . to keep things simple ,
we have chosen to investigate zigzag aligned linear grain boundaries separating two grains with the same orientation ( also referred to in the literature as degenerate or zero misorientation - angle grain boundaries ) .
several such gbs were proposed in the context of ab - initio works both on graphene and on boron nitride : the t7t5 grain boundary,@xcite the 7557 grain boundary@xcite and the 8484 grain boundary.@xcite
their transport properties have been recently investigated under the perspective of the tight - binding model.@xcite for simplicity we will consider a pentagon - only like grain boundary@xcite with a periodicity three times greater than that of pristine graphene ( along the zigzag direction
[ fig : scheme-3fpentonly ] ) .
such a periodicity ensures that the dirac points are mapped into the @xmath1 point
[ fig : fbz3-projected ] . in the remaining of this text
we will call this gb by _
3-periodic pentagon - only _ grain boundary . along the @xmath2-direction .
the lattice vectors used are @xmath3 and @xmath4 .
the hopping parameters at the defect alternate in value reading ( in units of @xmath5 ) : @xmath6 , @xmath7 and @xmath8 . ]
as is well known , graphene s low energy charge carriers behave as massless dirac fermions .
these are governed by a hamiltonian composed of two copies of the 2d dirac hamiltonian , each one of them valid around one of the two dirac points.@xcite in this limit the grain boundary essentially acts as a one - dimensional line that imposes a boundary condition on the dirac spinors living on the semi - planes above and bellow the grain boundary .
such boundary condition will result in a discontinuity in the spinors across the defect and will control its scattering properties.@xcite in alternative , the grain boundary can also be thought of as a finite width strip containing a generalized potential that constrains the dynamics of the massless dirac fermions.@xcite the specific form of the boundary condition seen by the massless dirac fermions at the 3-periodic pentagon - only grain boundary is determined by the details of its microscopic tight - binding model . in calculating it , we will follow the methodology developed for the cases of the pentagon - only , @xmath9 and @xmath10 grain boundaries.@xcite below , we show that the boundary condition obtained from the tight - binding gives rise , in the low - energy limit , to a boundary condition explicitly introducing intervalley scattering .
furthermore , we will show how , starting from the grain boundary s microscopic details , can we determine the generalized potential associated with viewing the grain boundary as a finite width strip with a potential constraining the dirac fermions dynamics . before proceeding , we detail the structure of this text . in section
[ sec : lowen-3periodic ] we discuss general properties of dirac fermion scattering off 3-periodic grain boundaries . in section [ sec:3po ]
we solve the electronic scattering off a 3-periodic pentagon - only grain boundary : we start by computing the tight - binding boundary condition matrix relating electronic amplitudes on each side of the grain boundary ( see sub - section [ sec : tbmodel-3po ] ) ; we then derive the boundary condition matrix _ seen _ by the low - energy charge carriers ( i.e. by the massless dirac fermions ) at the grain boundary ( see sub - section [ sec : ca - atgb ] ) ; finally , in sub - section [ sec : transmittance ] we compute the transmission probabilities for different choices of the microscopic hopping parameters at the grain boundary .
we close with section [ sec : conclusion ] where we overview the main results of the manuscript .
let us start by considering the case of a general 3-periodic grain boundary , i.e. a grain boundary with a periodicity ( and orientation ) defined by the vector @xmath11 , where @xmath12 and @xmath13 is a multiple of @xmath14 see fig .
[ fig : scheme-3fpentonly ] where @xmath15 and @xmath16 .
the presence of such a grain boundary in graphene , breaks the translation symmetry along the direction perpendicular to the grain boundary .
furthermore , the periodicity of these grain boundaries happens to fold the first brillouin zone in such a way that the two dirac points are mapped into the @xmath1-point of the projected brillouin zone .
it is thus natural to expect that intervalley scattering off these nanostructures is generally allowed at low energies .
direction ( perpendicular to the defect ) . in green ,
the spectrum of pristine graphene ( with the lattice vector along the zigzag direction given by @xmath17 ) .
its dirac points are located at @xmath18 . in red ,
the folded spectrum arising from choosing the lattice vector along the zigzag direction to be @xmath19 . in this case
, we can clearly see that the two dirac points are mapped into @xmath20 . ] as a consequence , when addressing the problem of electronic scattering across such kind of grain boundaries , we need to consider both valleys . instead of working with two separate copies of the dirac hamiltonian @xmath21 valid in the vicinity of each dirac point @xmath22
, we have to work with both copies simultaneously , i.e. with the hamiltonian @xmath23 where @xmath24 and @xmath25 ( @xmath26 ) stand for the @xmath27 pauli matrices acting , respectively , on the valley and pseudo - spin degrees of freedom , while @xmath28 and @xmath29 stand for the @xmath27 identity matrix . in the above equation ( and in the remaining of this text ) we have set @xmath30 . as we are considering a hamiltonian that is independent of ( real ) spin , we will always omit the spin degree of freedom .
the presence of a periodic grain boundary in a graphene flake imposes a discontinuity between the dirac fermion s spinors on each of the grain boundary s sides.@xcite however , grain boundaries that are @xmath14-periodic also connect the two valleys through non - zero intervalley scattering matrix elements . for simplicity ,
let us consider the case of a zigzag - aligned ( i.e. @xmath2-aligned ) @xmath14-periodic grain boundary located at @xmath31 .
such a grain boundary imposes the following general boundary condition on the dirac spinors : @xmath32 where @xmath33^{t}$ ] are @xmath34-spinors since @xmath35^{t}$ ] stands for the 2-spinor describing dirac fermions living in the @xmath36 valley .
the @xmath37 matrix @xmath38 can be written in the general form @xmath39 \ , , \label{eq : genmca}\end{aligned}\ ] ] where @xmath40 ( @xmath41 ) are @xmath27 matrices controlling the valley preserving ( intervalley ) scattering across the grain boundary .
the matrix @xmath38 must satisfy the flux conservation condition @xmath42 , where @xmath43 stands for the conserved current along the direction perpendicular to the grain boundary .
this stems from the hermiticity of the tight - binding hamiltonian ( that enforces current conservation at the gb in the tight - binding model ) .
furthermore , whenever we deal with non - magnetic grain boundaries , the boundary condition must as well be time - reversal invariant @xmath44 .
recall that the time - reversal operation exchanges the two dirac cones and applies complex conjugation , @xmath45 . a dirac fermion from the @xmath46 valley ( @xmath47 ) , incoming from @xmath48 ( see fig .
[ fig : scheme-3fpentonly ] ) will be partially transmitted and partially reflected at the grain boundary . in the absence of a potential difference between the two sides of the grain boundary
, the wave function reads [ eq : wavefunctions_dirac-1 ] @xmath49 on the lower half - plane ( i.e. @xmath50 ) , while for the upper half - plane ( i.e. @xmath51 ) we have @xmath52 in eqs .
( [ eq : wavefunctions_dirac-1 ] ) the @xmath53 are 4-spinors which read [ eq : wavefunctions_dirac-2 ] @xmath54^{t } e^{i(q_{x}x+q_{y}y ) } \ , , \\ \psi_{\mathbf{q } s}^{-,>}(\mathbf{r } ) & = & \frac{1}{\sqrt{2 } } \left[\begin{array}{cccc } 0 , & 0 , & s \ ,
e^{-i\theta_{\mathbf{q}}^{- } } , & 1 \end{array}\right]^{t } e^{i(q_{x}x+q_{y}y ) } \ , , \\
\psi_{\mathbf{q } s}^{+,<}(\mathbf{r } ) & = & \frac{1}{\sqrt{2 } } \left[\begin{array}{cccc } s \ , e^{-i \bar{\theta}_{\mathbf{q}}^{+ } } , & 1 , & 0 , & 0 \end{array}\right]^{t } e^{i(q_{x}x - q_{y}y ) } \ , , \\
\psi_{\mathbf{q } s}^{-,<}(\mathbf{r } ) & = & \frac{1}{\sqrt{2 } } \left[\begin{array}{cccc } 0 , & 0 , & s \ , e^{-i \bar{\theta}_{\mathbf{q}}^{- } } , & 1 \end{array}\right]^{t } e^{i(q_{x}x - q_{y}y ) } \ , .
\end{aligned}\ ] ] in the above expressions @xmath55 , @xmath56 , while @xmath57 and @xmath58 stand for the complex phases of , respectively , @xmath59 and @xmath60 . furthermore , @xmath61 stands for the sign of the energy , distinguishing electrons and and holes , while @xmath62 ( @xmath63 ) and @xmath64 ( @xmath65 ) respectively stand for the valley preserving ( intervalley ) reflection and transmission coefficients . since the conserved current associated with a given propagating mode is the same for all modes , then the four transmission and reflection probabilities are simply given by @xmath66 and @xmath67 , for @xmath47 and @xmath68 .
the transmission and reflection coefficients are obtained by solving the system of linear equations originating from imposing the boundary condition eq .
( [ eq : generalbc - m ] ) on the wave function written in eqs .
( [ eq : wavefunctions_dirac-1 ] ) and ( [ eq : wavefunctions_dirac-2 ] ) .
therefore , the transmission and reflection probabilities will not only depend on the angle of incidence of the electron into the gb , but will also ( strongly ) depend on the microscopic properties of the grain boundary through the matrix elements of @xmath38 .
below , following refs .
, we will show how can we compute the boundary condition matrix @xmath38 from the tight - binding model of the grain boundary .
however , before proceeding we will briefly discuss an often useful alternative viewpoint for such scattering problems ( see ref . ) .
instead of considering that in the low - energy limit the grain boundary simply imposes a discontinuity on the massless dirac fermions spinors along a line parallel to the @xmath2-axis , we will consider that the grain boundary can be viewed as a finite strip of width @xmath69 that exists in @xmath70 and extends along the @xmath2 direction see right - hand - side of fig .
[ fig : two - perspectives ] . on each side of this strip
the dirac fermions will be governed by eq .
( [ eq : h2k ] ) , while inside the strip there will also be a general local potential of the form @xmath71 , which gives rise to a discontinuity of the dirac spinors at the line separating the two half - planes [ where the dirac hamiltonian governs the physics : @xmath72 , with @xmath73 . on the right - hand - side , these two half - planes are separated by a finite width strip where the dirac fermions are subject to a generalized potential @xmath74 . ]
the equivalence between these two viewpoints becomes obvious when we integrate out the dirac equation in the finite width strip ( i.e. between @xmath75 ) while satisfying the constrain @xmath76 as @xmath77 .
as we show in detail in appendix [ app : genpot ] , this integration gives rise to the following boundary condition matrix @xmath78 which allows us to connect the matrix elements of @xmath38 ( determined from a tight - binding microscopic model ) and the effective generalized local potential felt by the dirac fermions inside the finite width strip . within this perspective
, there are two interfaces at which we must ensure the continuity of the wave function , namely @xmath79 and @xmath80 where @xmath81 .
these two equalities correspond to eight conditions ( @xmath82 are 4-spinors ) that will determine the eight unknown scattering coefficients ( region inside the strip requires four additional coefficients ) .
in what follows we will make the statements of the previous section concrete by investigating the electronic transport across the 3-periodic pentagon - only grain boundary ( see fig . [
fig : scheme-3fpentonly ] ) .
we will start by briefly sketching how can we compute the tight - binding boundary condition matrix relating the wave function above and below the grain boundary@xcite see sub - section [ sec : tbmodel-3po ] . from that result
we will then compute the boundary condition matrix seen by the massless dirac fermions at the grain boundary see sub - section [ sec : ca - atgb ] .
finally , in sub - section [ sec : transmittance ] we will work out the scattering problem and analyze the valley preserving and intervalley transmittance for specific sets of microscopic parameters defining the 3-periodic pentagon - only grain boundary .
consider a first - neighbor tight - binding model for electrons in the @xmath83-orbitals of graphene , where we define the pristine honeycomb direct lattice vectors as ( see fig .
[ fig : scheme-3fpentonly ] ) : @xmath84 and @xmath4 .
as we want to study a zigzag - oriented grain boundary with periodicity @xmath19 , we choose a _ bulk _ unit cell defined by the lattice vectors @xmath19 and @xmath85 as sketched in fig .
[ fig : scheme-3fpentonly ]
. fourier transforming along the @xmath19 direction diagonalizes the system s hamiltonian with respect to the variable @xmath86 , introducing the quantum number @xmath87 .
the corresponding bulk tight - binding equations can then be written as [ eq : bulktbeqs ] @xmath88 where @xmath89^{t}$ ] for @xmath90 and the matrix @xmath91 is defined in eq .
( [ eq : wamatrix ] ) .
( [ eq : bulktbeqs ] ) can be condensed in the form of a transfer matrix equation@xcite relating amplitudes at the atoms of the unit cell located at @xmath92 with the amplitudes at the atoms of the unit cell located at @xmath93 .
such an equation reads @xmath94 with @xmath95^{t}$ ] , while the transfer matrix @xmath96 is given by @xmath97 in the above equation , matrix @xmath98 is simply used to change from the basis @xmath99 into the basis @xmath100 .
this matrix is written in eq .
( [ eq : rmatrix ] ) , while matrices @xmath101 and @xmath102 are written in eqs .
( [ eq : q1q2 ] ) . as was shown in ref .
, we can employ a basis transformation @xmath103 that makes the transfer matrix block diagonal , @xmath104 $ ] .
we will use the notation @xmath105 to identify the elements of a vector in this basis .
the three matrices @xmath106 , @xmath107 and @xmath108 on the diagonal of @xmath109 are @xmath27 matrices that depend on both @xmath110 and @xmath111 .
they are written in eq .
( [ eq : tmdiags ] ) , while the matrix @xmath112 is written in eq .
( [ eq : lambdamat ] ) .
each one of the @xmath106 , @xmath107 and @xmath108 matrices correspond to one of the three propagation modes of the problem . around @xmath113 two of these modes are low - energy ( corresponding to each of the two dirac cones ) , while the other one is a high - energy mode
see appendix [ app : bulktb ] . in a similar way
, we can compute the boundary condition matrix that relates electronic amplitudes below and above the 3-periodic pentagon - only grain boundary ( see fig .
[ fig : scheme-3fpentonly ] ) .
we start by writing the tight - binding equations in the grain boundary region@xcite ( we neglect out - of - plane relaxations in the gb region ) [ eq : dfctbeqs ] @xmath114 where the matrix @xmath115 is written in eq .
( [ eq : sigmaxp ] ) .
note that @xmath115 depends both on @xmath111 and on the hopping parameters at the grain boundary , @xmath6 , @xmath7 and @xmath8 .
we can express these equations ( see appendix [ app : defecttb ] ) as a boundary condition equation connecting the electronic wave function on the two sides of the grain boundary @xmath116 the boundary condition matrix @xmath117 above is a @xmath118 matrix given by @xmath119 where , for the sake of simplicity , we have omitted the dependence of the matrices @xmath117 and @xmath120 on @xmath121 , @xmath87 and @xmath122 . the matrices @xmath120 are written in eqs . ( [ eq : nmatricesdfcln ] ) . finally , we can write the boundary condition matrix @xmath117 in the basis uncoupling the three pairs of modes of the transfer matrix , @xmath123^{-1}$ ] . by inspecting the boundary condition
( [ eq : tbdc ] ) in this basis , @xmath124 , we readily conclude that , in general , @xmath125 mixes all the three modes of the transfer matrix ( both the high - energy and the two low - energy ones ) . at very low energies , @xmath126 , and
very near the dirac points , @xmath127 , the matrix @xmath125 acquires a somewhat simple form see eq .
( [ eq : matdiracpnt ] ) . in this limit ,
the high - energy modes are evanescent , one of them increasing and the other one decreasing exponentially with @xmath128 see eq .
( [ eq : tmdiags1-lowe - limit ] ) and subsequent paragraph .
since we must require the wave function to be normalizable , we conclude that when @xmath126 and @xmath129 the wave function must have the following form [ eq : lowen - states ] @xmath130 , \,\ , \textrm{for $ n>0 $ , } \end{aligned}\ ] ] and @xmath131 , \,\ , \textrm{for $ n<0 $ , } \end{aligned}\ ] ] where , in order to keep the notation lighter , we have omitted the dependence on @xmath87 of both the vectors @xmath132 and @xmath133 , and of the amplitudes @xmath134 and @xmath135
. we can now substitute eqs .
( [ eq : lowen - states ] ) in the boundary condition @xmath136 to eliminate the high energy modes from our problem , ending up with an effective boundary condition that only involves the low - energy modes .
such a manipulation generates an effective boundary condition matrix @xmath137 that we can express in terms of the matrix elements of @xmath125 as follows @xmath138 \ , .
\label{eq : meffgen}\end{aligned}\ ] ] the matrix @xmath137 for the case of the 3-periodic pentagon - only grain boundary is written in eqs .
( [ eq : meff ] ) and ( [ eq : fandg ] ) . as is widely known , in the low - energy continuum limit the tight - binding amplitudes , @xmath139 , can be expressed in terms of slowly varying fields , @xmath140 , as @xmath141 we can thus cast the tight - binding 4-spinor valid at low - energies , @xmath142^{t}$ ] , in terms of slowly varying dirac fields as @xmath143^{t}$ ] , where @xmath144^{t}$ ] .
we can finally write the boundary condition that dirac fermions _ see _ at the 3-periodic pentagon - only grain boundary as @xmath145 , where @xmath146^{t } = [ \psi_{a_{-}}(\mathbf{r}),\psi_{b_{-}}(\mathbf{r}),\psi_{a_{+}}(\mathbf{r}),\psi_{b_{+}}(\mathbf{r})]^{t}$ ] . in the case of the 3-periodic pentagon - only grain boundary , the matrix @xmath38 ( see appendix [ app : bc - ca ] ) reads @xmath147 \ , , \label{eq : bcmatrixp3}\end{aligned}\ ] ] where @xmath148 and @xmath149 are written in eqs .
( [ eq : fandg ] ) .
interestingly , in eq .
( [ eq : bcmatrixp3 ] ) we clearly see that the off - diagonal blocks of matrix @xmath38 , those which control the intervalley scattering , are not _ a priori _ zero .
we can thus conclude that in general this grain boundary gives rise to intervalley scattering .
in particular , the intervalley scattering mixes the @xmath150 component of one valley with the same component of the other valley .
notice that when all the three hoppings are equal ( i.e. @xmath151 ) , we get back to the simple case of the pentagon - only grain boundary ( with periodicity @xmath152 ) that , as we know,@xcite does not give rise to intervalley scattering . owing to the fact that @xmath153 and @xmath154 [ see eqs .
( [ eq : fandg ] ) ] , its boundary condition matrix reads @xmath155 \ , .
\label{eq : bcmatrix - simplepo}\end{aligned}\ ] ] but this is natural since in such a case we are effectively dealing with a grain boundary with periodicity @xmath17 which maps the projected dirac points into distinct values of @xmath87 see fig .
[ fig : fbz3-projected ] . moreover , there are a few cases where , despite the 3-periodicity of the grain boundary , intervalley scattering is suppressed . in these cases , the microscopic details of the grain boundary ,
i.e. the precise values of @xmath6 , @xmath7 and @xmath8 , force @xmath156 thus forbidding intervalley scattering .
examples of such cases are : @xmath157 and @xmath158 ; @xmath159 and @xmath160 ; and @xmath161 and @xmath162 . in the context of the perspective where we consider the grain boundary to be a finite width strip with a generalized potential
( [ eq : genpotv ] ) and end of section [ sec : lowen-3periodic ] ] , we show in appendix [ app : genpot ] that the generalized potential originating from the 3-periodic pentagon - only grain boundary both has valley preserving terms ( such as @xmath163 - a scalar potential - and @xmath164 - a constant gauge potential ) , and valley mixing terms ( such as @xmath165 , @xmath166 , @xmath167 and @xmath168 ) . as a consequence , in general
, a 3-periodic pentagon - only grain boundary will not only generate intervalley scattering , but it will also prevent the existence of an angle of perfect transmission ( see appendix [ app : genpot ] ) . as discussed in section [ sec : lowen-3periodic ] we can now compute the transmission and reflection coefficients @xmath169 , @xmath170 , @xmath171 and @xmath172 . in particular , the transmission probability for an incoming electron living on the @xmath173 ( @xmath174 ) valley to be transmitted into the same valley is given by @xmath175 ( @xmath176 ) , while the probability for it to be transmitted into the other valley is given by @xmath177 ( @xmath178 ) . in eqs .
( [ eq : transmission - coefs ] ) we write the expressions of the @xmath179 , @xmath180 , @xmath181 and @xmath182 , for the 3-periodic pentagon - only grain boundary with general hoppings ( in units of @xmath5 ) , @xmath6 , @xmath7 and @xmath8 .
these were obtained by solving the system of linear equations defined by @xmath145 where @xmath38 is given by eq .
( [ eq : bcmatrixp3 ] ) . in fig .
[ fig : transmittances-3fpentonly ] we have plotted the transmission probabilities @xmath183 and @xmath184 for the case where the hopping parameters are @xmath185 , @xmath186 and @xmath187 . .
the @xmath188 ( @xmath189 ) [ @xmath190 ( @xmath191 ) ] stands for the probability of transmission of an incident dirac fermion living on the @xmath174 [ @xmath173 ] valley to be transmitted into the @xmath174 ( @xmath173 ) valley .
the hopping parameters at the grain boundary were set to : @xmath193 , @xmath194 and @xmath195 . ] for a 3-periodic pentagon - only grain boundary with the above hopping parameters , the intervalley scattering is weak , with the intervalley transmission probabilities being considerably smaller than the valley preserving ones .
however , this picture can be greatly modified if we choose an appropriate set of hopping parameters at the grain boundary . as an example , in fig .
[ fig : transmittances-3fpentonly__2 ] we plot the transmission probabilities for a case where @xmath196 , @xmath197 and @xmath198 , which shows much stronger intervalley scattering than fig .
[ fig : transmittances-3fpentonly ] . , but now with the hopping parameters at the grain boundary set to : @xmath199 , @xmath200 and @xmath201 . ] this robust increase of the intervalley transmission ( compare figs .
[ fig : transmittances-3fpentonly ] and [ fig : transmittances-3fpentonly__2 ] ) can be traced back to a strong amplification of the terms @xmath165 and @xmath167 of the generalized potential existent inside the strip that mimics the effect of the grain boundary in the low - energy limit .
such an amplification ( by nearly two orders of magnitude ) gives rise to a set of generalized potential terms of similar magnitudes , thus increasing the amount of valley mixing of the eigenmodes living inside the strip .
the stronger the valley mixing of these modes , the more the electron s valley quantum number rotates while propagating inside the strip , and thus more wave function weight is transferred between valleys while the dirac fermion propagates inside the strip .
finally , we must note that , as we can quickly infer from the above results ( where we have considered different sets of values for the hopping renormalizations @xmath122 ) , the scattering properties of the grain boundary are strongly dependent on the microscopic details at the grain boundary , as previously remarked by other studies.@xcite such behaviour points towards the possibility of making use of this kind of nanostructures to control and explore the valley degree of freedom of graphene .
chemical decoration of the grain boundary region , application of strains , of electric and of magnetic fields , all likely modify the electronic scattering off these grain boundaries , thus suggesting their usage as sensors and current switchers .
to close , let us briefly summarize the contents of this manuscript .
we have started by analyzing in general terms the low - energy charge carrier transport across zigzag - aligned degenerate 3-periodic grain boundaries .
we have then demonstrated such results by working out the low - energy charge transport across a 3-periodic pentagon - only grain boundary . in particular , starting from its microscopic tight - binding model , we have derived the boundary condition seen by the massless dirac fermions at such grain boundary . with it
we have calculated the scattering coefficients , from which we concluded that the valley preserving and intervalley scattering probabilities are highly responsive to external manipulation through control of the grain boundary s microscopic details .
we have also made use of the generalized potential representation of the grain boundary to gain insight into the obtained results .
j. n. b. r. thanks joo m. b. lopes dos santos and nuno m. r. peres for many insightful discussions and helpful comments .
the author also acknowledges the faculdade de cincias da universidade do porto for the warm support during the initial stages of this work , as well as the singapore national research foundation , the prime minister office , the singapore ministry of education and yale - nus college for their fellowship programs ( nrf - nrff2012 - 01 , r-144 - 000 - 295 - 281 and r-607 - 265 - 01312 ) .
in this section we will give details of the derivations presented in section [ sec : tbmodel-3po ] concerning the microscopic tight - binding model of monolayer graphene with a 3-periodic pentagon - only grain boundary .
the calculations below closely follow what was done in ref . .
in sub - section [ app : bulktb ] we concentrate on the calculations leading to the bulk transfer matrix , while in sub - section [ app : defecttb ] we focus on the calculations giving rise to the tight - binding boundary condition that originates from the presence of the 3-periodic pentagon - only grain boundary .
the matrix @xmath91 present in the bulk tight - binding equations [ see eqs .
( [ eq : bulktbeqs ] ) ] reads @xmath202 \ , .
\label{eq : wamatrix}\end{aligned}\ ] ] the bulk tight - binding equations [ see eqs .
( [ eq : bulktbeqs ] ) ] can be cast in the form [ eq : dfctbeqsbulk ] @xmath203 & = & \mathbb{q}_{1 } \left[\begin{array}{c } \mathbf{a}(n)\\ \mathbf{b}(n-1 ) \end{array}\right ] \,,\label{eq : bulk1}\\ \left[\begin{array}{c } \mathbf{a}(n)\\ \mathbf{b}(n-1 ) \end{array}\right ] & = & \mathbb{q}_{2 } \left[\begin{array}{c } \mathbf{b}(n-1)\\ \mathbf{a}(n-1 ) \end{array}\right ] \ , , \label{eq : bulk2 } \end{aligned}\ ] ] where the matrices @xmath101 and @xmath102 read [ eq : q1q2 ] @xmath204 \ , , \\
\mathbb{q}_{2 } & = & -\left[\begin{array}{cc } \frac{\epsilon}{t}\big(w_{a}\big)^{-1 } & \big(w_{a}\big)^{-1}\\ -\mathbb{i}_{3 } & 0 \end{array}\right ] \ , , \end{aligned}\ ] ] with @xmath205 standing for the @xmath206 unit matrix .
( [ eq : dfctbeqsbulk ] ) can be written in the form of a transfer matrix , as was done in eq .
( [ eq : transfmateq ] ) , where the transfer matrix is given by eq .
( [ eq : transfermatrixcoup ] ) . the matrix @xmath98 present in the latter equation reads @xmath207 \ , .
\label{eq : rmatrix}\end{aligned}\ ] ] this matrix simply changes from the basis @xmath208 into the basis @xmath209 .
the matrix @xmath210 enforcing the basis change that uncouples the modes of the transfer matrix reads @xmath211 \ , , \label{eq : lambdamat } \end{aligned}\ ] ] where @xmath212 . in this basis
, the three @xmath27 matrices in the diagonal of the transfer matrix @xmath213 are noted by @xmath106 , @xmath107 and @xmath108 .
the matrix [ eq : tmdiags ] @xmath214 \ , , \nonumber \\ \label{eq : tmdiags1 } \end{aligned}\ ] ] corresponds to the high - energy mode when we are around @xmath113 .
similarly , the matrices corresponding to the low - energy modes ( one from each dirac point ) read @xmath215 \ , , \label{eq : tmdiags2 } \\ \mathbb{t}_{l+}(\epsilon , k_{x}a ) & = & \theta(k_{x } ) \left [ \begin{array}{cc } -1 & -\epsilon \\ \epsilon & \frac{e^{-i ( k_{x } a + \frac{\pi}{3})}-1}{g(k_{x } ) } + \epsilon^{2 } \end{array } \right ] \ , , \label{eq : tmdiags3 } \end{aligned}\ ] ] where we have defined @xmath216 and @xmath217 as @xmath218 the matrix @xmath115 present in the tight - binding equations at the grain boundary region [ see eqs .
( [ eq : dfctbeqs ] ) ] reads @xmath219 \ , , \label{eq : sigmaxp}\end{aligned}\ ] ] where @xmath6 , @xmath7 and @xmath8 stand for the hopping parameters at the grain boundary region as represented in fig .
[ fig : scheme-3fpentonly ] . eqs .
( [ eq : dfctbeqs ] ) can be condensed in the form [ eq : dfctbeqs3 ] @xmath220 & = & \mathbb{n}_{1 } \left[\begin{array}{c } \mathbf{a}(1)\\ \mathbf{b}(0 ) \end{array}\right ] \,,\label{eq : bc1}\\ \left[\begin{array}{c } \mathbf{a}(1)\\ \mathbf{b}(0 ) \end{array}\right ] & = & \mathbb{n}_{2 } \left[\begin{array}{c } \mathbf{b}(0)\\ \mathbf{d}(0 ) \end{array}\right],\label{eq : bc2}\\ \left[\begin{array}{c } \mathbf{b}(0)\\ \mathbf{d}(0 ) \end{array}\right ] & = & \mathbb{n}_{3 } \left[\begin{array}{c } \mathbf{d}(0)\\ \mathbf{a}(0 ) \end{array}\right ] \,,\label{eq : bc3}\\ \left[\begin{array}{c } \mathbf{d}(0)\\ \mathbf{a}(0 ) \end{array}\right ] & = & \mathbb{n}_{1 } \left[\begin{array}{c } \mathbf{a}(0)\\ \mathbf{b}(-1 ) \end{array}\right ] \,,\label{eq : bc4}\\ \left[\begin{array}{c } \mathbf{a}(0)\\ \mathbf{b}(-1 ) \end{array}\right ] & = & \mathbb{n}_{2 } \left[\begin{array}{c } \mathbf{b}(-1)\\ \mathbf{a}(-1 ) \end{array}\right ] \,.\label{eq : bc5 } \end{aligned}\ ] ] where the matrices @xmath221 , @xmath222 and @xmath223 are @xmath224 matrices which read [ eq : nmatricesdfcln ] @xmath225 \ , , \\ \mathbb{n}_{2 } & = & -\left[\begin{array}{cc } \frac{\epsilon}{t}\big(w_{a}\big)^{-1 } & \big(w_{a}\big)^{-1}\\ -\mathbb{i}_{3 } & 0 \end{array}\right ] \
, , \\
\mathbb{n}_{3 } & = & -\left[\begin{array}{cc } \big(\frac{\epsilon}{t}\mathbb{i}_{3}+\sigma_{x}'\big ) & \mathbb{i}_{3}\\ -\mathbb{i}_{3 } & 0 \end{array}\right ] \ , .
\end{aligned}\ ] ] the above matrices depend on the reduced energy , @xmath121 , on the longitudinal momentum , @xmath87 , and on the hopping parameters at the defect , @xmath6 , @xmath7 and @xmath8 .
when @xmath126 and @xmath226 , the boundary condition matrix ( expressed in the basis uncoupling the modes of the transfer matrix @xmath227 ) @xmath125 reads @xmath228 \ , .
\label{eq : matdiracpnt}\end{aligned}\ ] ] in this limit , the @xmath27 matrix @xmath106 describing the high energy modes of the transfer matrix @xmath109 [ see eq .
( [ eq : tmdiags1 ] ) ] reads @xmath229 \ , .
\label{eq : tmdiags1-lowe - limit}\end{aligned}\ ] ] it is then straightforward to understand what will be the relation between the high - energy modes amplitudes @xmath230 at position @xmath231 , and the amplitudes @xmath232 at position @xmath233 : since @xmath234 , the _ upper _ high - energy mode , i.e. @xmath235 , is going to decrease exponentially with @xmath128 , while the _ lower _ one , i.e. @xmath236 , is going to exponentially increase because @xmath237 .
therefore , the requirement that the wave function be normalizable , implies that @xmath238 must have the form of eq .
( [ eq : lowen - states ] ) . hence ,
as described in the main text , we can then eliminate the high - energy modes from the problem , and write the effective boundary condition seen by a low - energy electron ( hole ) inciding in the 3-periodic pentagon - only from infinity .
in particular , the matrix @xmath239 obtained from eq .
( [ eq : meffgen ] ) reads @xmath240 \ , , \label{eq : meff } \end{aligned}\ ] ] where @xmath148 and @xmath149 can be written as [ eq : fandg ] @xmath241 the boundary condition seen by the massless dirac fermions is finally given by substituting eq .
( [ eq : tb = d ] ) in @xmath242 .
since @xmath243 , such condition can be recast as @xmath244 where @xmath245^{t}$ ] ( we have omitted the dependence of the @xmath246 components on @xmath2 and @xmath247 ) and @xmath248 $ ]
( [ eq : bc - provis ] ) can be written as @xmath145 with @xmath38 given in eq .
( [ eq : bcmatrixp3 ] ) .
finally , the transmission coefficients @xmath179 , @xmath181 , @xmath180 and @xmath182 for the 3-periodic pentagon - only grain boundary can be shown to have the following analytic expressions : [ eq : transmission - coefs ] @xmath249 it is possible to show that , for the 3-periodic pentagon - only grain boundary with any hopping parameters values @xmath6 , @xmath7 and @xmath8 , the intervalley scattering is always the same for incoming electrons either living on the valley @xmath174 or on the valley @xmath173 , i.e. @xmath250 . similarly , we can also show that @xmath251 , and thus the transmission plots ( see figs . [
fig : transmittances-3fpentonly ] and [ fig : transmittances-3fpentonly__2 ] ) are always symmetric ( upon interchanging of valley ) over the @xmath252 angle .
in this appendix we will show how can we connect the two perspectives discussed in section [ sec : lowen-3periodic ] for the low - energy electronic scattering off a periodic grain boundary .
in particular , we will show how can we compute the generalized potential @xmath74 in eq .
( [ eq : genpotv ] ) in terms of the boundary condition matrix @xmath38 originating from the tight - binding model of the grain boundary . as discussed in section [ sec : lowen-3periodic ] , in the low energy continuum limit of the tight - binding model we can see the grain boundary as a finite width strip where the dirac fermions are governed by the following hamiltonian @xmath253 where the @xmath24 and @xmath25 ( @xmath26 ) stand for the @xmath27 pauli matrices acting on , respectively , the valley and the pseudo - spin degrees of freedom . similarly , @xmath28 and @xmath29 stand for the @xmath27 identity matrix acting on each of these sub - spaces . note that in the above equation ( and in the remaining of this appendix ) we have set @xmath30 .
the term @xmath74 in eq .
( [ eq : genham ] ) stands for a generalized potential acting on graphene s massless dirac fermions . by forcing this generalized potential @xmath74 to be hermitian
[ see general expression in eq .
( [ eq : genpotv ] ) ] we ensure that the boundary condition matrix @xmath38 conserves the flux , @xmath254 , as required .
furthermore , the time - reversal invariance of @xmath38 ( whenever the grain boundary is non - magnetic ) is ensured by requiring that @xmath74 is also time - reversal invariant .
given this , and before proceeding , let us briefly analyze the effect of each of the terms @xmath255 on the eigenmodes living inside the finite width strip .
the terms @xmath256 act equally on both valleys .
the term proportional to @xmath163 represents an electrostatic potential analogous to that generated by gating graphene or by the presence of charge impurities in the vicinity of the graphene flake .
the term @xmath257 is a mass term equivalent to that originating whenever the atoms of each sub - lattice have different onsite energies .
terms proportional to @xmath164 and @xmath258 are analogous to the @xmath2- and @xmath247-component of a vector potential @xmath259 and @xmath260 arising from the presence of a magnetic field perpendicular to the graphene layer .
the terms @xmath261 can be viewed as analogous to those originating from a pseudo - magnetic field generated by deformations of the honeycomb lattice .
all the other terms , @xmath262 ( with @xmath263 and @xmath264 ) , give rise to eigenstates that _ live _ in both valleys simultaneously ( see below ) , thus giving rise to intervalley scattering .
we will now show how can we express the boundary condition matrix @xmath38 in eq .
( [ eq : generalbc - m ] ) in terms of the generalized potential @xmath74
. we shall start by using the fact that the problem is translation invariant along the grain boundary direction , @xmath265 , so that we can write the eigenspinors as @xmath266 which allows us to rewrite eq .
( [ eq : genham ] ) as @xmath267 this expression can be cast as @xmath268 where the operator @xmath269 reads @xmath270 integrating the differential equation , one obtains the following relation between the two sides of the strip @xmath271 which , if we take the limit @xmath272 when @xmath77,@xcite then becomes @xmath273 in eq .
( [ eq : phi - bc ] ) the boundary condition matrix @xmath38 reads @xmath274 with @xmath275 reading @xmath276 where @xmath277
. the generalized potential @xmath74 will be hermitian if all the @xmath278 are real numbers .
time - reversal symmetry requires that @xmath279 .
therefore , @xmath275 reads @xmath280 thus , the argument of the exponential in eq .
( [ eq : bcexp ] ) can be recast as @xmath281 \,,\end{aligned}\ ] ] in order to determine which generalized potential terms are present whenever we have a boundary condition matrix as that of eq .
( [ eq : bcmatrixp3 ] ) we can use lagrange - sylvester interpolation,@xcite which allows us to express the function of a diagonalizable matrix @xmath134 as @xmath282 where @xmath283 are the eigenvalues of the matrix @xmath134 . the matrices @xmath284 stand for the frobenius covariants of matrix @xmath134.@xcite these are given by @xmath285 where @xmath286 identifies the identity matrix . by computing @xmath287
, we will be able to express the coefficients @xmath278 as functions of @xmath288 and @xmath289 appearing in the expression for the boundary condition matrix @xmath38 , eq .
( [ eq : bcmatrixp3 ] ) .
the non - zero terms of the generalized potential originating from the 3-periodic pentagon - only grain boundary read ( the principal value of the logarithm was taken ) [ eq : vs - expr2 ] @xmath295 \ , , \\
x_{\pm } & \equiv & \sqrt{-4 + f_{r}^{2 } + \vert g \vert^{2 } \pm 2 f_{r } \vert g \vert } \ , , \\
y_{\pm } & \equiv & \vert g \vert^{2 } \pm f_{r } \vert g \vert \ , , \\
z_{\pm } & \equiv & f_{r } \pm \vert g \vert \ , , \end{aligned}\ ] ] we can readily conclude from the above expressions that , for a general choice of the hopping parameters of the 3-periodic pentagon - only grain boundary , @xmath6 , @xmath7 and @xmath8 , the dirac fermions will feel a generalized potential both containing terms that do not mix the valleys ( namely , @xmath300 and @xmath301 ) , and terms that do mix valleys ( such as @xmath302 , @xmath303 , @xmath304 and @xmath305 ) .
let us briefly examine the implications of the presence and absence of these terms
. start by noting that when we force @xmath306 the valley mixing terms vanish ( i.e. , @xmath307 ) , and only the valley preserving terms ( @xmath300 and @xmath301 ) are present inside the strip . as a consequence
, there will be no intervalley scattering , just as expected : remember that when @xmath308 we recover the pentagon - only grain boundary which has a periodicity that does not map the dirac points into the same @xmath111 , thus forbidding low - energy intervalley scattering.@xcite this can be also concluded from the boundary condition matrix @xmath38 expression when @xmath308 [ given in eq .
( [ eq : bcmatrix - simplepo ] ) ] : it has no off - diagonal ( intervalley scattering ) elements . when @xmath308 there will always be an angle @xmath309 with perfect transmittance , i.e. with @xmath310
this can be understood by noting that for this particular angle of incidence @xmath309 it is possible to perfectly match the wave - function immediately inside the strip ( at @xmath311 ) and that immediately outside the strip ( at @xmath312 ) without the need to use reflected modes . as argued in ref .
, we can see this by comparing the spinors of the modes inside the strip ( dirac modes subject to a generalized potential with the terms @xmath300 and @xmath301 ; these are non - chiral due to @xmath313 ) and the spinor of the incident mode : for the angle @xmath309 the incident mode s spinor is exactly equal to that of a positive - propagating mode inside the strip .
let us now focus on the valley - mixing terms of the generalized potential @xmath74 .
both the terms @xmath302 and @xmath304 give rise to a shift of the energy cones along the @xmath314-direction ( which causes a deflection of the incoming mode ) , resembling what happens when a constant gauge potential term @xmath315 is present . the latter term s eigenstates ( as well as those of @xmath301 terms )
have a well defined valley quantum number , but are non - chiral ( pseudo - spin not aligned with momentum ) .
similarly , the @xmath302 and @xmath304 eigenstates are non - chiral . more importantly , and unlike what happens with the gauge term @xmath315 ( and @xmath301 ) , the terms @xmath302 and @xmath304 mix the two valleys ,
i.e. their eigenstates do not have a well defined valley quantum number .
similarly , we can also show that both the terms @xmath303 and @xmath305 open a gap in the spectrum , resembling what happens when a mass term ( i.e. @xmath316 ) is present . however , unlike the latter , the former potential terms eigenstates are both non - chiral and mix the two valleys .
the fact that ( for general values of the hopping renormalizations @xmath122 ) there are valley mixing potential terms inside the strip , implies that its modes do not have a well defined valley quantum number , i.e. strip eigenstates _ live _ in both valleys .
therefore , a wave function ( living only on the valley @xmath317 ) incoming from @xmath318 , will in general require reflected modes ( in both valleys ) in order to match the wave function inside the strip .
that is , in general there will not be an angle of perfect transmission ( of low - energy carriers ) at the 3-periodic pentagon - only grain boundary , i.e. @xmath319 . only for very particular cases , and by fine tuning the values of the hopping parameters at the grain boundary will perfect transmission occur .
42ifxundefined [ 1 ] ifx#1 ifnum [ 1 ] # 1firstoftwo secondoftwo ifx [ 1 ] # 1firstoftwo secondoftwo `` `` # 1''''@noop [ 0]secondoftwosanitize@url [ 0 ]
+ 12$12 & 12#1212_12%12@startlink[1]@endlink[0]@bib@innerbibempty link:\doibase 10.1126/science.1171245 [ * * , ( ) ] link:\doibase 10.1021/nl801827v [ * * , ( ) ] link:\doibase 10.1038/nature07719 [ * * , ( ) ] link:\doibase 10.1038/nnano.2010.132 [ * * , ( ) ] link:\doibase 10.1021/nl801386 m
[ * * , ( ) ] link:\doibase 10.1038/nnano.2010.53 [ * * , ( ) ] link:\doibase 10.1038/nature09718 [ * * , ( ) ] link:\doibase 10.1021/nn1033423 [ * * , ( ) ] link:\doibase 10.1063/1.3610941 [ * * , ( ) ] link:\doibase 10.1103/physrevb.92.205402 [ * * , ( ) ] link:\doibase 10.1038/nnano.2014.166 [ * * , ( ) ] link:\doibase 10.1002/adma.201401389 [ * * , ( ) ] link:\doibase 10.1038/ncomms5911 [ * * ( ) , 10.1038/ncomms5911 ] http://stacks.iop.org/2053-1583/2/i=2/a=024008 [ * * , ( ) ] link:\doibase 10.1126/science.1196893 [ * * , ( ) ] link:\doibase 10.1126/science.1235126 [ * * , ( ) ] link:\doibase 10.1038/nmat3010 [ * * , ( ) ] link:\doibase 10.1016/j.ssc.2011.05.023 [ * * , ( ) ] link:\doibase 10.1126/science.1218948 [ * * , ( ) ] link:\doibase 10.1021/acs.nanolett.5b04234 [ * * , ( ) ] link:\doibase 10.1021/jp411464w [ * * , ( ) ] link:\doibase 10.1039/c6nr01359 g [ * * , ( ) ] link:\doibase 10.1038/nmat2830 [ * * , ( ) ] link:\doibase 10.1038/srep12508 [ * * , ( ) ] link:\doibase 10.1103/physrevb.91.075425 [ * * , ( ) ] link:\doibase 10.1021/acsnano.5b05879 [ * * , ( ) ] link:\doibase 10.1021/nl200369r [ * * , ( ) ] link:\doibase 10.1021/nl301141 g [ * * , ( ) ] link:\doibase 10.1103/physrevb.89.121407 [ * * , ( ) ] link:\doibase 10.1021/ja5054847 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.106.136806 [ * * , ( ) ] link:\doibase 10.1016/j.physleta.2011.10.043 [ * * , ( ) ] link:\doibase 10.1103/physrevb.86.165433 [ * * , ( ) ] link:\doibase 10.1103/physrevb.86.214206 [ * * , ( ) ] link:\doibase 10.1088/0953 - 8984/25/7/075303 [ * * , ( ) ] link:\doibase 10.1088/0953 - 8984/26/12/125502 [ * * , ( ) ] link:\doibase 10.1103/physrevb.92.045426 [ * * , ( ) ] link:\doibase 10.1039/c0nr00820f [ * * , ( ) ] link:\doibase 10.1039/c4cp02552k [ * * , ( ) ] link:\doibase 10.1088/0957 - 4484/25/11/115702 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.53.2449 [ * * , ( ) ] http://dx.doi.org/10.1017/cbo9780511840371[__ ] ( , )
|
we study how low - energy charge carriers scatter off periodic and linear graphene grain boundaries oriented along the zigzag direction with a periodicity three times greater than that of pristine graphene .
these defects map the two dirac points into the same position , and thus allow for intervalley scattering to occur . starting from graphene s first - neighbor tight - binding model
we show how can we compute the boundary condition seen by graphene s massless dirac fermions at such grain boundaries .
we illustrate this procedure for the 3-periodic pentagon - only grain boundary , and then work out the low - energy electronic scattering off this linear defect .
we also compute the effective generalized potential seen by the dirac fermions at the grain boundary region .
|
spectral imaging in general , and solar spectral imaging in particular , suffer from a fundamental problem in detector physics .
spectral images have three independent variables @xmath0 , while current image detectors only support two independent variables @xmath2 and integrate over wavelength @xmath3 . conventional techniques to overcome this problem generally use time - multiplexing : in filtergraph imaging spectroscopy ,
a narrow band filter is tuned slowly across the spectral range of interest and an image collected at each discrete @xmath3 ; in conventional scanned - slit imaging spectroscopy , the light is passed through a spatial filter ( the slit ) , selecting a single @xmath4 , and the remaining light is dispersed to project @xmath3 onto the detector s @xmath4 axis .
even more sophisticated techniques such as fourier imaging spectroscopy use time multiplexing to collect multiple two - dimensional basis images of the three - dimensional data space .
multiplexing in time is photon - inefficient as photons that are not part of the current sample are discarded .
that is a problem because solar measurements are photon starved : instruments must race to collect sufficient photons for a spectral measurement , before the features on the sun change .
stereoscopic spectral imaging overcomes the problem of spectral imaging by analogy between the spectral imaging problem and the stereoscopic problem of determining feature distance in ordinary 3-space
. dispersed spectral images are integrals along diagonal lines in @xmath0 space ; they are analogous to images of a 3-space subject , with the images collected at a `` look angle '' that depends on the dispersion of the instrument . collecting
multiple dispersed images in different spectral orders yields data that can be inverted stereoscopically to measure some spectral characteristics of the line everywhere in the image plane simultaneously .
@xcite and @xcite have described using this stereoscopic approach to simultaneous imaging and spectroscopy in euv emission lines .
moses _ rocket payload , launched 2006 february 8 , obtained doppler measurements of explosive events and jets in the he ii line at @xmath5 nm ( @xcite ) .
another related technique , computed - tomography imaging spectroscopy ( ctis ) uses multiple spectral orders for hyperspectral imaging and spectropolarimetry at low spectral resolution in the visible and infrared ( e.g. @xcite ) .
deforest et al . (
2004 ) have developed a first - order theory of stereoscopic inversion that is applicable to a much simpler spectral context : measuring doppler shift and the longitudinal zeeman splitting in a narrow visible absorption line in the solar photosphere . that work is conceptually similar to the _
moses _ stereoscopic imaging , but there are two key differences : there is only a single spectral line in the instrument passband , and it is an absorption line ; and the line width is narrower than a pixel in dispersion .
deforest et al . demonstrated the technique with proof - of - concept observations of the quiet sun , a decayed active region , and a clean sunspot .
however , in that work they did not analyze the noise characteristics of the observing and inversion technique , only report the _ a posteriori _ noise measured in a quiet sun region . in this article
, we derive expressions for the _ a priori _ noise level ( hence sensitivity ) and systematic error of a stereoscopic instrument in the narrow - line regime described by @xcite using hybrid differential stereoscopy to determine line shift .
we use _ a priori _ statistical calculation to derive the effect of the dominant source of uncorrelated noise - photon counting statistics - and also address quasi - random systematic errors in the inversion using some basic assumptions about the scene being viewed .
the analytic expressions are general and may be applied to stereoscopic spectrographs viewing either absorption or emission lines , but we consider them in the specific context of an absorption - line zeeman magnetograph viewing the relative line shift between the right- and left - circular polarizations . in [ sec : hybrid - differential - stereoscopy ] we analyze each of several noise sources individually ; in [ sec : summary-&-specific ] the results of the individual analyses are applied to generate a noise budget for an example baseline instrument observing the quiet photosphere with ~0.1 arc sec resolution and 0.03 arc sec pixels , in the 617.34 nm fe i absorption line .
finally , in [ sec : conclusions ] we draw conclusions about the types of observation for which hybrid differential stereoscopic spectral imaging offers the best prospects for advances over the state of the art . throughout the discussion we have used @xmath6 to indicate a single difference from the correct or expected value of the following quantity , and angle - brackets to denote rms averages over an image .
here we develop a theory of noise in a hybrid differential analysis of absorption line data .
following @xcite , consider an image in the @xmath2 plane produced by observing , through a slitless dispersing spectrograph , a narrow absorption line of rest wavelength @xmath7 , offset @xmath8 from that wavelength , constant width @xmath9 , and integrated total intensity absorption @xmath10 over that width ; limiting the wavelength range on the detector is a filter with transmission profile @xmath11 and total admitted continuum intensity @xmath12 in the absence of the spectral line . without loss of generality ,
take the dispersion direction of the spectrograph to be along the @xmath4 axis and consider the images collected in the @xmath13 spectral order .
then the @xmath14 variation of the image drops out of the analysis and inversion may be performed independently along each line parallel to the @xmath4 axis .
variations in the central wavelength of the spectral line result in spatial distortions of the images ; these distortions are antisymmetric across the two spectral orders , and can be used to recover low spatial frequencies of the function @xmath15 by _
correlation stereoscopy_. small patches of the images are cross - correlated to determine the offset functions @xmath16 , and the wavelength offset can then be determined by @xmath17 where @xmath18 is the dispersion of the + 1 order .
the @xmath19 subscripts refer to low spatial frequencies ; the maximum spatial frequency that may be resolved by this method is determined by the patch size used for the cross - correlation .
the inversion used for eq .
[ eq : low - freq dispersion ] is analogous to the inversion carried out by the human visual system , determining the @xmath20 coordinate of observed objects by cross - correlating the visual fields from the left and right eye . recovering higher spatial frequencies
requires a more sophisticated inversion .
deforest ( 2003 ) and deforest et al .
( 2004 ) developed the process of _ differential stereoscopy _ to determine high spatial frequencies in @xmath21 based on first - order intensity variation in the dispersed images .
the difference between the + 1 and -1 order dispersed images yields several terms in first approximation order , one of which is proportional to @xmath22 . by integrating the difference signal along the @xmath4 direction it is possible to recover an approximation to the high spatial frequencies @xmath23 throughout the image plane because the primary detection term is to the spatial derivative , low spatial frequencies are not well measured with that technique .
furthermore , differential stereoscopy is sensitive to `` leakage '' of intensity and other signals into the @xmath22 measurement .
these terms are proportional to @xmath21 itself and must therefore be compensated in some way in strong field regions . by resampling the dispersed images to eliminate the spatial shifts
related to @xmath24 before applying differential stereoscopy , it is possible to eliminate image - related noise in the differential signal , to first approximation order .
the differential process is used to recover high spatial frequencies , and the low spatial frequencies are inserted via a mathematical filter . the resulting _ symmetric hybrid inversion equation _ for first - order stereoscopic inversion ( from deforest et al .
2004 ) is:@xmath25 where @xmath24 is again the low - spatial - frequency component of the central - wavelength shift image , as determined by explicit cross - correlation of patches of the + 1 image with corresponding regions of the -1 image , @xmath26 , and @xmath27 are again distorted spatial coordinates that use the correlation signal to remove measured offsets in the images , @xmath28 is the intensity profile of each image as a function of focal - plane coordinate @xmath4 or the distorted focal - plane coordinate @xmath29 , @xmath18 is the dispersion of the instrument , and @xmath30 is a small convergence factor that splices the frequency spectra of the two derivation methods .
note that eq .
[ eq : hybrid_inversion ] is implicit rather than closed - form : @xmath31is present both on the left hand side of the equality , and also inside the integral on the right hand side .
the presence of the corrective filter term @xmath32 is what makes eq .
[ eq : hybrid_inversion ] a _ hybrid _ inversion : it preserves the high spatial frequencies present in the main portion of the integrand , while forcing the low spatial frequencies in the integral to match the @xmath24 signal obtained by direct cross - correlation of image patches .
if @xmath24 happens to be constant and zero , then the impulse response function of the resulting spatial filter is a decaying exponential . in practice ,
the integral should be performed numerically in both the positive - x and negative - x direction , and the two results summed .
the forward integral is causal along the @xmath4 axis and the reverse integral is anti - causal , so that the resulting impulse response function is symmetric in @xmath4 .
the discrete form of equation [ eq : hybrid_inversion ] , then , is best calculated using the sum of two discrete filter transforms .
let @xmath33 be the normalized difference image in compensated coordinates , interpolated to ( integer ) non - distorted pixel locations , and let @xmath34 run over the pixel coordinates in the @xmath4 direction
. then @xmath33 may be written:@xmath35 where the @xmath36 subscript is to indicate that the @xmath28 images have been interpolated so that the @xmath37 pixel is found from the focal - plane location @xmath38 in the first - order theory developed by deforest et al .
2004 , @xmath33 is important because , when @xmath39@xmath40 which allows recovery of @xmath21 by integrating and scaling appropriately .
the integral becomes a sum of two discrete spatial filters in the _ discrete symmetric hybrid inversion equation_:@xmath41 where pixel centers are considered to exist at all integers between 0 and @xmath42 , and@xmath43 is the average ratio of admitted continuum intensity to line absorption intensity .
the @xmath44 terms merely produce a smoothed version of @xmath24 : if @xmath24 were constant , and the image were infinite in extent , then each of the two summations would return exactly @xmath24 by the law of geometric sums .
the low pass filter from this smoothing effect exactly matches the high - pass filter applied to the integral by the exponential decay term , so that the two versions of @xmath21 are combined together seamlessly .
equation [ eq : discrete inversion ] , together with the process of cross - correlation to determine the relationship between @xmath29 and @xmath4 , is a complete discrete image inversion .
step response of the filter in eq .
[ eq : discrete inversion ] shows how the different components of hybrid inversion combine to reconstruct the wavelength shift signal .
the source wavelength ( black ) is a discrete heaviside function with step at pixel 100 .
the differential stereoscopy step ( in the absence of noise ) yields a high - pass filtered version of the heaviside function ( blue ) , while the correlation step yields a low - pass filtered version ( cyan ) .
because the filters are constructed mathematically to match one another , the sum of the two filters ( magenta ) reproduces the step .
the low - pass curve suffers from edge effects at the far sides of the image , but in a typical detector with more than @xmath45 pixel width , edge effects are minimal . ]
figure [ fig : step - response ] shows the step response of the filter described by eq .
[ eq : discrete inversion ] , to illustrate how the different parts of the inversion work together .
the step is located at position 100 .
the high - pass response to the differential stereoscopy signal ( dark blue ) is due to the interaction between the differentiation inherent in the detection method , and the `` imperfect '' ( high - pass ) integration imposed by the @xmath30@xmath31 decay term in the integrand of eq .
[ eq : hybrid_inversion ] .
the exponential decay is matched by the low - pass response to the correlation stereoscopy signal ( light cyan ) .
the sum of the two signals is the value of the the rhs of eq .
[ eq : discrete inversion ] , and reproduces the original signal .
the slight overshoot of the reconstructed ( magenta ) curve is a discretization effect due to the phase of the heaviside function relative to the pixel grid : a heaviside function with the opposite phase relative to the pixel grid would yield an equal - magnitude undershoot . as with any discrete sampling filter , eq .
[ eq : discrete inversion ] operates best on spatial frequencies less than 2/3 of the pixel nyquist frequency , because the reconstructed amplitude of those frequency components is independent of their phase relative to the pixel grid .
the heaviside function contains frequencies both below and above that limit , and the high frequencies reconstructed amplitude depends on phase . in practice ,
errors in both the measurement of intensity and the inversion affect the analysis , so that the measured @xmath31 is not exactly the same as @xmath21 even if the image frequency spectrum is appropriately limited . in this error analysis
we consider six principal sources of error : photon counting noise , @xmath46 , that represents an additive , uncorrelated noise source at each pixel in each of the differential and correlation stereoscopy steps ; misalignment noise @xmath47 in the cross - correlation step ; the effect of intensity gradients ( both @xmath48 and @xmath49 ) on the line shift signal ; a nonzero slope @xmath50 of the instrument filter transmission function @xmath11 ( the slope is assumed to be zero in eq .
[ eq : hybrid_inversion ] ) ; and an excessively high @xmath22 signal in the solar image .
the first two noise sources arise ( directly or indirectly ) from imprecision in the measurement of @xmath51 in each pixel , and the last four sources arise from the non - ideal nature of the instrument and/or the solar field being imaged . in the following subsections
we treat each error source independently .
photon counting noise is due to uncertainty in the measured intensity @xmath28 in each focal plane , through the statistics of photon counting .
each measurement does not determine intensity directly , but rather fluence of energy ( or , equivalently , photon count ) on each pixel during an exposure time .
intensity can be derived by knowing the size of the pixel and the length of the exposure .
each pixel measures a fluence @xmath52 with an uncertainty @xmath53 given by:@xmath54 so that the rms value @xmath55 .
this effect contributes to @xmath31 primarily via a photon noise term that must be propagated through equation [ eq : discrete inversion ] : the much higher photon counts in many - pixel image `` patches '' greatly reduce photon counting noise in the cross - correlation portion of the inversion ( [ sub : misalignment - noise ] ) .
the rms photon noise contribution @xmath56 is just the incoherent sum of the photon noise terms in each term on the rhs of eq .
[ eq : discrete inversion ] . each of those terms , in turn , is ( to first order in @xmath57 and neglecting @xmath58 for the calculation of noise in each pixel):@xmath59 where @xmath60 is the per - exposure photon count in the corresponding pixel in each of the two orders .
[ eq : delta_ph j ] simply reproduces the familiar behavior of incoherent counting statistics for intrinsic ( normalized ) quantities .
the treatment may , at first glance , appear overly simple as the individual pixels in the images are interpolated and hence do not represent individual samples of the @xmath61 random variable but rather scaled incoherent sums of adjacent samples .
this is acceptable to ignore because , in the absence of strong gradients in @xmath62 ( i.e. gradients comparable to unity per pixel ) , the samples noise profile is preserved by the inversion .
adjacent pixels are co - added in the discrete inversion . under linear interpolation , adjacent interpolated fractions of the signal from each detector pixel are approximately conserved by the interpolation and added coherently in the cross - pixel sum in eq .
[ eq : discrete inversion ] . because addition is commutative
, the noise effect of the interpolated pixels on the final sum is the same as if no interpolation had taken place , up to a negligible factor of @xmath63 ) . put another way , although considering interpolation would further reduce the noise in @xmath64 for individual pixels , co - adding @xmath64 for adjacent pixels in the inversion eq .
[ eq : discrete inversion ] would then include coherent cross - pixel noise terms that would exactly cancel the reduction
so we may safely choose , instead , to ignore the effects of the @xmath29 interpolation and use incoherent sums throughout .
carrying eq .
[ eq : delta_ph j ] forward through eq .
[ eq : discrete inversion ] yields:@xmath65 neglecting any field edge effects ( in other words , taking @xmath66 ) permits taking the infinite sums directly , yielding ( to first order in epsilon):@xmath67 thus the statistics of the differential stereoscopy step are worsened by the width of the correction term .
this is reflective of the fact that random walks diverge , and is the reason for using a hybrid inversion rather than a direct integral inversion : using direct integration in equation [ eq : discrete inversion ] , rather than high - pass integration with correction , is equivalent to setting @xmath30 to zero , which would yield an arbitrarily large amount of photon - counting noise in the individual measurements . in actual use , it is necessary to balance @xmath30 and the cross correlation patch size : larger patch sizes improve correlation behavior but require smaller @xmath30 and thus worsen the photon noise due to integration of the differential signal .
it is important to note that the photon noise from the discrete inversion is not independently sampled at each pixel : it is the sum of many noise terms along the @xmath4 direction , so that the _ difference _ measurement between adjacent pixels has a much smaller noise level , yielding higher precision for small features .
in particular , shifting by 1 pixel re - samples only one @xmath68 of the population of random samples in the calculated @xmath31 , reducing the photon noise by a factor of @xmath69 so that small features are only subject to : @xmath70 which is to be expected considering that the difference signal is the original signal being measured by the differential stereoscopy - the difference in measured wavelength between two adjacent pixels is simply a linearly scaled form of @xmath64 .
differential stereoscopy is _ more sensitive _ to small features than to large ones , at least from the standpoint of photon statistics .
taking `` typical '' values of @xmath71 ( fwhm of 15 pixels ) , @xmath72 , and @xmath73 yields an absolute partial sensitivity @xmath74 pixel , and a gradient sensitivity @xmath75 ( the gradient sensitivity is unitless : pixels per pixel ) , due to photon counting alone .
the second source of error in equation [ eq : discrete inversion ] is misalignment in the cross correlation step that is used to derive @xmath29 from the original images . in the 2004 derivation of eq .
[ eq : hybrid_inversion ] , it was necessary to assume that the displacement of the images was small , leading to @xmath58 terms that enter in first order because the + 1 and -1 order intensities are measured at slightly different spatial locations . here , we consider the effect of small misalignments due to errors in the correlation co - alignment process . in the test measurements made by deforest et al .
, the cross - correlation step used a full two - dimensional patch of image around each sampled point , rather than ( as in the differential stereoscopic portion of the inversion ) merely the particular horizontal line surrounding each point .
this is reasonable in part because the image is , in fact , a two - dimensional field rather than a collection of independent rasters ; and in part because the technique was found empirically to reduce correlation `` misses '' and improve the image signal . here , we consider the potential sources of error injected by cross - correlation alignment of patches of image around each point . because correlation co - alignment is a nonlinear process and depends on the field in the image
, it is necessary to make some assumptions about the field being viewed .
we here confine our analysis to magnetic measurements of the quiet sun , which consists principally of nearly uniformly bright granules separated by dark , well defined lanes .
some other types of field ( for example , the uv chromospheric network ) are analogous and may be considered with the same or similar analysis , but other types of field ( such as sunspot penumbrae ) are likely too complex for simple analysis and require numerical modeling .
sample correlation field used for our correlation photon analysis .
correlation between two images is used to align two patches of pixels with radius @xmath76 .
the image field consists of dark lanes against a background of approximately uniform brightness ( left ) .
the lanes consist of side - walls with thickness @xmath77 and depth @xmath78 ( right ) .
a fraction @xmath79 of the image is considered to be lane wall.,width=288 ] in this analysis we consider cross - correlation of an idealized small patch of quiet sun with @xmath80 pixels , of nearly uniform intrinsic brightness everywhere except in a fraction @xmath79 of the pixels of which are occupied by narrow , dark features with steep sides that are , on average , @xmath77 pixels across , and an absolute intensity deficit of @xmath78 ( figure . for typical quiet - sun visible light observations , @xmath79is about 0.1 and @xmath78is about @xmath81 .
assuming that the correlation algorithm does not miss ( it achieves a match between corresponding features in the opposite order images ) , we assume that errors in the resulting overall position of the patches is due to randomization errors from the displacements ( real and apparent ) of the dark features between the two images . in the worst non - pathological case ,
every intensity feature in the image patch is displaced by a different random amount , representing a strong high - spatial - frequency field . in that case , each feature s apparent location is the incoherent sum of two random variable samples : one due to photon counting statistics in the sloping sides of the feature , and one due to the random displacements of the features themselves .
the former depends on the steepness of feature sides , the overall brightness of the image , and the number of pixels in each feature wall .
the latter depends on the dispersion of the instrument and the variability of spectral offset across the image the uncertainty in location of each feature due to photon counting statistics is related to the slope at the sides of the feature : correlation fitting converts intensity in the sloping sides of a feature to position , so we can estimate the position / wavelength noise due to error in the intensity , by simply scaling the intensity noise to account for this conversion .
the cross - correlation requires comparison between two features , so one can write:@xmath82 where @xmath83 is the uncertainty in brightness in the side - of - feature pixels , and the right - hand expression accounts for the typical number of pixels in the sides of a particular feature .
the @xmath60 comes from the fraction @xmath84and is the number of photons counted during the exposure .
the unitless @xmath85 arises because the average brightness in a feature sidewall is halfway between the background intensity and the dark feature intensity , the extra factor of 2 under the radical comes from the fact that the noise in offset is the incoherent sum of the location noise in each of the two features being compared .
the ensemble average error position @xmath86 is then just the incoherent average of the offsets of each feature wall , or @xmath87 taking the patch to be about 15 pixels across , and assuming a scene of solar granulation with @xmath88 , @xmath89 , @xmath90 , and @xmath73 yields a noise estimate of @xmath91 pixels .
photon noise in the correlation step is hence negligible compared to photon noise in the differential step .
this is not surprising considering the large number of pixels that are averaged over by the correlation step . put another way , the noise spectrum is dominated by high spatial frequencies , which are `` smoothed out '' by the correlation process .
there is an additional source of error in the correlation step : because the wavelength signal @xmath15 includes high spatial frequencies ( the signal that is being measured by the differential stereoscopy step ) , different parts of the image must be shifted in different directions to find the average offset over the entire patch being correlated .
these high spatial frequencies introduce an error into the measurement of the shifts @xmath16 , because the correlation step weights the importance of the different offsets according to the average sidewall slope in each portion of the patch being correlated , rather than uniformly throughout the patch of image .
this error term is independent of the photon noise , and is introduced even for successful cross - correlations .
it depends on the inequality of the weighting of the features , as well as the amount of displacement:@xmath92 where @xmath93 is used to indicate actual differences from the patch - wide mean ( different from @xmath94which is being used to indicate statistical error in a quantity ) , @xmath95 is the signed difference between @xmath21 at each location in the patch being correlated and the mean for the whole patch , @xmath58 is the slope of the intensity function at each location in the image , and @xmath96 is the difference between the current value of the slope and its patch - wide average . as above ,
angle - brackets symbolize rms averaging over a complete data set ; here , an over bar is used to represent signed averaging over the patch .
the inclusion of both @xmath93-quantities ( on the rhs ) and @xmath6-quantities ( on the lhs ) is because eq .
[ eq : weighting - delta ] describes errors in the measurement of @xmath36 that are due to the interpretation of the images themselves even in the absence of injected noise , rather than to an injected noise source in the direct measurement of brightness in each image plane . the numerator in the rhs of eq .
[ eq : weighting - delta ] depends only on the scene being viewed .
the signed average simply applies an @xmath97 scaling of the overall error term , according to the number of independent samples @xmath98 of image data in the patch .
the quantity being averaged is the product of two random variables with zero mean and hence can be expanded simply in terms of their covariance @xmath12:@xmath99 where @xmath100 is the number of `` features ''
separately resolved regions that each represents an independent sample of image data in the patches being correlated .
assuming that the wavelength and brightness signals are uncorrelated eliminates the covariance term:@xmath101 so that , if wavelength and intensity of emissions be uncorrelated , the location uncertainty of a given cross - correlated patch is proportional to the rms variation in central wavelength and the rms variation in _ slope _ of the image intensity .
for example , taking the patch to be about 15 pixels across , @xmath102 ( one unique feature every four pixels ) , @xmath103= 0.1 pixel , @xmath104=0.2 , and @xmath105=0.1 yields @xmath106 pixel . in practice ,
the image intensity and wavelength variation are not completely uncorrelated .
for example , small magnetic features tend to be found in dark intergranular lanes , rather than randomly across the quiet sun , so that @xmath107 .
it is not clear , however , how or whether @xmath58 and @xmath21 are correlated . in the worst case , they are proportional and the covariance term is equal to the product term in eq.[eq : weighting - covariance ] , doubling the effective noise level from this source to ( in the example ) 0.012 pixel . in the best case , @xmath108 would be near zero everywhere except where @xmath58 is small . in that case , the covariance term would nearly exactly cancel the product term in eq .
[ eq : weighting - covariance ] and eliminate this noise source . in the context of a photospheric magnetograph
, some further information about the covariance term may be used . in the quiet sun for example , the strongest magnetic fields are observed to exist in the intergranular lanes , near local minima in continuum intensity .
this suggests that the covariance between @xmath58 and @xmath109 is in fact negative in real scenes of solar granulation , so that eq .
[ eq : no - covariance ] likely overestimates the strength of this noise source in quiet sun magnetic measurements . in scenes that contain more contrast , particularly magnetic - correlated contrast
, image correlation problems can dominate the zeeman shift signal .
in particular , near sunspots very strong variations in intensity are strongly correlated to very strong variations in magnetic field , reducing the effectiveness of this technique for inverting stereoscopic data of sunspot fields .
this was observed by deforest et al .
( 2004 ) , who found that correlation introduced strong artifacts into a not - well - resolved sunspot penumbra .
equation [ eq : hybrid_inversion ] includes several important engineered cancellations .
failure of those cancellations causes leakage of intensity or other information into the derived wavelength signal .
the most important , and most likely to fail , cancellation is the elimination of an intensity spatial - derivative signal from the @xmath33 introduced in eq .
[ eq : j - definition ] . in the narrow - line approximation described by deforest et al .
( 2004 ) , the single - image brightness is given by ( their equation 7):@xmath110(x/\alpha_{n})}{\alpha_{n}}-l(x)+\alpha_{n}l(x)\frac{d\lambda'}{dx}+\alpha_{n}\lambda'(x)\frac{dl}{dx}-2\alpha_{n}^{2}\lambda'(x)\frac{dl}{dx}\frac{d\lambda'}{dx}\label{eq : single - intensity}\ ] ] ( where , again , @xmath111is the continuum brightness per unit wavelength , @xmath112 is the prefilter function , @xmath113 is the total line depth intensity , and @xmath114 is the dispersion of the @xmath115 spectral order ) .
the differential inversion arises from noticing that the difference between two opposite orders contains a useful @xmath22 term : @xmath116 in which the @xmath22 term is desired and the @xmath49 term is undesired . in hybrid stereoscopy the images are resampled into @xmath117 coordinates with the substitution : @xmath118 that eliminates the undesired @xmath49 term in eq .
[ eq : intensity - equation ] to first order .
this process inevitably introduces some other undesired terms .
in particular , substituting eq .
[ eq : x - def ] into eq .
[ eq : single - intensity ] yields ( writing @xmath119 for the convolution term ) : @xmath120 where @xmath121 is the convolution term from eq .
[ eq : single - intensity ] . as before
, subtracting opposite orders removes all terms that contain an even power of @xmath18 , and doubles the remaining terms .
grouping all terms and discarding terms beyond first order in @xmath21 yields:@xmath122 where the first term is desired and the other terms are noise .
the @xmath123 term is normally small both because the filter function is symmetric in wavelength and because the convolution integral smooths across several pixels .
the @xmath23 term represents leakage of the @xmath49 signal into the data .
( recall , @xmath23 is just the high spatial frequency component of the wavelength shift , given by @xmath124 ) .
the low - frequency image contribution to the noise has been reduced by the square of the ( normally small ) factor @xmath125 .
the lhs of equation [ eq : not - so - evil ] is merely the numerator of the @xmath33 definition in equation [ eq : j - definition ] .
the denominator is the sum of the two intensities , which retains the even powers of @xmath114 while canceling the odd powers .
again , many cancellations occur , leaving:@xmath126 so that the summed brightness of the resampled images has a systematic noise term that is proportional to the second derivative of the line depth . combining the terms yields a more complete expression for @xmath33:@xmath127 where first order approximations to the fraction have been taken , and squares of perturbations have been ignored . the first term of eq .
[ eq : evil - j ] reproduces the desired signal ( eq .
[ eq : dlambda / dx ] ) and the additional terms represent noise sources due to the image itself . writing them as an error term @xmath128,@xmath129 where the first terms represent leakage of intensity information into the wavelength signal ( both from continuum variation and from line variation ) , and the final term represents leakage of rapid brightness variations into the j gain and is negligible in nearly all cases .
the continuum intensity leakage term is dependent on large - scale variations in continuum brightness , because @xmath130is smoothed ( by convolution with the filter function ) over a distance of @xmath131 , where @xmath132 is the width of the entrance filter .
in typical applications , @xmath131 might be 5 - 15 pixels .
a scene of solar granulation with @xmath133~10 , @xmath134~0.1 pixel and @xmath131~10 pixels , with intergranular lanes 3 pixels wide and continuum dips of 30% in the lanes , will have a continuum intensity leakage error of ~@xmath135 in j / i in the regions on either side of the lane , equivalent to the noise from @xmath136 photons .
the error accumulates coherently over half of the dispersed filter width , or about 5 pixels , yielding a peak injected noise signal in @xmath137 of ~0.025 pixel in the lane center from just the continuum term .
the continuum intensity leakage is much stronger in regions with strong large scale variations in intensity , coupled with strong line shifts ( such as found in sunspots ) . in a sunspot scene with @xmath133~10 ,
@xmath138 pixels , and a continuum dip of 75% within the sunspot , pixels near the sunspot edge will have a continuum leakage term of 0.2 in j / i , much larger than any photon noise in the system .
the error accumulates coherently for the full dispersed width of the entrance filter , yielding integrated errors in @xmath134 as high as 20 pixels near the edge of the sunspot .
this large error injection is the primary reason why deforest et al .
( 2004 ) found sunspot measurements to be challenging with this inversion technique .
the line intensity leakage term is largest in regions of very rapid variation of the central wavelength , but is negligible compared to the continuum leakage in most scenes . considering a `` challenging '' scene of solar plage with @xmath134~ 1 pixel , @xmath139 amplitude , @xmath140~ 0.1 , @xmath141 , and @xmath142 yields intensity leakage of about @xmath143 into j / i from the central term .
this is equivalent to the photon noise from @xmath144photons , and if sustained across a full correlation patch of width 15 pixels could inject an error signal of up to ~0.1 pixel into the final inversion ; this represents about 20% of the background field . with a more typical quiet - sun case of @xmath145pixel ( 30 g ) , @xmath146pixel , and @xmath147 ,
the intensity leakage noise is under @xmath148 in j / i and thus leads to errors of under 0.015 pixel after accumulating over a 15-pixel fwhm patch .
the rapid brightness variation leakage depends on rapid changes in the line depth , in regions that are magnetized .
it is negligible in nearly all cases other than sunspot penumbrae . in the quiet sun
, the strongest value of the term occurs where @xmath149-band bright points are present in intergranular lanes . in a typical g - band bright point , with @xmath150 , @xmath151and @xmath152 , @xmath33 can be expected to have a relative error of only @xmath153 from this term . in sunspot penumbrae , where both @xmath134 and @xmath137 may be larger than order unity , and where @xmath154 may be a factor of 4 - 10 higher than in intergranular lanes due to the fine nature of penumbral roll structure , @xmath33 can be expected to have a relative error of order close to unity from this term , severely degrading results in sunspot penumbrae even if the continuum leakage signal ( above ) were not a problem . in all cases ,
the intensity gradient leakage terms are proportional to @xmath155 and therefore represent a gain error rather than an uncorrelated additive noise source that affects detection of lone magnetic features .
filter passband gradients enter because the overall measured intensity is affected by the throughput function of the prefilter . in conventional filtergraph instruments ,
the slope of the side - lobes of the main filter passband is used to convert wavelength offset into an intensity signal ; in stereographic instruments , that signal is undesired and represents a source of noise . even in a properly tuned stereoscopic instrument , large excursions of @xmath21
carry the spectral line out of the flat central region of the prefilter s passband and into the sloping wings .
this modifies eq .
[ eq : single - intensity ] , to include the effect of the filter on the measured l(x):@xmath156(x/\alpha_{n})}{\alpha_{n}}-l(x)f(\lambda')+\alpha_{n}l(x)f(\lambda')\frac{d\lambda'}{dx}\\ + \alpha_{n}\lambda'(x)\left(f(\lambda')\frac{dl}{dx}+l(x)\frac{df}{d\lambda'}\frac{d\lambda'}{dx}\right)\\ -2\alpha_{n}^{2}\lambda'(x)\left(f(\lambda')\frac{dl}{dx}\frac{d\lambda'}{dx}+l(x)\frac{df}{d\lambda'}\left(\frac{d\lambda'}{dx}\right)^{2}\right)\end{array}\right\ } \label{eq : asymmetric - filter}\end{aligned}\ ] ] where @xmath157 is taken to be unity . of the two new terms , only one is asymmetric , so the difference equation becomes:@xmath158 where both the overall factor of f and the additional unitless derivative @xmath159 represent perturbations on the gain of the measured @xmath22 , and the final term is the intensity leakage that is removed by the hybridization step in the overall inversion . in a well - tuned instrument ,
the maximum value of @xmath112 occurs near @xmath160 , so we may expand f and f around that value : @xmath161 where @xmath162 is the difference between the center wavelength of the filter and the rest wavelength of the spectral line , and @xmath163 is the half - width at half maximum of the filter passband
. then the derivative @xmath164 is given by @xmath165 .
as an example , if @xmath162 is about -w/10 and @xmath21is about w/4 , this yields a gain error of 7% in the @xmath22 signal .
rapid variation of the central wavelength of the observed spectral line affects the measured signal from differential stereoscopy .
the differential inversion ( eq . [ eq : discrete inversion ] ) is derived using a first - order expansion of the brightness in terms of the slope @xmath22 of the central wavelength versus position .
if the condition @xmath166 does not hold ( the slope is significant compared to the reciprocal of the instrument s dispersion ) then the differential inversion equation is not valid and the derived @xmath31 will differ from the true @xmath21 even in the absence of any other noise source .
this is due to the effect being exploited to extract the @xmath22 signal from the intensities : small shifts in the @xmath3 direction project into the dispersed spatial direction , so that the line of integration in the @xmath167 plane that contributes to a single pixel may have more or less contribution from the line core depending on the relative angle between @xmath114 and @xmath22 .
the resultant variation in intensity is extremely nonlinear when @xmath114 and @xmath22 are nearly parallel .
further , the @xmath22 intensity variations are due to small shifts in image position as@xmath21 varies .
if @xmath21 grows larger than 1 pix/@xmath114 , then the image shift becomes noticeable and distorts the resultant line - shift images .
the various effects of large excursions in @xmath21 on spatial distributions are illustrated in figure [ fig : multi ] , which illustrates the geometry of the @xmath167 plane in a stereoscopic measurement of line center for a spectral line with structured doppler signal .
three forms of perturbation on the line are shown : a sinusoidal modulation of the line center wavelength ; a triangle wave modulation ; and a dual step function .
the sinusoidal modulation fits the criteria outlined by deforest et al .
( 2004 ) for differential stereoscopic inversion : @xmath168 and @xmath169 is not large compared to the spatial features of interest .
the others fail in different ways .
the triangle wave fails because @xmath170 , and the offset fails because @xmath169is large compared to the width of the transition features at the edges of the step in central wavelength .
hybrid differential stereoscopy reduces the effects by eliminating the low spatial frequency components , so that the amplitude and slope of the high spatial frequency component @xmath171 rather than of @xmath21 itself , are of import : structures larger than roughly the width of the patch used for correlation are attenuated or eliminated by the high - pass filtering that generates @xmath23 . by considering carefully the geometry of the differential inversion in the @xmath167 plane it is possible to estimate the systematic errors due to slope and line - displacement effects alone on differential stereoscopy , even in the absence of image effects such as variations in line or continuum intensity .
figure [ fig : step ] shows the geometry of dispersed integration through a spectral line .
the contribution @xmath172 to the measured intensity @xmath173 is approximated to first order in eq .
[ eq : dlambda / dx ] , but the contribution may also be written exactly in the case of constant @xmath113 , constant @xmath22 , and a broad filter:@xmath174 applying this expression to eq .
[ eq : j - definition ] gives an analytic formula for the relation between j and @xmath22 . letting @xmath175 and @xmath176 , @xmath177 where the factor @xmath178 is a linear calibration coefficient on @xmath22 and
the right - hand term is a nonlinearity term .
the nonlinearity in @xmath179 becomes strongly evident when @xmath180 grows to about @xmath181 , where ( with @xmath182 ) the nonlinearity term has a value of 1.38 , and grows to infinity where @xmath183 .
the strong nonlinearity in @xmath33 is partially compensated by spatial distortion .
the nonlinear term causes growth in @xmath33 compared to the proportionality with @xmath22 , but that growth is accompanied by spatial distortions that partially cancel the nonlinearity . in particular , the measurable quantity of interest is @xmath21 rather than @xmath22 . in a pure differential inversion , with @xmath180 positive and non - negligible
, the @xmath179 signal arises from a slight weakening of the spectral line in the @xmath184 spectral order and a strong strengthening in the @xmath185 order .
however , the image in the @xmath185 channel is strongly foreshortened while the image in the @xmath184 channel is weakly fore - lengthened : the total @xmath4-integrated intensity in each spectral order remains the same under small to moderate ( < 1 pixel ) @xmath21 perturbations in the image . consider a finite shift in central wavelength @xmath21 as illustrated in figure [ fig : step ] .
accompanying the shift in brightness is a spatial distortion that affects the two spectral orders differently . in the positive order
the line depth is more strongly expressed , because it is concentrated into fewer pixels than might be expected from the spatial extent of the feature . likewise , in the negative order the line depth is more weakly expressed , because it is spread across more pixels than the usual spatial extent of the feature .
the corresponding variation of @xmath179 thus has three regimes in this simple system ( neglecting the line width):@xmath186 where ( from eq .
[ eq : analytic ] ) @xmath187 , @xmath188 , and @xmath189 .
simple analysis shows that@xmath190 and @xmath191 so that the measured wavelength shift may be computed directly : @xmath192 where the last row expands the @xmath180-dependent error factor to first order in @xmath180 .
equation [ eq : deltalambda ] shows that the linear approximations used to derive [ eq : hybrid_inversion ] are quite good , at least for this simple geometry : the first nontrivial term in the error factor is two orders of approximation smaller than the extracted signal ( taking both @xmath180 and @xmath193 to be small , as did deforest et al .
2004 ) . the high quality of the approximation can be seen in figure [ fig : multi ] : the measured peak and feature - averaged @xmath194 in each feature remains close to the injected @xmath195 of the model , despite gross violation of the assumptions used to derive equation [ eq : hybrid_inversion ] .
the astute reader will have noticed that equation [ eq : deltalambda - exact ] has a pole at @xmath196 .
that is because we have neglected finite length effects in the region where the line shift occurs .
equation [ eq : deltalambda - exact ] is more complex but exact for this geometry , provided that the line width is small compared to @xmath197 .
when that condition is violated ( which happens when the center locus is exceptionally narrow on the detector ) , individual pixel lines of integration do not integrate across the entire foreshortened spectral line but are limited instead by the ends of the spectral feature this limitation keeps the signal response @xmath198 finite . equation [ eq : deltalambda - exact ] neglects this effect , leading to a spurious pole . for nonzero line widths , equation
[ eq : deltalambda - exact ] reveals a constant correction factor of @xmath199 that must be applied to stereoscopic inversions in general .
this correction ( of order 10% ) is a true calibration coefficient , in the sense that it is independent of @xmath180 .
five plots showing arbitrary variations of spectral line central wavelength show the manner in which the approximation of equation [ eq : discrete inversion ] breaks down at high line slopes . in order : the x - lambda plane ; the x - lambda plane , dispersed in two opposite directions ( pixel brightnesses are formed by vertical integration in these two plots ) ; the original and derived spatial derivative of line center wavelength ; and an overlay of the original and reconstructed signals . the sinusoidal segment at left is well matched .
the pathological sawtooth and square wave segments are distorted by the coupling between wavelength and spatial position.,width=384 ] a finite linear feature in @xmath7 demonstrates the interplay between projection , brightness , and feature size in a stereoscopic instrument . at top , a spectral line undergoes a finite shift @xmath200 between the points @xmath201 and @xmath202 in the ( @xmath203 ) plane .
the instrument collects two spectral orders with dispersion @xmath204 .
each order integrates pixels along lines of the appropriate slope : positive ( cyan ) or negative ( red ) and projects onto the detector @xmath4 coordinate . at bottom
the line depth brightness integral @xmath205 varies as a function of pixel location ( projected back to the @xmath206 line at top ) .
, width=288 ]
the effects in [ sec : hybrid - differential - stereoscopy ] above separate into two primary classes : additive and multiplicative noise in the data .
the additive effects , such as photon noise , affect sensitivity for detection of small magnetic or doppler features ; the multiplicative effects , affect the precision of magnetic measurements . here
we calculate a noise budget for a plausible observing scenario using a stereoscopic instrument as a photospheric magnetograph .
consider an instrument observing quiet sun through the dunn solar telescope ( dst ; @xcite ) and the medium - order adaptive optics system ( @xcite ) , at the fe i 617.34 nm line , with 0.07 arcsec pixels ( the diffraction limit is 0.2 arcsec ) , a 67 pm prefilter passband width detuned from @xmath7 by 6.7 pm ( 10% of the bandwidth ) , @xmath207 , dispersion @xmath208 pixel nm@xmath209 , @xmath73 photons per pixel , and a correlation patch fwhm of 15 pixels ( @xmath71 ) .
consider that a `` typical '' resolved feature strength is between 100g-1 kg , with length scales of a few pixels ( so that only stereoscopic measurements apply , without the stabilizing influence of the more precise correlation stereoscopy ) .
the 617.34 nm line has a land @xmath149 factor of 2.5 , so that the zeeman splitting equation is just @xmath210 where @xmath211 * * * * is measured resolution - smoothed magnetic field strength ( `` flux density '' ) , @xmath212 is planck s constant , @xmath213 is the speed of light , @xmath149 is the land factor , @xmath214 is the bohr magneton , and @xmath215 is the _ splitting _ of the line ( twice the shift ) .
the splitting measurement is the result of _ two _ independent measurements of a shift @xmath195 : one in right - circularly polarized light ( stokes i+v ) and one in left - circularly polarized light ( stokes i - v ) .
hence the noise @xmath216 is a factor of @xmath217 higher than might be supposed by simple scaling of the @xmath195 noise terms in [ sec : hybrid - differential - stereoscopy ] with equation [ eq : splitting ] .
the noise budget for a quiet - sun dst observation may be summarized as in -table [ tab : additive - budget ] .
two principal types of error enter into the measurement : random or quasi - random noise as described in [ sub : photon - counting - noise]-[sub : image - brightness - gradients ] ; and calibration nonlinearities that enter as a result of filter detuning and the first - order stereoscopic inversion as described in [ sub : filter - passband - gradients]-[sub : rapid - variation - of ] .
.[tab : additive - budget]noise budget for a sample stereoscopic magnetograph measurement as described in the text .
this measurement would have ~ 6 g rms noise even in field - free regions of the sun , and 3%-12% calibration errors in the measured flux of small magnetized regions of 100 - 1000 g , no more than a few pixels across . [ cols="^,^,^,^,^",options="header " , ] because of the multiple sources of error , calculation of error budgets for actual observations must be carried out on a case - by - case basis ; however , we note that for magnetograph observations of the photosphere , the dominant noise term is ( as expected ) photon noise for a plausible instrument setup and a range of targets that includes most non - sunspot - related targets on the sun .
the expected formal single - exposure magnetogram sensitivity is ~10 gauss ( twice the rms background noise level ) and the expected magnetic flux calibration accuracy is expected to be close to 10% in most cases including plage .
the error budget in table [ tab : additive - budget ] is broadly consistent with test measurements collected by deforest et al .
( 2004 ) at the dunn solar telescope . in those measurements , exposures of only @xmath218 photons were collected ; 16 exposures were combined to create each image for an effective exposure of @xmath219 photons ; and an _ asymmetric _ inversion ( whose sensitivity was reduced by a factor of @xmath217 compared to the symmetric inversion discussed here ) was used .
the resulting calculated photon noise level is thus a factor of @xmath220 higher than shown here , or 9.5 g ( 0.45 g for the correlation noise ) , and the non - photon noise sources are scaled up a factor of @xmath217 compared to the sample measurement described here .
this leads to a total _ a priori _ noise level estimate of 10 g for that measurement .
indeed , in a small quiet region with no detected flux features , deforest et al . found _ a posteriori _ rms noise levels between 9 - 12 gauss , in good agreement with the current _ a priori _ calculation .
we have calculated a detailed noise budget for hybrid differential inversion of the line - shift signal from a stereoscopic spectrograph , with specific application as a stereoscopic magnetograph observing a specific absorption line ( fe i 617.34 nm ) in the solar photosphere .
the noise budget is photon dominated and careful _ a priori _ analysis shows that it offers similar calibration performance to existing and past filtergraph ( soho / mdi , @xcite ) and fourier tachometer ( gong ; @xcite ) instruments ( @xcite ) : instrument response to field is within a few to ten percent of linear over a wide range of pixel - averaged field strengths .
the principal difference between the stereoscopic technique and the filtergraph or fourier tachometer techniques is that the instrument can operate much more rapidly , because all required photons are collected simultaneously in a dual - polarized - beam instrument .
this is very important for ground based observations in which atmospheric seeing effects severely limit the precision with which line shifts can be measured in time - multiplexing instruments such as filtergraphs .
some instruments , such as zimpol ( @xcite ) , have been able to limit crosstalk of atmospheric effects by multiplexing extremely rapidly ; but such instruments are still subject to motion blur even in the absence of severe crosstalk .
a stereoscopic spectral imager should be able to acquire a diffraction limited magnetogram with few - gauss sensitivity in a small fraction of a second at most major observing facilities .
we expect this rapidity of acquisition to become important even in space - based observing as higher resolution instruments ( e.g. _ hinode / sot _ ; @xcite ) release data to the scientific community : the race between solar evolution and data acquisition ( photon counting ) give rise to a tradeoff between sensitivity and spatial resolution for any spectral imager , even above the atmosphere . particularly for features that are spatially small , stereoscopic imagers offer a surprising advantage : because the fundamental signal extracted from the images is proportional to the _ spatial derivative _ of the line shift signal , such instruments are particularly sensitive to small features , right down to the spatial resolution of the measurement . like dolby@xmath221 noise reduction for audio recordings ( e.g. @xcite )
, spatial derivative measurement enhances high spatial frequencies , boosting the high frequency signal relative to the photon noise floor .
conventional spectral measurements sample each location in the image plane independently , so that in a roughly circular feature of linear size @xmath19 and area @xmath222 the uncertainty due to photon counting in the feature s total magnetic flux scales as @xmath223 or @xmath224 . in a differential stereoscopic measurement , the _ average slope _ uncertainty , rather than the total flux uncertainty , scales as @xmath224
so the total flux uncertainty is independent of feature size for features smaller than the correlation patches described in section [ sec : summary-&-specific ] .
stereoscopic spectral imaging is a special case of the more general technique of _ spectral tomography _ ( e.g. @xcite ) , and appears to offer significant advantages in the regime we have described : observations of a narrow spectral line , where only one or two moments of the line are desired to be measured .
spectral tomography using regularized inverse or conjugate analysis techniques ( e.g. @xcite ) could potentially be used to extract much more information about the spectrum , using multiple orders or , as did wilson et al . , crossed gratings ; but as the number of additional channels grows , so do the potential paths for signal contamination and crosstalk between different aspects of the measured spectrum .
we conclude that stereoscopic techniques represent a sweet spot for imaged doppler or magnetic measurements of solar scenes , offering superior photon efficiency and comparable performance to filtergraph measurements .
this is already important as ground - based spectral imagers are already photon - starved near the diffraction limit in existing meter - class telescopes ( such as the dunn solar telescope , @xcite , and the swedish vacuum tower , @xcite ) and will grow more so near the diffraction limit of future telescopes such as the four - meter advanced technology solar telescope ( @xcite ) .
the authors thank jack harvey for helpful discussion of the current state of magnetograph cross - calibration , and steve tomczyk for pointing out the neceessity of this paper .
this work was funded internally by the southwest research institute .
j. l. fox , c. c. kankelborg , and t. r. metcalf . .
in _ optical spectroscopic techniques and instrumentation for atmospheric and space research v. edited by larar , allen m. ; shaw , joseph a. ; sun , zhaobo .
proceedings of the spie , volume 5157 , pp .
124 - 132 ( 2003 ) .
_ , 2003 .
k. ichimoto , y. suematsu , t. shimizu , y. katsukawa , s. tsuneta , t. d. tarbell , r. a. shine , c. m. hoffmann , a. m. title , b. w. lites , d. f. elmore , and k. v. streander . .
in _ astronomical society of the pacific conference series _ , 2006 .
t. r. rimmele . .
in d. bonaccini calia , b. l. ellerbroek , and r. ragazzoni , editors , _ advancements in adaptive optics . edited by domenico b. calia , brent l. ellerbroek , and roberto ragazzoni .
proceedings of the spie , volume 5490 , pp .
34 - 46 ( 2004 ) .
_ , page 34 , october 2004 .
g. b. scharmer , k. bjelksjo , t. k. korhonen , b. lindberg , and b. petterson . .
in _ innovative telescopes and instrumentation for solar astrophysics . edited by stephen l. keil , sergey v. avakyan .
proceedings of the spie , volume 4853 , pp .
341 - 350 ( 2003 ) .
_ , 2003 .
p. h. scherrer , r. s. bogart , r. i. bush , j. t. hoeksema , a. g. kosovichev , j. schou , w. rosenberg , l. springer , t. d. tarbell , a. title , c. j. wolfson , i. zayer , and mdi engineering team .
_ , 162:0 129188 , 1995 .
|
stereoscopic spectral imaging is an observing technique that affords rapid acquisition of limited spectral information over an entire image plane simultaneously .
light from a telescope is dispersed into multiple spectral orders , which are imaged separately , and two or more of the dispersed images are combined using an analogy between the @xmath0 spectral data space and conventional @xmath1 three - space . because no photons are deliberately destroyed during image acquisition ,
the technique is much more photon - efficient in some observing regimes than existing techniques such as scanned - filtergraph or scanned - slit spectral imaging .
hybrid differential stereoscopy , which uses a combination of conventional cross - correlation stereoscopy and linear approximation theory to extract the central wavelength of a spectral line , has been used to produce solar stokes - v ( line - of - sight ) magnetograms in the 617.34 nm fe i line , and more sophisticated inversion techniques are currently being used to derive doppler and line separation data from euv images of the solar corona collected in the neighboring lines of he - ii and si - xi at 30.4 nm . in this paper
we develop an analytic _ a priori _ treatment of noise in the line shift signal derived from hybrid differential stereoscopy .
we use the analysis to estimate the noise level and measurement precision in a high resolution solar magnetograph based on stereoscopic spectral imaging , compare those estimates to a test observation made in 2003 , and discuss implications for future instruments .
|
bl lacertae objects are the most extreme class of active galactic nuclei ( agn ) , exhibiting strong , rapidly variable polarization and continuum emission , and core - dominated radio emission with apparent superluminal motion ( see e.g. kollgaard et al .
1992 for references ) .
these properties have led to the commonly accepted view that bl lacs are dominated by doppler - boosted synchrotron emission from a relativistic jet nearly along our line - of - sight ( blandford & rees 1978 ) .
the line emission of bl lacs is absent or weak , making their redshift determination rather difficult . in the current unified models of radio - loud agn ( e.g. urry & padovani 1995 ) ,
bl lacs are identified as low luminosity , core - dominated f - r i ( fanaroff & riley 1974 ) radio galaxies ( rg ) viewed nearly along the axis of the relativistically boosted jet .
this model appears supported by the comparison of their extended radio luminosity and morphology ( e.g. perlman & stocke 1993 ) , host galaxy luminosity and morphology ( e.g. wurtz et al .
1996 ; hereafter wsy96 ) and space density and beamed luminosity functions ( e.g. padovani & urry 1990 ; morris et al . 1991 ; celotti et al . 1993 ) .
knowledge of the properties of the host galaxies and environments where agn live is fundamental for the understanding of the formation of agn in galaxies .
comparison of orientation - independent properties of bl lacs , such as the host galaxies and environment , with those of rgs , allow one to test the unified model based on orientation ( antonucci 1993 ; urry & padovani 1995 ) . the frequency of close companions will determine whether interactions are important for triggering of the bl lac activity , as seems to be the case for quasars ( e.g. heckman 1990 ; hutchings & neff 1992 ) .
possible cosmological evolution in the properties of the hosts and environments can be studied by comparing agn properties at different redshifts .
finally , the alternative gravitational lensing hypothesis for bl lac activity ( ostriker & vietri 1990 ) can be tested by measuring the predicted offsets between the bl lacs and their host galaxies .
recent studies of bl lac hosts and close environments from the ground and with the hst ( see pesce et al 1995 ; falomo 1996 ; owen et al . 1996 ; wsy96 ; wurtz et al .
1997 ; falomo et al .
1997 ; scarpa et al 1999a ; urry et al . 1999a ; urry et al . 1999b , hereafter u99b ; heidt et al .
1999 ) have shown that their host galaxies are luminous ( m@xmath5 23 to 24 ) and large ( r(e ) = 10@xmath37 kpc ) elliptical galaxies .
they are on average @xmath61 mag brighter than l@xmath7 galaxies ( e.g. mobasher et al .
1993 ) and of similar luminosity or slightly fainter than the brightest cluster galaxies ( bcg ; e.g. hoessel et al .
intriguingly , bl lac hosts appear slightly fainter than f - r i hosts and resemble better typical f - r ii hosts .
detections or claims of disk hosts and/or inner disks components have been made in a handful of sources ( e.g. ms 0205.7 + 3509 stocke wurtz and perlman 1995 ; oq 530 abraham et al 1991 ; pks 1413 + 135 , mchardy et al .
1991 ; pks 0548 - 322 , falomo et al .
1995 ; 1es 1959 + 650 , heidt et al . 1999 ; ) . while it is not unreasonable to find small inner disk structures similarly to what found in normal ellipticals , the objects with a disk dominated host galaxy are rather controversial ( for details
see : falomo et al 1997 ; stocke et al 1992 ; lamer , newsam , and mchardy 1999 ; scarpa et al 1999a , b ) .
close companions have been found around many bl lacs , some with signs of interaction , but the physical association has been confirmed only in some cases through spectroscopic measurements . bl lacs usually reside in poor clusters , unlike f - r i rgs .
a modified unified model has therefore been proposed in which bl lacs either include partly f - r iis in addition to f - r is , or bl lacs avoid the brightest f - r is and the f - r is in rich clusters at low redshift ( wsy96 ) .
this scenario seems also supported by recent measurements of radio polarization of bl lacs ( stanghellini et al 1997 ) that show signature of a f - r ii population .
similarly to the case of quasars ( e.g. hutchings et al . 1994 ; rnnback et al . 1996
; bahcall et al .
1997 ) , the study of bl lac host galaxies is rather problematic because of the presence of the bright nucleus that swamps the light of the host galaxy , although the nucleus / host luminosity ratio of the bl lacs is smaller than that of quasars .
the use of hst images can substantially improve the ability to study the host close to the nucleus , however , as we shall show in section 4.2.2 , hst data are usually not deep enough to properly investigate the external fainter regions of the host galaxies ( see also e.g. hutchings et al . 1994 ; bahcall et al 1997 ) .
in this paper we report on a large , homogeneous data set of observations of bl lacs secured mainly with sub - arcsec resolution from the nordic optical telescope ( not ) in la palma .
we have observed the complete sample of 26 x - ray selected ( or high frequency peaked , hbl ) bl lacs derived from the _ einstein _ medium sensitivity survey ( emss ; stocke et al .
this sample is almost complete for multiwavelength information and redshift ( although some redshifts are tentative and probably wrong , see the appendix ) and most of the objects are at [email protected] , ensuring the detection of the host in most cases with reasonable observing time .
additionally , we obtained images for 26 bl lacs in the _ einstein _ slew survey ( perlman et al 1996 ) .
these bl lacs are also of hbl type and are compared with the emss targets .
based on the current data , we have previously published a separate study of the peculiar bl lac object ms 0205.7 + 3509 ( falomo et al .
1997 ) .
the structure of this paper is as follows : in section 2 we describe the samples of the objects observed . in section 3
we give details of the observations , data reduction and describe the image analysis .
results and discussion on the overall properties of the host galaxies , including comparison with previous studies of bl lacs and other types of agn are given in section 4 .
the main conclusions from this study are summarized in section 5 . in the appendix
, we give comments for individual objects . throughout this paper ,
are adopted .
the emss survey ( gioia et al . 1990 ; morris et al 1991 ; stocke et al . 1991 ; maccacaro et al .
1994 ) is a flux
limited complete sample of faint x
ray sources discovered serendipitously in numerous _
imaging proportional counter ( ipc ) fields centered on high galactic latitude ( b @xmath1 20 @xmath8 ) targets .
it covers 780 @xmath9 in the 0.3 3.5 kev soft x ray band with limiting sensitivity ranging from @xmath65 @xmath10 10@xmath11 to @xmath63 @xmath10 10@xmath12 erg @xmath13 s@xmath14 .
the emss includes 34 bl lac objects and 4 bl lac candidates ( as listed in padovani & giommi 1995 ) , selected to have the observed equivalent width of any emission line ew @xmath45 @xmath15 .
moreover , if a ca ii h+k break is present due to starlight in the bl lac host galaxy , its contrast must be @xmath16 25 @xmath17 , much less than for a typical giant non - active elliptical galaxy ( @xmath650 % ) .
the ipc slew survey ( perlman et al . 1996 ) was constructed using the _ einstein _ slew data taken when the satellite was moving from one target to the next ( elvis et al .
1992 ) and covers a large fraction of the sky with limiting sensitivity ranging from @xmath65 @xmath10 10@xmath12 to @xmath161 @xmath10 10@xmath12 erg @xmath13 s@xmath14 in the 0.3 3.5 kev soft x ray band .
padovani & giommi ( 1995 ) list a total of 60 slew bl lacs and 9 bl lac candidates extracted from the slew survey adopting the same classification criteria as for the emss survey .
the observed bl lacs were selected from the emss and slew samples to have declination @xmath1815 @xmath8 , to be observable at the not .
this limit excluded five bl lacs and three bl lac candidates of the emss sample .
of the remaining 29 bl lacs and one bl lac candidate , 28 ( 93 % ) were observed at the not , the only exceptions being ms 1019.0 + 5139 and ms 1207.9 + 3945 . in the slew sample , seven bl lacs and two bl lac candidates do not satisfy our declination limit .
furthermore , of the remaining 53 bl lacs and 7 candidates , 14 are of low frequency peak ( lbl ) type while 4 belong also to the emss sample . of the final sample of 35 slew bl lacs and 7 slew bl lac candidates , 26 ( 62 % ) were observed at the not .
the selection of the slew objects observed was based only on observability conditions .
general properties of the observed bl lacs are given in table 1 , columns ( 1)-(7 ) , where column ( 1 ) gives the name of the object , column ( 2 ) the redshift , column ( 3 ) the apparent @xmath19band magnitude , columns ( 4 ) and ( 5 ) the 5 ghz and 2 kev flux densities , respectively , and columns ( 6 ) and ( 7 ) the optical x - ray and radio optical spectral indices , respectively .
llllllllllll + name & z & v & s(5 ghz ) & s(2 kev ) & @xmath20(o x ) & @xmath20(r o ) & date & t(int ) & sky & seeing & a(r ) + & & & ( mjy ) & ( @xmath21jy ) & & & & ( sec ) & ( mag ) & ( @xmath22 ) & ( mag ) + 1es 0033 + 595 & ... & 19.5 & 66.0 & 2.22 & 0.45 & 0.61 & 21/09/98 & 600 & 21.0 & 0.64 & 1.94 + 1es 0120 + 340 & 0.272 & 15.2 & 33.6 & 1.86 & 1.06 & 0.28 & 23/12/95 & 1200 & 19.72 & 0.51 & 0.24 + ms 0122.1 + 0903 & 0.339 & 20.0 & 1.4 & 0.15 & 0.91 & 0.34 & 22/09/98 & 1800 & 21.01 & 0.58 & 0.20 + ms 0158.5 + 0019 & 0.229 : & 18.0 & 11.3 & 0.67 & 0.88 & 0.36 & 22/09/98 & 3600 & 20.9 & 0.63 & 0.12 + ms 0205.7 + 3509 & 0.318 : & 19.2 & 3.6 & 0.11 & 1.08 & 0.36 & 23/12/95 & 2400 & 19.97 & 0.68 & 0.29 + 1es 0229 + 200 & 0.139 & 14.7 & 49.1 & 1.13 & 1.22 & 0.27 & 22/09/98 & 2400 & 21.01 & 0.60 & 0.41 + ms 0257.9 + 3429 & 0.247 & 18.5 & 10.0 & 0.25 & 1.02 & 0.40 & 23/12/95 & 1800 & 20.95 & 0.53 & 0.41 + ms 0317.0 + 1834 & 0.190 & 18.1 & 17.0 & 2.56 & 0.71 & 0.41 & 24/12/95 & 1800 & 20.88 & 0.62 & 0.46 + 1es 0347 - 121 & 0.188 & 18.2 & 9.0 & 2.32 & 0.64 & 0.36 & 23/09/98 & 1800 & 21.01 & 1.12 & 0.17 + 1es 0414 + 009 & 0.287 & 17.5 & 70.0 & 3.64 & 0.67 & 0.47 & 24/02/98 & 2400 & 20.2 & 0.87 & 0.41 + ms 0419.3 + 1943 & 0.512 : & 20.3 & 8.0 & 0.53 & 0.62 & 0.50 & 24/12/95 & 3600 & 19.9 & 0.73 & 0.84 + 1es 0446 + 449 & 0.203 : & 18.5 & 139.4 & 0.63 & 0.81 & 0.60 & 24/12/95 & 1800 & 20.79 & 0.59 & 2.52 + 1es 0502 + 675 & 0.341 & 17.0 & 32.7 & 1.36 & 0.92 & 0.37 & 21/09/98 & 3600 & 21.01 & 0.78 & 0.53 + 1es 0525 + 713 & 0.249 : & 19.0 & 9.0 & 0.80 & 0.69 & 0.42 & 21/09/98 & 1200 & 20.9 & 0.68 & 0.41 + ms 0607.9 + 7108 & 0.267 & 19.6 & 18.2 & 0.27 & 0.86 & 0.52 & 24/12/95 & 3300 & 20.67 & 0.67 & 0.41 + 1es 0647 + 250 & ... & 15.8 & 73.4 & 2.36 & 1.01 & 0.35 & 21/09/98 & 900 & 20.9 & 0.65 & 0.65 + ms 0737.9 + 7441 & 0.315 & 16.9 & 24.0 & 0.46 & 1.10 & 0.40 & 24/12/95 & 1500 & 20.68 & 0.77 & 0.17 + 1es 0806 + 524 & 0.136 : & 15.0 & 171.9 & 1.38 & 1.22 & 0.36 & 24/02/98 & 3000 & 20.73 & 0.96 & 0.20 + ms 0922.9 + 7459 & 0.638 : & 19.7 & 3.3 & 0.22 & 0.90 & 0.39 & 24/12/95 & 3600 & 20.66 & 0.71 & 0.12 + 1es 0927 + 500 & 0.188 & 17.2 & 18.3 & 0.67 & 1.00 & 0.34 & 24/12/95 & 2100 & 20.84 & 0.93 & 0.08 + ms 0950.9 + 4929 & @xmath10.5 & 19.3 & 3.3 & 0.21 & 0.88 & 0.36 & 02/06/95 & 1320 & 20.26 & 0.83 & 0.05 + ms 0958.9 + 2102 & 0.334 & 19.8 & 1.5 & 0.04 & 1.16 & 0.34 & 02/06/95 & 1320 & 19.88 & 0.88 & 0.12 + 1es 1011 + 496 & 0.210 & 16.1 & 286.0 & 0.54 & 1.21 & 0.48 & 24/02/98 & 2400 & 20.75 & 1.66 & 0.05 + 1es 1028 + 511 & 0.239 : & 16.6 & 44.2 & 1.88 & 0.92 & 0.37 & 24/12/95 & 1500 & 20.96 & 0.57 & 0.05 + 1es 1106 + 244 & ... & 18.7 & 18.1 & 0.56 & 0.80 & 0.45 & 25/02/98 & 2400 & 20.75 & 2.02 & 0.05 + 1es 1118 + 424 & 0.124 : & 17.0 & 35.0 & 1.41 & 0.91 & 0.38 & 24/12/95 & 1200 & 21.04 & 0.60 & 0.10 + 1es 1212 + 078 & 0.130 & 16.0 & 94.0 & 0.27 & 1.25 & 0.43 & 25/02/98 & 2400 & 20.23 & 1.53 & 0.08 + 1es 1218 + 304 & 0.182 : & 16.4 & 56.0 & 2.51 & 0.90 & 0.37 & 25/02/98 & 2400 & 20.39 & 0.98 & 0.08 + ms 1221.8 + 2452 & 0.218 : & 17.6 & 26.4 & 0.26 & 1.20 & 0.41 & 31/05/95 & 1200 & 19.95 & 1.14 & 0.10 + ms 1229.2 + 6430 & 0.164 & 16.9 & 42.0 & 0.70 & 1.15 & 0.39 & 01/06/95 & 1200 & 20.83 & 0.59 & 0.10 + ms 1235.4 + 6315 & 0.297 & 18.6 & 7.0 & 0.39 & 0.99 & 0.37 & 01/06/95 & 1200 & 20.78 & 0.70 & 0.08 + 1es 1255 + 244 & 0.141 & 15.4 & 7.4 & 3.66 & 1.30 & 0.17 & 24/12/95 & 900 & 20.36 & 0.58 & 0.08 + ms 1256.3 + 0151 & ... & 20.0 & 8.0 & 0.05 & 1.11 & 0.49 & 31/05/95 & 1200 & 19.41 & 1.34 & 0.08 + ms 1402.3 + 0416 & 0.344 : & 17.1 & 20.8 & 0.10 & 1.33 & 0.34 & 03/06/95 & 1920 & 20.6 & 0.84 & 0.10 + ms 1407.9 + 5954 & 0.495 & 19.7 & 16.5 & 0.41 & 0.81 & 0.52 & 31/05/95 & 1800 & 19.69 & 1.36 & 0.08 + ms 1443.5 + 6349 & 0.299 & 19.6 & 11.6 & 0.33 & 0.85 & 0.49 & 02/06/95 & 1920 & 20.82 & 0.76 & 0.08 + ms 1458.8 + 2249 & 0.235 : & 16.8 & 29.8 & 0.22 & 1.35 & 0.36 & 02/06/95 & 1320 & 20.97 & 0.64 & 0.17
+ 1es 1517 + 656 & @xmath10.7 & 15.9 & 39.0 & 1.19 & 1.11 & 0.30 & 21/09/98 & 2400 & 19.5 & 0.70 & 0.10 + ms 1534.2 + 0148 & 0.312 & 18.7 & 34.0 & 0.43 & 0.94 & 0.51 & 01/06/95 & 1920 & 20.41 & 0.65 & 0.24 + ms 1552.1 + 2020 & 0.222 & 17.7 & 37.5 & 0.89 & 0.97 & 0.44 & 02/06/95 & 2100 & 20.88 & 0.65 & 0.17 + ms 1704.9 + 6046 & 0.280 & 19.1 & 1.8 & 0.10 & 1.13 & 0.30 & 31/05/95 & 1800 & 19.31 & 1.95 & 0.10 + ms 1757.7 + 7034 & 0.407 & 18.3 & 7.2 & 0.48 & 0.98 & 0.35 & 24/09/98 & 2400 & 20.9 & 1.20 & 0.20 + 1es 1853 + 671 & 0.212 & 16.4 & 12.1 & 0.28 & 1.19 & 0.28 & 21/09/98 & 2400 & 21.01 & 0.64 & 0.26 + 1es 1959 + 650 & 0.048 : & 13.7 & 251.6 & 3.64 & 1.19 & 0.32 & 23/09/98 & 1800 & 20.99 & 0.62 & 0.48 + 1es 2037 + 521 & ... & 19.0 & 32.5 & & & & 22/09/98 & 2400 & 20.91 & 0.77 & 2.21 + ms 2143.4 + 0704 & 0.237 & 18.0 & 50.0 & 0.46 & 1.03 & 0.49 & 01/06/95 & 1320 & 20.57 & 0.82 & 0.22 + 1es 2321 + 419 & 0.059 : & 17.0 & 19.0 & 0.27 & 1.19 & 0.32 & 24/09/98 & 1200 & 21.14 & 0.69 & 0.44 + 1es 2326 + 174 & 0.213 & 16.8 & 18.4 & 0.56 & 1.02 & 0.35 & 21/09/98 & 1800 & 20.9 & 0.70 & 0.17 + ms 2336.5 + 0517 & ... & 20.3 & 4.9 & 0.10 & 0.93 & 0.47 & 24/12/95 & 3600 & 18.84 & 0.61 & 0.26 + 1es
2343 - 151 & 0.226 & 19.2 & 8.2 & 0.30 & 0.83 & 0.42 & 23/09/98 & 3000 & 20.9 & 0.85 & 0.10 + 1es 2344 + 514 & 0.044 & 15.5 & 215.2 & 1.14 & 1.18 & 0.41 & 22/09/98 & 1200 & 20.9 & 0.99 & 0.74 + ms 2347.4 + 1924 & 0.515 & 20.8 & 3.2 & 0.10 & 0.86 & 0.47 & 23/09/98 & 1800 & 20.9 & 0.65 & 0.20 + + the redshift distribution of the observed objects from emss and slew is shown in fig .
1 . the average redshifts of the bl lacs with known redshift in the samples are : [email protected] ( emss , all ) ; [email protected] ( emss , observed ) ; [email protected] ( slew , all ) and [email protected] ( slew , observed ) .
it can be seen that a ) the observed and full samples do not differ significantly in their redshift distribution and b ) the slew survey tends to select bl lacs at somewhat lower redshift than the emss survey because of the brighter x - ray flux limit of the slew survey .
optical images were obtained during four observing runs using the 2.5 m not telescope at la palma .
we used the brocam camera ( 1024@xmath23 px , 0176 px@xmath14 ) for observations in 1995 while the hirac camera ( 2048@xmath23 px , 0.11 px@xmath14 ) was used for observations in 1998 ( for details , see table 2 ) . in all observations
the cousins @xmath0 filter was used to image the objects .
the observations were performed mostly during photometric conditions and photometric calibration of each night was obtained from frequent observations of landolt ( 1992 ) standard stars .
some objects were imaged during non photometric conditions , therefore we secured additional short exposure images of these targets during photometric nights to calibrate these frames using reference stars .
seeing conditions were generally very good with average and median seeing fwhm = 0.84 and 0.70 , respectively .
.description of the observing runs .
[ cols="<,<,<,<",options="header " , ] in most cases both a short ( typically 2 minutes ) and a long ( ranging from 10 to 60 minutes , average 20 minutes ) integration were obtained .
the short and long exposures were then combined to form an image of the target well exposed in the external ( fainter ) regions while avoiding saturation of the bright nucleus . in some cases where the nucleus was much brighter than the surrounding nebulosity ,
several intermediate length integrations were combined in order to obtain a final well exposed image . moreover ,
the combination of multiple images allowed for identifying and removing cosmic ray hits . in table 1 , columns ( 8)-(12 ) , we give a journal of the observations with details for each object .
column ( 8) gives the date of observation , columns ( 9 ) and ( 10 ) report the total exposure time and the sky brightness , while in columns ( 11 ) and ( 12 ) the seeing fwhm measured from stellar images and the galactic extinction used are given .
data reduction was performed following standard procedures ( including bias subtraction , flat fielding and cosmic ray rejection ) available in iraf .
the level of the sky was derived sampling several regions over the image and checking for residual gradients in the background level . since no significant gradient
was found a single flux level for the sky was used . from these final images
we have extracted for each object the azimuthally averaged radial brightness profile down to a surface brightness @xmath24 @xmath6 26 mag arcsec@xmath25 .
any obvious extra features ( e.g. companions and foreground stars ) were removed ( masked ) from the image in order to avoid contamination of the radial profile . in order to derive the shape of the point spread function ( psf ) ,
we have similarly extracted the radial profile of a large number of stars in each field . since the field of view of our images is sufficiently large ,
this was always easily obtained using the object frame .
many stellar profiles were combined in order to obtain a good psf both in the core and in the wings . subsequent analysis consisted in comparing each profile of the bl lac object with its psf and , for the resolved sources , fitting the observed luminosity profile with a model .
we used the simple approach of assuming the observed object is composed of a nuclear ( unresolved ) source , described by a psf , plus a galaxy , modeled by either a de vaucouleurs ( @xmath21(r ) @xmath26 r@xmath27 ) or an exponential disk ( @xmath21(r ) @xmath26 r ) , convolved with the proper psf .
more complex models , for example a generalized de vaucouleurs law ( e.g. the sersic law @xmath28 ) , increases the number of free parameters and do not offer real advantages for the characterization of the host galaxies . from the best fit of the profiles
we have determined the parameters of the host galaxy ( @xmath29 , r@xmath30 , total magnitude ) as well as the magnitude of the nuclear source .
these parameters are reported in table 3 , columns ( 1)-(7 ) , where column ( 1 ) gives the name of the object , column ( 2 ) its redshift , columns ( 3 ) and ( 4 ) the central surface brightness of bulge component and its scale length , columns ( 5)(6 ) the apparent magnitude of the nuclear and galactic components , and column ( 7 ) the reduced @xmath31 of the best fit .
absolute quantities were derived after applying correction for galactic extinction and redshift ( k - correction ) .
the former was determined using the bell lab survey of neutral hydrogen n@xmath32 converted to e@xmath33 ( stark et al .
1992 ; shull & van steenberg 1985 ) , while the latter was computed from the model of coleman et al ( 1980 ) for elliptical galaxies .
lllllllllllll name & z & @xmath34 & r@xmath30 & m@xmath35 & m@xmath36 & @xmath31 & k cr .
& m@xmath37 & m(host ) & r(e ) & @xmath21(e ) & note + & & & ( @xmath22 ) & ( mag ) & ( mag ) & & & ( mag ) & ( mag ) & ( kpc ) & & + 1es 0033 + 595 & ... & 12.00 & 0.40 & 18.88 & 18.92 & 1.50 & ... & ... & ... & ... & ... & + 1es 0120 + 340 & 0.272 & 14.97 & 3.40 & 16.39 & 17.24 & 7.17 & 0.29 & -25.19 & -24.63 & 18.89 & 21.72 & a ) + ms 0122.1 + 0903 & 0.339 & 13.60 & 0.90 & 15.43 & 18.75 & 0.52 & 0.38 & -26.65 & -23.70 & 5.79 & 20.08 & + ms 0158.5 + 0019 & 0.229 : & 14.70 & 2.05 & 18.82 & 18.07 & 0.26 & 0.32 & -22.87 & -23.94 & 12.15 & 21.45 & + ms 0205.7 + 3509 & 0.318 : & 14.50 & 1.20 & 17.53 & 19.03 & 0.06 & 0.35 & -24.48 & -23.33 & 7.41 & 20.99 & a ) + 1es 0229 + 200 & 0.139 & 13.84 & 4.00 & 18.25 & 15.76 & 0.04 & 0.13 & -21.91 & -24.53 & 13.33 & 21.06 & + ms 0257.9 + 3429 & 0.247 & 14.44 & 2.05 & 19.07 & 17.81 & 1.32 & 0.27 & -22.45 & -23.98 & 10.64 & 21.13 & + ms 0317.0 + 1834 & 0.190 & 12.70 & 1.10 & 19.01 & 17.42 & 4.10 & 0.20 & -21.93 & -23.72 & 4.70 & 19.61 & a ) + 1es 0347 - 121 & 0.188 & 11.80 & 0.65 & 19.37 & 17.66 & 1.16 & 0.20 & -21.26 & -23.17 & 2.76 & 19.01 & + 1es 0414 + 009 & 0.287 & 12.54 & 1.00 & 16.97 & 17.47 & 0.41 & 0.31 & -24.91 & -24.72 & 5.76 & 19.05 & + ms 0419.3 + 1943 & 0.512 : & 14.10 & 0.74 & 18.65 & 19.68 & 3.31 & 0.75 & -25.12 & -24.84 & 6.05 & 19.04 & + 1es 0446 + 449 & 0.203 : & 16.11 & 22.90 & ... & 14.24 & ... & 0.21 & ... & -29.13 & 102.92 & 20.90 & + 1es 0502 + 675 & 0.341 & 10.98 & 0.35 & 16.83 & 18.19 & 5.25 & 0.38 & -25.60 & -24.62 & 2.26 & 17.12 & a ) + 1es 0525 + 713 & 0.249 : & 14.35 & 2.30 & ... & 17.47 & 1.27 & 0.27 & ... & -24.34 & 12.01 & 21.03 & + ms 0607.9 + 7108 & 0.267 & 14.1 & 1.85 & 18.31 & 17.69 & 1.35 & 0.29 & -23.40 & -24.31 & 10.1 & 20.7 & + 1es
0647 + 250 & ... & ...
& ... & 15.03 & ... & 12.95 & 0.21 & -26.26 & ... & ... & ... & + ms 0737.9 + 7441 & 0.315 & 12.97 & 1.10 & 18.29 & 17.69 & 4.01 & 0.34 & -23.58 & -24.52 & 6.75 & 19.60 & + 1es 0806 + 524 & 0.136 : & 12.05 & 1.70 & 15.77 & 15.83 & 2.32 & 0.13 & -24.16 & -24.24 & 5.63 & 19.48 & + ms 0922.9 + 7459 & 0.638 : & 15.26 & 1.03 & 20.34 & 20.12 & 16.87 & 1.15 & -23.30 & -24.67 & 9.40 & 20.17 & + 1es 0927 + 500 & 0.188 & 15.50 & 3.00 & 17.65 & 18.04 & 2.20 & 0.20 & -22.89 & -22.70 & 12.72 & 22.80 & + ms 0950.9 + 4929 & @xmath10.5 & ... & ... & 18.78 & ... & 0.04 & 0.22 & -21.95 & ... & ... & ... & + ms 0958.9 + 2102 & 0.334 & 15.18 & 1.82 & 20.66 & 18.81 & 0.10 & 0.37 & -21.30 & -23.52 & 11.60 & 21.76 & + 1es 1011 + 496 & 0.210 & 14.37 & 2.73 & 16.07 & 17.12 & 15.92 & 0.21 & -24.58 & -23.74 & 12.13 & 21.64 & + 1es 1028 + 511 & 0.239 : & 13.75 & 1.30 & 16.69 & 18.11 & 5.27 & 0.26 & -24.39 & -23.23 & 6.59 & 20.83 & + 1es 1106 + 244 & ... & 15.60 & 1.40 & 18.71 & 19.8 & 5.35 & 0.48 & -23.64 & -23.03 & 9.97 & 21.93 & a ) + 1es 1118 + 424 & 0.124 : & ... & & 16.88 & ... & 1.83 & 0.11 & -22.71 & ... & ... & ... & + 1es 1212 + 078 & 0.130 & 14.64 & 5.51 & 17.26 & 15.86 & 0.33 & 0.12 & -22.42 & -23.94 & 17.38 & 22.23 & + 1es 1218 + 304 & 0.182 : & 13.55 & 2.17 & 16.32 & 16.8 & 0.18 & 0.19 & -24.14 & -23.85 & 8.97 & 20.88 & + ms 1221.8 + 2452 & 0.218 : & 12.46 & 0.62 & 17.76 & 18.43 & 75.64 & 0.23 & -23.15 & -22.71 & 2.94 & 19.60 & + ms 1229.2 + 6430 & 0.164 & 13.51 & 2.90 & 17.35 & 16.13 & 2.06 & 0.17 & -22.89 & -24.28 & 11.05 & 20.91 & + ms 1235.4 + 6315 & 0.297 & 13.80 & 1.39 & 19.20 & 18.01 & 0.11 & 0.32 & -22.43 & -23.95 & 8.20 & 20.60 & + 1es 1255 + 244 & 0.141 & 13.10 & 1.90 & 18.27 & 16.64 & 1.23 & 0.14 & -21.59 & -23.36 & 6.41 & 20.63 & + ms 1256.3 + 0151 & ... & ... & ... & 19.50 & ... & 0.14 & ... & ... & ... & ... & ... & + ms 1402.3 + 0416 & 0.344 : & ... & ... & 16.88 & ... & 0.60 & 0.39 & -25.14 & ... & ... & ... & + ms 1407.9 + 5954 & 0.495 & 14.80 & 1.00 & 18.69 & 19.73 & 5.53 & 0.70 & -24.23 & -23.89 & 8.04 & 20.60 & + ms 1443.5 + 6349 & 0.299 & 15.49 & 2.90 & 19.38 & 18.11 & 2.90 & 0.32 & -22.27 & -23.86 & 17.18 & 22.28 & + ms
1458.8 + 2249 & 0.235 : & 10.50 & 0.46 & 16.27 & 17.12 & 0.98 & 0.25 & -24.89 & -24.29 & 2.30 & 17.49 & + 1es 1517 + 656 & @xmath10.7 & ... & ... & 16.38 & ... & 6.86 & 1.35 & -27.49 & ... & ... & ... & + ms 1534.2 + 0148 & 0.312 & 15.81 & 4.05 & 19.22 & 17.7 & 10.38 & 0.34 & -22.70 & -24.56 & 24.68 & 22.38 & b ) + ms 1552.1 + 2020 & 0.222 & 13.80 & 2.35 & 17.97 & 16.87 & 0.59 & 0.24 & -23.05 & -24.39 & 11.29 & 20.84 & + ms
1704.9 + 6046 & 0.280 & 14.95 & 2.40 & 19.49 & 17.98 & 0.20 & 0.30 & -22.02 & -23.83 & 13.60 & 21.80 & + ms 1757.7 + 7034 & 0.407 & 12.14 & 0.30 & 17.89 & 19.68 & 0.84 & 0.50 & -24.65 & -23.36 & 2.16 & 18.28 & a ) + 1es 1853 + 671 & 0.212 & 15.64 & 2.50 & 19.47 & 18.58 & 0.51 & 0.23 & -21.53 & -22.65 & 11.61 & 22.64 & + 1es 1959 + 650 & 0.048 : & 13.25 & 4.10 & 14.86 & 15.12 & 10.85 & 0.04 & -22.97 & -22.75 & 5.34 & 20.85 & + 1es 2037 + 521 & ... & 14.86 & 5.80 & 19.64 & 15.97 & 2.17 & 0.04 & -20.01 & -23.72 & 7.84 & 20.72 & + ms 2143.4 + 0704 & 0.237 & 14.62 & 2.17 & 18.24 & 17.87 & 0.18 & 0.26 & -22.99 & -23.62 & 10.94 & 21.54 & + 1es 2321 + 419 & 0.059 : & ... & ... & 17.31 & ... & 1.35 & 0.05 & -20.94 & ... & ... & ... & + 1es 2326 + 174 & 0.213 & 13.06 & 1.60 & 18.41 & 16.97 & 2.42 & 0.23 & -22.51 & -24.18 & 7.46 & 20.15 & + ms 2336.5 + 0517 & ... & 12.60 & 0.60 & 19.83 & 18.64 & 0.08 & ... & ... & ... & ... & ... & + 1es 2343 - 151 & 0.226 & 13.27 & 1.60 & 20.36 & 17.18 & 0.67 & 0.24 & -20.63 & -24.05 & 7.79 & 20.37 & b ) + 1es 2344 + 514 & 0.044 & 12.87 & 5.80 & 17.00 & 13.98 & 0.25 & 0.03 & -20.89 & -23.95 & 6.96 & 20.24 & + ms 2347.4 + 1924 & 0.515 & 13.27 & 0.40 & 21.89 & 20.19 & 0.75 & 0.76 & -21.26 & -23.72 & 3.28 & 18.83 & + + + +
in fig . 2 we report for each object the radial luminosity profile together with its best - fit decomposition into nucleus and host galaxy components .
we are able to resolve 45 objects out of the 52 observed sources . for almost all of the clearly resolved sources we are able to fit the luminosity profile of the host galaxy with an elliptical model while a disk model gave a significantly worse fit .
for a few distant and marginally resolved sources ( see table 3 and fig .
2 ) , we can not rule out a disk model but even in these cases an elliptical model is a good representation of the host galaxy
. continued . continued .
the absolute magnitudes of the nucleus and the host galaxies are reported in table 3 , columns ( 8)-(12 ) , where column ( 8) gives the applied k correction , columns ( 9 ) and ( 10 ) the absolute magnitudes of the nuclear component and the host galaxy , and columns ( 11 ) and ( 12 ) the scale length and surface brightness at the effective radius .
the distributions of the host galaxy magnitude in the emss and slew samples are illustrated in fig .
the average absolute magnitude of the host galaxies is @xmath2 = [email protected] ( all ) , [email protected] ( emss ) and [email protected] ( slew ) , the average scale length of the host is @xmath4r(e)@xmath1 = [email protected] kpc ( all ) , [email protected] kpc ( emss ) and [email protected] kpc ( slew ) , while the average absolute magnitude of the nucleus is @xmath38 = [email protected] ( all ) , [email protected] ( emss ) and [email protected] ( slew ) .
the average host luminosity is in good agreement with previous studies ( e.g. @xmath39 = [email protected] , wsy96 ; @xmath39 = [email protected] , u99b ) and confirms that the host galaxies of bl lacs are almost without exception giant ellipticals .
no significant difference is found in the distribution of these values between the emss and slew samples . at high luminosities
the two distributions are very similar while at lower luminosities slew hosts appear slightly but not significantly fainter than emss sources .
this can be due differences in the selection procedure .
a k - s test between the two distributions yields p@xmath40 = 0.1 , confirming that the two distributions are practically indistinguishable . since we have 22 objects in common with the cfht survey of bl lac objects ( wsy96 ) .
it is interesting to compare our results for individual sources with those obtained by wsy96 . of the 22 bl lacs , three are unresolved by the cfht and/or the not data .
thus , the final comparison is based on 19 resolved sources in common between the not and cfht samples . before comparing the results we have transformed the wsy96 host magnitudes from the gunn @xmath41band they used into our cousins @xmath0band assuming @xmath41@xmath0 = 0.3 and applying a small correction ( @xmath6 0.1 mag ) for the different cosmology used ( q@xmath42 = 0.5 instead of q@xmath42 = 0 ) .
4 shows the apparent host magnitudes from the two studies plotted against each other , with a one to
one correspondence superimposed . on average , the agreement is quite good , even if in few cases the difference is quite large ( 1.7 mag for 0922 + 749 and 1.1 mag for 0419 ) . for both these cases there are observations obtained with hst ( see the appendix ) that agree with our values within few tens of mag .
the average and median difference in the host apparent magnitude is @xmath43 = [email protected] and 0.07 , respectively .
we have 40 objects in common with the hst snapshot survey of bl lac objects ( scarpa et al . 1999b ; u99b ) .
although the not and hst data are taken with different instruments and have different spatial resolution , the data have been analyzed homogeneously , facilitating comparison between the samples .
of the 40 bl lacs , four are unresolved by hst . of these four sources , 1es 0647 + 250 , ms 1402.3 + 0416 and 1es 1517 + 656
remain also unresolved by us , while for 1es 0033 + 595 we are able to detect a probable nebulosity .
this observation is however complicated because of the presence of a very bright star close to the target ( see individual notes ) .
the comparison is therefore based on 35 sources in common between the not and hst samples .
4 shows the apparent host magnitudes from the two studies plotted against each other , with a one to
one correspondence superimposed .
the average and median difference in the host magnitude with respect to the results of this study are @xmath44 = [email protected] and 0.1 , respectively .
in addition we show in fig .
5 the comparison of the distributions of the absolute host magnitudes from the two whole surveys . on average the agreement is also very good .
the average host and nuclear luminosities of the hst survey are @xmath45 = [email protected] and @xmath46 = [email protected] .
the difference of average and median host luminosity with respect to the results of this study are @xmath47 = -0.1 and -0.2 , respectively . in fig .
6 we compare the radial profiles observed in this study of three bl lacs with those derived from hst images ( scarpa et al 1999b ) .
the three objects were chosen to represent different well resolved ( ms 0257.9 + 3429 ) , poorly resolved ( 1es 1218 + 304 ) and unresolved sources ( ms 1402.3 + 0416 ) both in the not and hst images . while the hst data allow one to investigate the host much closer to the nucleus , the not images are clearly much deeper .
this is partly due to the longer exposure times and to the favorable pixel scale of the not data .
this translates into a better capability for the not images of mapping the faint outer regions of the galaxies .
also , fig . 6 clearly shows the effect of different seeing on the psf in the ground based images ( 05 for ms 0257.9 + 3429 and 10 for 1es 1218 + 304 ) .
an optimal characterization of the host galaxies would thus be obtained by combining the high resolution ( hst ) images with the deeper ground - based observations .
the relevance of this point in a larger sample of bl lacs will be discussed in a forthcoming paper . according to the unified schemes of radio - loud agn ( see e.g. urry & padovani 1995 ) , bl lacs and rgs of f - r i type morphology are identical objects seen along different angles with respect to the relativistic jet .
a number of optical studies have reported on the optical properties of rgs ( see govoni et al 1999 and references therein ) .
these have almost unanimously found that hosts are luminous ellipticals sometimes interacting with other galaxies , other times rather isolated .
a general trend has been reported suggesting that f - r i rgs are on average brighter and in denser environments than those hosting f - r ii rgs
. comparison among various samples is hindered by the selection procedure and by possible systematic effects introduced by the analysis ( e.g. extinction , k - correction , passband , method of measurement of galaxy magnitude , etc ) . in order to reduce as much as possible the systematic effects we compare our results with those obtained by govoni et al ( 1999 ) for a large sample of low redshift rgs .
this sample includes 79 rgs of both f - r i and f - r ii type in the redshift interval 0.01 to 0.1 .
@xmath0band imaging is used to investigate in detail the morphological and photometric properties of the radio galaxies .
this includes the analysis of the luminosity profile using the same procedure and the corrections used in this work .
the average host luminosity of the f - r i and f - r ii rg hosts are @xmath48 = [email protected] and @xmath49 = [email protected] .
the bl lac hosts appear therefore on average slightly brighter than f - r ii hosts , but also slightly fainter than the f - r i rg hosts , which are in turn quite similar to bcgs in moderately rich clusters at [email protected] ( @xmath50 = [email protected] ; hoessel et al .
note that since the rgs are at lower redshift , any cosmological evolutionary correction makes the difference from fr i even larger .
a similar trend was also noted by lamer mc hardy newsam ( 1999 ) comparing bl lacs and fr i galaxies in the near infrared .
on the other hand the comparison between bl lac hosts and the whole sample of rg shows a general good agreement . in fig .
7 we report the histogram and cumulative distribution of the absolute host magnitudes of the bl lacs and the rgs from govoni et al .
the two distributions are rather similar with only a significant excess of more bright rg that are not present in the bl lacs .
if we compare bl lacs with f - r i and f - r ii separately the agreement is formally good for f - r ii but not for f - r i. a k - s test yields formally p@xmath40 = 0.250 and 0.004 for f - r ii and f - r i respectively compared with bl lac host luminosities .
since the selection procedure that identifies samples of bl lacs and of rg is different , we can not rule out that selection effects may introduce some bias .
these effects should be related to correlations between host optical luminosity and x - ray and radio properties which are used to classify the objects .
the present result indicates that from the point of view of host luminosities either the parents of bl lacs are rg of both types ( i and ii ) or for some reason brightest f - r i must be excluded from the parent population as proposed by wsy96 .
the first association of bl lacs with f - r i was done on the basis of similar extended radio emission ( browne 1983 , wardle et al 1984 ; antonucci & ulvestad 1985 ) .
there are however some observations ( e.g. kollgaard et al 1992 , murphy et al 1993 ) that indicate that at least some bl lac object may have diffuse emission more similar to that of f - r ii than f - r i. the same conclusion is reached from the kpc scale radio polarization study of six bl lacs ( stanghellini et al 1997 ) where the magnetic field is parallel to the radio jet axis for most of its length as found in f - r ii sources .
if the radio morphology of these bl lacs be of f - r i type the magnetic field should be dominated by the component perpendicular to jet axis .
our results therefore are consistent with the idea that bl lacs avoid the bcgs in rich clusters at low redshift ( wurtz et al.1997 ) although there are examples of bl lacs in very luminous hosts and members of group / clusters of galaxies ( see 1es 0414 + 009 and pks 0548 - 322 ; this study and falomo et al 1995 ; 1es1741 + 196 heidt et al 1999 ) .
this idea is also supported by the narrow - angle tail radio morphology seen in many bl lacs and non - bcgs ( e.g. owen & laing 1989 ; perlman & stocke 1993 ) and by the correlation between cluster richness and the bcg and bl lac luminosity ( thuan & romanishin 1981 ; wurtz et al . 1997 ) . on
the other hand evidence is growing that also from the point of view of the extended radio luminosities many bl lacs are quite different from fr i and share the characteristic luminosity of fr ii sources ( cassaro et al 1999 ) .
this evidence together with our results on the host luminosities led to argue that both types of rg form the parent population of bl lacs .
it is well established that elliptical galaxies form families of homologous systems with characteristic parameters r(e ) , @xmath21(e ) and velocity dispersion @xmath51 .
these are commonly represented in the fundamental plane ( see e.g. djorgovski & davis 1987 ) .
we have investigated the properties of the bl lac hosts in the projected fundamental plane ( f - p ) concerning the central surface brightness @xmath21(e ) and the scale length r(e ) .
surface brightness @xmath21(e ) derived from the fit ( see table 3 ) was corrected for galactic extinction , k correction and for the ( 1+z)@xmath52 cosmological dimming .
8 shows the correlation between @xmath21(e ) and log r(e ) for the bl lac hosts and the low redshift rg hosts ( govoni et al .
it can be seen that the behavior of the emss and slew bl lac hosts are similar . both bl lac hosts and the rg hosts ( govoni et al .
1999 ) follow the kormendy relation for giant massive ellipticals ( e.g. capaccioli et al . 1992 ) , with a best - fit correlation @xmath21(e ) = 16.45 + 4.6 @xmath10 ( log r(e ) ) .
practically no host galaxy is in the ( scatter ) area at log r(e ) @xmath4 0.5 kpc .
this confirms that the bl lac hosts are almost exclusively drawn from the population of giant ellipticals and not from normal field ellipticals .
note that the bl lac hosts seem to avoid the area of the brightest and largest galaxies in the bottom right hand corner of fig .
8 , similarly to the result based on the total host luminosities ( fig . 7 ) . in fig .
9 we show the apparent r magnitude vs. redshift hubble diagram for the bl lac hosts ( this work ) , together with data for rgs ( govoni et al .
1999 ) , compared with the expected relationship for elliptical galaxies derived from passive stellar evolution models ( bressan et al 1994 ) normalized to the average redshift and magnitude of the low redshift rgs from govoni et al .
the resolved bl lac hosts lie within about 1 mag spread on the h z relation , as do the rgs .
based on this diagram we can estimate the redshift of the four objects with unknown redshift but resolved in our images . these are 0033 + 59 , 1106 + 24 , 2037 + 52 , and 2336 + 05 ( see fig .
9 ) for which we derive a hubble law redshift of 0.43 , 0.58 , 0.14 and 0.40 respectively .
optical images of bl lacs are characterized by the superposition of two components : the extended starlight emission from the galaxy and the non - thermal unresolved bright nucleus .
we discuss here the properties of these two components as derived from our sample .
10 shows the relationship between absolute magnitude of hosts and redshift of our bl lacs and of other samples of agn .
all bl lac hosts at all redshifts have luminosities between that of a passively evolving m@xmath7 and m@xmath72 mag galaxies , where m@xmath7 is the characteristic luminosity of a shechter luminosity function . while there are rg hosts ( govoni et al .
1999 ) with luminosities larger than m@xmath72 , none of the observed bl lacs are found in such a luminous galaxy .
note that any correction due to different average redshift would make this difference even larger .
there is a suggestion of a positive correlation of host luminosity with redshift , consistently with what is expected from passive stellar evolution models for elliptical galaxies ( e.g. bressan et al .
1994 ; fukugita et al .
1995 ) , and the evolution of galaxies in clusters ( ellingson et al 1991 ) , although the scatter is large .
this trend is similar to that suggested by wsy96 and is consistent with the average value found for higher redshift flat spectrum radio quasar ( fsrq ) host galaxies ( kotilainen et al 1998a ) : @xmath53 = 0.673 , @xmath54 = 25.3 and even with high redshift rlqs ( lehnert et al 1992 ) : @xmath55 = 2.34 , @xmath56 = 26.3 .
11 shows the histogram of the ratio between nuclear and host luminosity for the resolved bl lac objects .
this ratio ranges from 0.03 to 3 , with average @xmath60.8 and median @xmath60.4 .
interestingly , we find an apparently bimodal distribution with two peaks around log(ln / lh ) @xmath60.6 and log(ln / lh ) @xmath60.2 .
this behavior is present in both the emss and slew sub - samples .
while the detected large range in the luminosity ratio within individual objects can be due to differences in the intrinsic nuclear or host luminosity , or a difference in the beaming factor from one object to another , it is difficult to explain the apparent bimodality present in both emss and slew samples . in comparison , the luminosity ratio in the @xmath19band for the low redshift quasars studied by bahcall et al .
( 1997 ) is @xmath610 while for the rgs observed in the @xmath0band ( govoni et al 1999 ) this ratio is less than 0.1 . for bl lacs therefore the observed nuclear optical luminosity seems intermediate between that of rg and quasars . in order to investigate the relationship of the optical nuclear luminosity with that in different bands
we have performed a partial correlation analysis among radio , optical , and x - ray luminosities .
we used data in table 1 together with nuclear optical data of table 3 and used the procedure described in padovani ( 1992 ) .
for all objects with known redshift we computed the spearman rank order correlation coefficients between luminosities and between luminosity in a given band and redshift .
the results of the correlation analysis are given in table 4 .
it turns out that apart the weak correlation with redshift ( which is marginally significant only for l@xmath57 ) the only significant correlation among luminosities after subtraction of the redshift dependency is between l@xmath58 and l@xmath59 .
a similar result was obtained from the analysis of a smaller sample of x - ray bl lacs by padovani ( 1992 ) but with a smaller correlation coefficient .
lllll + & l@xmath59 & l@xmath58 & l@xmath57 & z + l@xmath59 & 1.0 & 0.74 & 0.55 & 0.30 + l@xmath58 & ... & 1.0 & 0.48 & 0.27 + l@xmath57 & ... & ... & 1.0 & 0.45 + z & ... & ... & ... & 1.0 + fig .
12 shows the relationship between the absolute host and nuclear magnitudes of the bl lacs in emss and slew .
there is an indication of a weak correlation ( spear correlation coefficient .
= 0.3 ) between the two parameters , in the sense that more luminous nuclei reside in more luminous hosts .
however , we note the obvious selection effects that may de - populate the upper left hand ( faint nuclei in luminous hosts ) and lower right hand ( faint hosts swamped by luminous nuclei ) corners of the diagram . while the first effect could bias the original samples ( see e.g. browne & marcha 1993 ) the second should have marginal effect since , excluding misclassified objects we are able to resolve 90% of observed targets .
the putative correlation is , however , consistent with the luminosity / host - mass limit found when considering agn samples at higher redshift and with more luminous nuclei and hosts ( kotilainen et al .
1998a ; lehnert et al .
1992 ) . also , assuming that bl lac activity results from accretion of material onto a super - massive black hole , it is in agreement with the relationship found by magorrian et al .
( 1998 ) from hst kinematic study between the mass ( luminosity ) of the black hole and the mass ( luminosity ) of the spheroid component in nearby galaxies .
mclure et al .
( 1999 ) detected a similar weak correlation for a small sample of 9 rlqs at [email protected] , and calculated that most of their rlqs radiate at a few percent of the eddington luminosity .
the correlations found in this study and in mclure et al .
( 1999 ) indicate that the magorrian et al .
relationship extends to galaxy and host galaxy masses at cosmological distances . on the other hand the lack of a strong correlation found in this study of relatively nearby and modest luminosity agn
may indicate that the onset of the correlation occurs only after a certain level in nuclear and/or galaxy luminosity has been reached ( c.f . with kotilainen
we have presented high resolution homogeneous optical observations of a large data set of bl lac objects drawn from two complete samples of x ray selected ( high frequency peaked ) sources .
we are able to resolve @xmath6 90% of the observed targets and study the properties of their host galaxies .
it turns out that all galaxies are well represented by elliptical model with mean luminosity @xmath2 = 23.9 and an observed average nuclear source - to - host luminosity ratio of @xmath6 1 .
it is also shown that hosts of bl lacs are almost exclusively drawn from the population of massive elliptical galaxies .
no cases of disc dominated systems hosting bl lacs are found supporting the view that all radio - loud active galaxies are dominated by the spheroidal component .
the global properties of the galaxies hosting bl lacs follow the same relationships of normal ( non active ) giant ellipticals under passive stellar evolution .
the comparison of properties of bl lac hosts with those of low redshift radio galaxies indicates that from the point of view of galaxy luminosity both fr i and fr ii radio galaxies form the parent population of bl lacs . a result that is supported recent radio imaging and polarization studies that show many bl lacs exhibiting the characteristics of fr ii sources .
in spite of the good seeing conditions seven bl lacs remained unresolved .
for all these objects the redshift is unknown or uncertain . for two unresolved bl lacs ( ms 0950.9 + 4929 and 1es
1517 + 656 ) , the redshift ( or lower limit of redshift ) is high enough to explain the lack of detection of the host galaxy . for the other five unresolved objects ( 1es 0647 + 250 , 1es 1118 + 424 , ms 1256.3 + 0151 ms
1402.3 + 0416 and 1es 2321 + 419 ) , there are only tentative redshifts which are , however , inconsistent with the results from our images .
we determined a lower limit of the redshift assuming these objects are surrounded by a typical elliptical host galaxy as found in this study ( m(r ) = 23.8 and r(e ) = 10 kpc ) .
a simulated galaxy was produced and superposed onto the image of the observed sources assuming various redshifts .
this produced images of the object that appeared resolved or not depending on the redshift used .
a lower limit of the redshift was set when the galaxy become undetectable .
of course using host galaxies that are less luminous and smaller would make these limits lower .
the derived lower limits for the unresolved bl lacs are : z @xmath1 0.3 for 1es 0647 + 250 , z @xmath1 0.6 for ms 0950.9 + 4929 , z @xmath1 0.5 for 1es 1118 + 424 , z @xmath1 0.65 for ms 1256.3 + 0151 , z @xmath1 0.5 for ms 1402.3 + 0416 , z @xmath1 0.45 for 1es 1517 + 656 and z @xmath1 0.45 for 1es 2321 + 419 . with the exception of 1es 0647 + 250 ,
the observed magnitude of these objects is consistent with them being at moderately high redshift , thus remaining unresolved .
the case of 1es 0647 + 250 could be due to either a very bright ( or beamed ) nucleus or an under - luminous host galaxy . since some of these bl lacs have tentative redshift ( see individual notes ) either these objects are hosted by an atypically faint host galaxy or , more likely ,
the redshift is wrong . *
1es 0647 + 250 * : a tentative redshift z = 0.203 has been derived for this bl lac , which is unresolved in the not images taken with very good seeing ( fwhm 065 ) .
it is also unresolved in the hst snapshot survey ( scarpa et al .
1999b ) , who derived an upper limit m(host ) @xmath119.1 . at z = 0.203
, this corresponds to m(host ) @xmath121.7 , i.e. 2 magnitudes fainter than an average bl lac host . *
ms 0950.9 + 4929 * : this bl lac , for which a lower limit to the redshift of z @xmath1 0.5 has been derived ( perlman , priv .
comm . ) , remains unresolved in the not data , even in good seeing .
it was also unresolved by wsy96 , who derived an upper limit m(host ) @xmath119.2 .
the corresponding absolute luminosity limit from wsy96 is m(host )
@xmath1 24.4 , in good agreement with the average value of the bl lac hosts derived here .
* 1es 1118 + 424 * : this bl lac , at an uncertain redshift z = 0.124 ( perlman et al , priv . comm ) , remains unresolved in the not data , even in very good seeing ( fwhm 060 ) .
the redshift is inconsistent with results from our image . *
ms 1256.3 + 0151 * : this bl lac , at tentative redshift z = 0.162 , is unresolved in the not data but the seeing during this observation was rather poor ( fwhm 134 ) .
however , even taking into account the poor seeing , if it is at z = 0.162 its host galaxy should be easily detectable ( unless it is atypically faint ) .
we estimate it should be at @xmath60 0.5 if hosted by a standard galaxy . *
ms 1402.3 + 0416 * : this bl lac , at a tentative redshift z = 0.344 , remains unresolved in the not data .
it was also unresolved by wsy96 , who derived an upper limit m(host ) @xmath118.4 , which at z = 0.344 corresponds to m(host ) @xmath1 23.9 .
similarly , it remained unresolved in the hst snapshot survey ( scarpa et al 1999b ) , who derived an upper limit m(host ) @xmath119.4 , which at z = 0.344 corresponds to m(host ) @xmath1 22.9 .
these upper limits still allow for the existence of a typical elliptical host galaxy which at the moderately high redshift remains unresolved .
* 1es 1517 + 656 * : this bl lac , for which beckmann ( priv .
comm . ) has derived a spectroscopic lower limit of z @xmath10.7 based on the presence of mgii and feii absorption lines , remains unresolved in the not data , even in good seeing ( fwhm 070 ) .
it was also unresolved in the hst snapshot survey ( scarpa et al 1999b ) , with an upper limit m(host ) @xmath119.9 , which at z @xmath10.7 corresponds to m(host ) @xmath4 25.2 . at this upper limit
the observations are consistent with the presence of a luminous host galaxy . at hst resolution
, this bl lac shows an unusual morphology with three non homogeneous arclets surrounding the point source at 1 - 2 arcsec of distance ( scarpa et al 1999a ) .
we are able to see clearly the more external arc and can detect the two more internal features after subtraction of a scaled psf .
our deeper images show that at least the most external ring does not extend more than what it is seen on hst data .
moreover we detect many other faint sources in the immediate 5 arcsec environment .
it was proposed that these arcs are effects of gravitational lensing produced by a foreground galaxy .
our image is much deeper than that obtained with hst but no signature of foreground galaxy is found .
this therefore weakens the lens hypothesis . *
1es 2321 + 419 * : this bl lac , at a tentative redshift z = 0.059 , remains unresolved by our images . since the seeing conditions were very good ( fwhm 069 ) we believe the correct redshift is considerably larger than that proposed . *
1es 0446 + 449 * : this bl lac at a tentative redshift z = 0.203 was resolved by the hst snapshot survey ( scarpa et al .
1999b ) , but no point source was detected while a pure exponential ( disk ) brightness profile was observed .
the not data confirm that no point source is present in this object .
instead , we detect a very luminous galaxy with m(host ) = 15.1 , corresponding to an extremely luminous galaxy ( m@xmath61 = 29.1 at the proposed redshift ) making it a very unusual source . however , a more likely explanation is that the correct redshift for this object is much lower than z = 0.203 and that this source is not a bl lac object . from re - inspection of the optical spectrum for this object ( perlman et al .
1996 ) , we note that the redshift determination is probably wrong and the correct redshift is [email protected] , which would make the identified counterpart to be a normal low redshift elliptical galaxy , and not a bl lac object . *
1es 0525 + 713 * : the not data failed to detect a point source at the tentative redshift z = 0.249 , while the elliptical host galaxy has m(host ) = 17.7 and m(host ) = 24.3 .
similar results was obtained from hst images ( scarpa et al 1999b ) .
imaging results together with the lack a power law continuum in the optical spectrum ( perlman et al . 1996 ) strongly suggest it is not a bl lac source . * 1es 0033 + 595 * : the hst observations of this bl lac at unknown redshift ( although a tentative redshift z = 0.086 was derived by perlman , priv . comm . )
have been discussed by scarpa et al ( 1999a ) .
it is a gravitational lens candidate consisting of two objects ( a and b ; see scarpa et al 1999a ) with similar @xmath0band brightness and separation of @xmath616 .
in addition to @xmath0band images we obtained also data in the @xmath62 , @xmath63 and @xmath19bands .
this allows us to compare colors of the two objects .
these turned out to be significantly different with @xmath62@xmath63 = 0.4 and -0.1 , @xmath63@xmath19 = 1.4 and 1.7 , and @xmath19@xmath0 = 0.9 and 1.9 for a and b , respectively .
due to the bluer color it is likely that the bl lac is the object b while a is a red galactic star ( the source is close to the galactic plane ) .
the probable bl lac in the pair ( component b ) remained unresolved in the hst snapshot survey ( scarpa et al 1999b ) , with an upper limit m(host ) @xmath120.0 . from the analysis of the not image
we are able to detect an excess of light at radii larger than 2 arcsec corresponding to surface brightness fainter than 24 mag / arcsec@xmath25 .
if the nebulosity is attributed to the host galaxy this would correspond to an object of 19.0 mag .
a caution is however needed for this detection because of the presence of a very bright star in the field that contaminates the background emission . *
ms 0205.7 + 3509 * : this source , for which wsy96 preferred a disk host galaxy surrounding a significantly de - centered nucleus , was recently studied in detail by us ( falomo et al .
we refuted both the disk model and the gravitational lens interpretation ( stocke et al .
1995 ) , and identified ms 0205.7 + 3509 as an elliptical host galaxy with no de - centering but with a close companion galaxy .
* ms 0317.0 + 1834 * : wsy96 could not distinguish between elliptical and disk host for this source , for which they derived m(r ) = 23.0 and r(e ) = 5.1 kpc . after masking out the close companion , we clearly prefer an elliptical host with m(r ) = 23.7 and r(e ) = 4.7 kpc , a somewhat brighter host . *
1es 0347 - 121 * : the host galaxy is clearly resolved with the radial profile well described by an elliptical model .
we derive for the host m(host ) = 23.2 , identical to that derived by u99b .
there is an interacting system of three galaxies located 12 arcsec n of the bl lac but no signs of a physical connection with the bl lac are apparent . *
1es 0414 + 009 * : this bl lac was resolved into an elliptical galaxy with m(r ) = 24.8 by the hst ( u99b ) .
it was also studied by falomo ( 1996 ) who derived for the host m(r ) = 24.3 and r(e ) = 5 kpc .
we derive for the elliptical host at z = 0.287 m(r ) = 24.7 and r(e ) = 5.8 kpc , in good agreement with the other measurements .
this source was also studied by wsy96 , who could not distinguish between elliptical and disk morphology .
they derived for the host m(r ) = 24.4 and r(e ) = 30.8 kpc .
while the host luminosity is in agreement , the scale length from wsy96 is clearly too large .
note that 1es 0414 + 009 is the dominant member of a moderately rich cluster at z = 0.287 ( abell class 0 ; mchardy 1992 ; falomo et al 1993 ) . * ms 0419.3 + 1943 * : wsy96 derived for the probably elliptical host m(r ) = 23.5 and r(e ) = 22.9 kpc , while u99b derived for the elliptical host m(r ) = 24.0 , and we identify the host as clearly an elliptical with m(r ) = 24.8 and r(e ) = 6.0 kpc .
* 1es 0502 + 675 * : the hst observations of this bl lac at z = 0.341 have been discussed by scarpa et al ( 1999a ) .
it is a gravitational lens candidate with a double source of similar magnitude with a separation of only @xmath603 .
scarpa et al ( 1999a ) found the host galaxy has m(host ) = 23.9 . using the not data we derive for the host galaxy m(host ) = 24.6 .
the difference is likely due to the fact that our image is much deeper than the one obtained with hst reaching @xmath24 = 26.5 at 6 arcsec from the center compared with @xmath24 = 24 at 2 arcsec from hst image . *
ms 0607.9 + 7108 * : for this bl lac at z = 0.267 , the morphology of its host galaxy has been controversial , mainly due to the presence of a bright nearby star .
wsy96 preferred a disk model ( although they could not rule out elliptical host ) and derived m(r ) = 24.8 and r(e ) = 8.9 kpc , while hst ( u99b ) indicated an elliptical host with m(host ) = 24.3 .
because of the presence of the close bright star we have first subtracted a scaled psf from the image removing the contribution from the star and then extracted the brightness profile from the bl lac .
after this correction the surrounding nebulosity turned out to be well represented by a an elliptical model similarly to the rest of the objects .
* 1es 0806 + 524 * : the hst observations of this bl lac at tentative z = 0.136 ( bade et al .
1998 ) have been discussed by scarpa et al ( 1999a ) .
noteworthy is an arc like structure @xmath62@xmath64 s of the nucleus . using the not data we derive for the host galaxy m(host ) = 24.2 to be compared with m(host ) = 23.5 reported by scarpa et al ( 1999a ) .
the difference can be due in part to poor photometry for this target and to the fact that not image is much deeper ( @xmath24 = 26.5 ) than hst image ( @xmath24 = 24.5 ) and allow us to detect the host up to 15 arcsec from the center . *
ms 0922.9 + 7459 * : wsy96 derived for the marginally resolved host m(r ) = 22.6 , while we clearly resolve the elliptical host with m(r ) = 24.7 and r(e ) = 9.7 kpc .
the large difference is likely attributable to the difference of seeing ( 0.7 versus 1.3 arcsec ) and therefore the ability to distinguish starlight from nuclear source .
this high redshift ( z = 0.638 ) source , which lies behind the rich cluster of galaxies abell 786 ( z = 0.124 ) and may itself be located in a moderately rich cluster ( wurtz et al .
1997 ) , was also resolved by hst ( u99b ) , with m(r ) = 24.6 , in good agreement with our result . *
1es 1011 + 496 * : our derived magnitude for the elliptical host galaxy , m(r ) = 23.7 , is in good agreement with that derived by the hst ( u99b ) , m(r ) = 23.6 .
the redshift of this object is uncertain , being based on the possible membership of the bl lac to the cluster abell 950 at z=0.20 ( wisniewski et al .
some galaxies are indeed detected in the field of view . *
1es 1028 + 511 * : our derived magnitude for the elliptical host galaxy , m(r ) = 23.2 , is significantly fainter than that derived by the hst ( u99b ) , m(r ) = 24.1 . again we believe the difference is attributable to different surface brightness limits of the images .
note that a reliable redshift of z = 0.361 based on caii h&k absorption lines was recently reported by polomski et al .
( 1997 ) , which is considerably larger than the z = 0.239 previously used for this target . *
1es 1218 + 304 * : this source has a recently determined redshift of z = 0.182 ( bade et al .
wsy96 derived for the host m(r ) = 16.6 , while we derive m(r ) = 17.6 .
hst derived for the absolute host magnitude m(r ) = 23.6 , in good agreement with our result , m(r ) = 23.8 . *
ms 1221.8 + 2452 * : u99b derived for the host m(r ) = 22.5 and r@xmath30 = 4.0 kpc , wsy96 derived m(r ) = 22.7 and r@xmath30 = 2.6 kpc , and jannuzi et al ( 1997 ) derived m(r ) = 22.8 and r@xmath30 = 3.2 kpc .
we derive in this study m(r ) = 22.7 and r@xmath30 = 2.9 kpc , in good agreement with the previous studies . *
ms 1229.2 + 6430 * : wsy96 derived for the host m(r ) = 24.1 and r(e ) = 10.6 kpc , and hst ( u99b ) derived m(r ) = 24.1 while we derive m(r ) = 24.3 and r(e ) = 11.0 kpc , in good agreement with the previous determinations .
the elliptical host looks quite symmetric despite the presence of a companion galaxy located 3.4 arcsec sw .
* 1es 1255 + 244 * : heidt et al ( 1999 ) derived for the host m(r ) = 23.2 and r(e ) = 7.2 kpc , and hst ( u99b ) derived m(r ) = 23.3 , while we derive m(r ) = 23.4 and r(e ) = 6.4 kpc , in good agreement with the previous determinations .
this bl lac seems to be embedded in a small cluster of galaxies ( heidt et al .
1999 ) .
* ms 1407.9 + 5954 * : u99b using hst derived for the host m(r ) = 24.8 and r@xmath30 = 11.1 kpc , wsy96 derived m(r ) = 24.3 and r@xmath30 = 9.7 kpc , while jannuzi et al ( 1997 ) derived m(r ) = 24.0 and r@xmath30 = 12.2 kpc .
we derive in this study m(r ) = 23.9 and r@xmath30 = 8.0 kpc , in good agreement with the previous studies .
this bl lac is the bcg in a moderately rich cluster ( wurtz et al .
1993 , 1997 ) .
* ms 1443.5 + 6349 * : wsy96 derived for the host which they classified as a probable disk , m(r ) = 23.6 and r@xmath30 = 11.3 kpc .
we clearly classify the host as an elliptical , with m(r ) = 23.9 and r@xmath30 = 17.2 kpc .
this source is surrounded by close companions . *
ms 1458.8 + 2249 * : wsy96 derived for the host m(r ) = 23.6 and r@xmath30 = 7.9 kpc , and hst ( u99b ) derived m(r ) = 23.7 , while we derive m(r ) = 24.3 and r@xmath30 = 2.3 kpc , somewhat brighter host .
the model fits are hampered by the presence of bright nearby stars . *
ms 1757.7 + 7034 * : wsy96 derived for the host m(r ) = 23.3 and r@xmath30 = 6.1 kpc , and hst ( u99b ) derived m(r ) = 23.6 , while we derive m(r ) = 23.4 and r@xmath30 = 2.2 kpc , in good agreement with the previous studies .
* 1es 1853 + 671 * : heidt et al ( 1999 ) derived for the host m(r ) = 22.9 and r@xmath30 = 9.4 kpc , and hst ( u99b ) derived m(r ) = 23.2 , while we derive m(r ) = 22.7 and r@xmath30 = 11.6 kpc , in reasonably good agreement .
this bl lac belongs to a very poor group of galaxies , however , it has a close companion 2@xmath64 to the nw ( heidt et al .
1999 ) .
* 1es 1959 + 650 * : the hst observations of this bl lac at a tentative redshift z = 0.048 have been discussed by scarpa et al ( 1999a ) .
it is hosted by a gas rich elliptical galaxy with a prominent dust lane .
they are able to clearly resolve the host galaxy whose luminosity is m(host ) = 22.5 .
this bl lac was also studied by heidt et al ( 1999 ) , who derived for the host galaxy m(host ) = 14.8 , m(host ) = 23.0 and r@xmath30 = 12.5 kpc .
we derive for the host galaxy m(host ) = 16.1 , m(host ) = 22.7 and r@xmath30 = 5.3 kpc .
all these determinations are in good agreement with each other , except for the scale length .
note that the absolute host luminosity is in the faintest end of the distribution for the xbls , suggesting that its distance could be larger than z = 0.048 .
* 1es 2037 + 521 * : heidt et al ( 1999 ) derive for the host m(r ) = 23.2 and r@xmath30 = 12.3 kpc , while we derive m(r ) = 23.7 and r@xmath30 = 7.8 kpc , in reasonable agreement .
hst ( u99b ) derived m(r ) = 16.1 , in good agreement with the value found here , m(r ) = 16.3 .
a small companion galaxy is visible 0.6 arcsec away from the nucleus , apparently well inside the host . *
ms 2143.4 + 0704 * : kotilainen et al .
( 1998b ) derived for the host m(h ) = 25.9 and r@xmath30 = 5.5 kpc , while we find in this study m(r ) = 23.6 and r@xmath30 = 10.9 kpc .
the scale length is in reasonable agreement .
the color of the host is @xmath0@xmath65 = 2.3 , in agreement with the average found for low redshift bl lacs , @xmath0@xmath65 = [email protected] ( kotilainen et al .
urry et al . ( 1999a ) and u99b derived for the host m(r ) = 23.7 , m(i ) = 24.1 and r@xmath30 = 8.8 kpc , wsy96 derived m(r ) = 23.8 and r@xmath30 = 11.6 kpc , while jannuzi et al ( 1997 ) derived m(r ) = 23.3 and r@xmath30 = 9,0 kpc , in good agreement with our result , m(r ) = 23.6 .
* 1es 2326 + 174 * : heidt et al ( 1999 ) derive for the host m(r ) = 23.4 and r@xmath30 = 8.5 kpc , and hst ( u99b ) derived m(r ) = 23.7 , while we derive m(r ) = 24.2 and r@xmath30 = 7.5 kpc , slightly brighter host .
part of the 0.8 mag difference with heidt et al is probably due to different extinction and k - correction applied since the difference of observed galaxy mag is just 0.5 mag .
three faint galaxies , possibly companions , are superimposed onto the outer parts of the host at 36@xmath64 distance from the nucleus ( heidt et al .
1999 ) .
this work was partly supported by the italian ministry for university and research ( murst ) under grant cofin98 - 02 - 32 .
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 .
abraham , r.g . ,
mchardy , i.m . , crawford , c.s .
1991 , mnras 252 , 482 + antonucci , r .
1993 , ara&a 31 , 473 + antonucci , r. ulvestad , j.s . 1985 ,
294,158 + bade , n . , beckmann , v . , douglas , n.g . 1998 , a&a 334 , 459 + bahcall , j.n . ,
kirhakos , s . ,
saxe , d.h . , schneider , d.p .
1997 , apj 479 , 642 + blandford , r.d . ,
rees , m.j .
1978 . in wolfe a.n .
pittsburgh conference on bl lac objects .
university of pittsburgh press , p. 328
+ bressan , a . ,
chiosi , c . ,
fagotto , f .
1994 , apjs 94 , 63 + browne , i. w. a. 1983 mnras , 204 , 23 + browne , i.w.a . , marcha , m.j.m .
1993 , mnras 261 , 795 + cassaro , p. , stanghellini , c. , bondi , m. , dallacasa , d. della ceca , r. and zappala , r.a .
1999 , a&a , in press + capaccioli , m . ,
caon , n . , donofrio , m .
1992 mnras 259 , 323 + celotti , a . ,
maraschi , l . ,
ghisellini , g . ,
caccianiga , a . , maccacaro , t .
1993 , apj 416 , 118 + coleman , g.d . ,
wu , c.c . , weedman , d.w .
1980 , apjs 43 , 393 + djorgovski , s . ,
davis , m .
1987 , apj 313 , 59 + ellingson , e . ,
yee , h.k.c . , green , r.f .
1991 , apj 371 , 49 + elvis , m . ,
plummer , d . ,
schachter , j . , fabbiano , g .
1992 , apjs 80 , 257 ( e92 ) + falomo , r. , pesce , j. e. , treves , a. 1993 , apj 411 , l63 + falomo , r . ,
pesce , j.e . , treves , a .
1995 , apj 438 , l9 + falomo , r .
1996 , mnras , 283 , 241 + falomo , r . ,
kotilainen , j . ,
pursimo , t . , et al .
1997 , a&a 321 , 374 + fanaroff , b . , riley , j.m .
1974 , mnras 167 , 31p + fukugita , m . ,
shimasaku , k . ,
ichikawa , t .
1995 , pasp 107 , 945 + gioia , i.m . ,
maccacaro , t . ,
schild , r.e . , et al .
1990 , apjs 72 , 567 + govoni , f. , falomo , r , fasano , g. and scarpa , r. 1999 , a&a in press + heckman , t.m .
1990 , paired and interacting galaxies ( ed .
j.w.sulentic , w.c.keel , c.m.telesco ) , p. 359
, nasa conf .
3098 + heidt , j . , nilsson , k .
, fried , j.w . , takalo , l.o . ,
sillanpaa , a . 1999 ,
a&a in press + heidt , j . , nilsson , k . , sillanpaa , a . ,
takalo , l.o . ,
pursimo , t .
1999 , a&a 341 , 683 + hoessel , j.g . ,
gunn , j.e .
, thuan , t.x .
1980 , apj 241 , 486 + hutchings , j.b . , neff , s.g .
1992 , aj 104 , 1 + hutchings , j.b . , holtzman , j . , sparks , w.b .
1994 , apj 429 , l1 + jannuzi , b.t . , yanny , b . ,
impey , c .
1997 , apj 491 , 146 + kollgaard , r.i . ,
wardle , j.f.c . ,
roberts , d.h . ,
gabuzda , d.c .
1992 , aj 104 , 1687 + kotilainen , j.k . ,
falomo , r . ,
scarpa , r .
1998a , a&a 332 , 503 + kotilainen , j.k . , falomo , r . ,
scarpa , r .
1998b , a&a 336 , 479 + lamer , g. , newsam , a.m. , mchardy , i.m .
1999 , mnras , in press + lamer , g. , mchardy , i.m .
newsam , a.m. 1999 . in l.o.takalo ,
a.sillanp ( eds ) bl lac phenomenon .
159 , 381 + landolt , a.u .
1992 , aj 104 , 340 + lehnert , m.d . ,
heckman , t.m . , chambers , k.c . , miley , g.k .
1992 , apj 393 , 68 + maccacaro , t . ,
wolter , a .
, mclean , b . , et al .
1994 , astroph .
29 , 267 + magorrian , j . ,
tremaine , s . ,
richstone , d . , et al .
1998 , aj 115 , 2285 + mchardy , i.m . ,
abraham , r.g . , crawford , c.s .
1991 , mnras 249 , 742 + mchardy , i.m . ,
luppino , g.a . ,
george , i.m . ,
abraham , r.j . , cooke , b.a .
1992 , mnras 256 , 655 + mclure , r.j . ,
dunlop , j.s . ,
kukula , m.j .
et al . 1999 ,
apj in press + mobasher , b . , sharples , r.m . , ellis , r.s .
1993 , mnras 263 , 560 + morris , s.l . ,
stocke , j.t . ,
gioia , i.m . , et al .
1991 , apj 380 , 49 ( m91 ) + murphy d.w . ,
browne w.a .
, perleyr.a .
1993 , mnras 264 , 298 + ostriker , j.p . , vietri , m .
1990 , nat 344 , 45 + owen , f.n . , laing , r.a .
1989 , mnras 238 , 357 + owen , f.n . ,
ledlow , m.j . ,
keel , w.c .
1996 , aj 111 , 53 + padovani , p .
1992 , mnras 257 , 404 + padovani , p . , urry , c.m .
1990 , apj 356 , 75 + padovani , p . ,
giommi , p .
1995 , mnras 277 , 1477 + perlman , e.s . , stocke , j.t .
1993 , apj 406 , 430 + perlman , e.s . ,
stocke , j.t . ,
schachter , j.f . , et al .
1996 , apjs 104 , 251 ( p96 ) + pesce , j.e . ,
falomo , r . , treves , a .
1995 aj 110 , 1554 + polomski , e . , vennes , s . , thorstensen , j.p . ,
mathioudakis , m . , falco , e.e .
1997 , apj 486 , 179 + rnnback , j . ,
van groningen , e .
, wanders , i . , rndahl , e .
1996 , mnras 283 , 282 + scarpa , r . ,
urry , c.m . ,
falomo , r . ,
et al . 1999a , apj 521 , 134 + scarpa , r .
, et al . 1999b , apjs in press + shull , j.m . ,
van steenberg , m.e .
1985 , apj 294 , 599 + stanghellini , c. , dallacasa , d. , bondi , m. , and della ceca , r .
1997 , a&a 325,911 + stark , a.a . ,
gammie , c.f . ,
wilson , r.w .
, et al . 1992 , apjs 79 , 77 + stocke , j.t . ,
morris , s.l . ,
gioia , i.m . , et al .
1991 , apjs 76 , 813 + stocke , j.t . ,
wurtz , r.e . ,
wang , q.d .
, elston , r. , januzzi , b.t .
1992 , ap.j , 400 , l17 + stocke , j.t . ,
wurtz , r.e . , perlman , e.s .
1995 , apj 454 , 55 + thuan , t.x . , romanishin , w .
1981 , apj 248 , 439 + urry , c.m . ,
padovani , p .
1995 , pasp 107 , 803 + urry , c.m . ,
falomo , r . ,
scarpa , r .
et al . 1999a , apj 512 , 88 + urry et al 1999b , apj , in press ( u99b ) + wisniewski , w.z . ,
sitko , m.l .
sitko , a.k .
1986 mnras 219 , 299 + wardle , j.f.c . ,
moore , r.l .
, and angel , j.r.p .
1984 , ap.j 279 , 93 + wurtz , r . ,
ellingson , e .
, stocke , j.t . , yee , h.k.c .
1993 , aj 106 , 869 + wurtz , r . , stocke , j.t . ,
yee , h.k.c .
1996 , apjs 103 , 109 ( wsy96 ) + wurtz , r . ,
stocke , j.t . , ellingson , e . ,
yee , h.k.c .
1997 , apj 480 , 547
|
we investigate the properties of the host galaxies of x ray selected ( high frequency peaked ) bl lac objects using a large and homogeneous data set of high spatial resolution @xmath0band observations of 52 bl lacs in the emss and slew samples .
the redshift distribution of the bl lacs ranges from z = 0.04 to [email protected] , with average and median redshifts z = 0.26 and z = 0.24 , respectively .
eight objects are at unknown redshift .
we are able to resolve 45 objects out of the 52 bl lacs .
for all the well resolved sources , we find the host to be a luminous elliptical galaxy . in a few cases a disk
is not ruled out but an elliptical model is still preferred .
the average absolute magnitude of the host galaxies is @xmath2 = [email protected] , while the average scale length of the host is @xmath4r(e)@xmath1 = 9@xmath35 kpc .
there is no difference in the host properties between the emss and slew samples .
we find a good agreement between the results derived by the surveys of wurtz et al .
( ground - based data ) and urry et al .
( hst data ) , and by our new deeper imaging .
the average luminosity of the bl lac hosts is between those of f - r i and f - r ii radio galaxies in govoni et al . ,
supporting the idea that both radio galaxy types could contribute to the parent population .
the bl lac hosts follow the f - p relation for giant ellipticals and exhibit a modest luminosity evolution with redshift .
finally , we find a slight correlation between the nuclear and host luminosity and a bimodal distribution in the nuclear / host luminosity ratio .
|
this section contains the proofs of our three theorems .
our proofs are constructive in nature , based on a procedure that constructs pair of matrices @xmath27 and @xmath28 .
the goal of the construction is to show that matrix @xmath29 is an optimal primal solution to the convex program , and that the matrix @xmath30 is a corresponding dual - optimal solution , meaning that it belongs to the sub - differential of the @xmath31-norm ( see lemma [ lemsubdiff ] ) , evaluated at @xmath29 .
if the construction succeeds , then the pair @xmath32 acts as a witness for the success of the convex program in recovering the correct signed support in particular , success of the primal - dual witness procedure implies that @xmath29 is the unique optimal solution of the convex program , with its row support contained with @xmath33 . to be clear
, the procedure for constructing this candidate primal - dual solution is _ not _ a practical algorithm ( as it exploits knowledge of the true support sets ) , but rather a proof technique for certifying the correctness of the block - regularized program .
we begin by providing some background on the sub - differential of the @xmath4 norm ; we refer the reader to the books @xcite for more background on convex analysis .
the sub - differential of a convex function @xmath34 at a point @xmath35 is the set of all vectors @xmath36 such that @xmath37 for all @xmath38 .
see the standard references @xcite for background on subdifferentials and their properties .
we state for future reference a characterization of the sub - differential of the @xmath4 block norm : [ lemsubdiff ] the matrix @xmath39 belongs to the sub - differential @xmath40 if and only if the following conditions hold for each @xmath41 . 1 .
if @xmath42 for at least one index @xmath43 , then @xmath44 where @xmath45 such that @xmath46 .
2 . if @xmath47 for all @xmath48 , then we require @xmath49 .
we now describe our method for constructing the matrix pair @xmath50 .
recalling that denotes the union of supports of the true regression vectors , let @xmath51 denote the complement of @xmath52
. with this notation , figure [ figwitness ] provides the four steps of the primal - dual witness construction .
the following lemma summarizes the utility of the primal - dual witness method : [ lemkey ] suppose that for each @xmath53 , the @xmath54 sub - matrix @xmath55 is invertible . then for any @xmath56 , we have the following correspondences : 1 . if steps ( a ) through ( c ) of the primal - dual construction succeed , then @xmath57 is the unique optimal solution of the original convex program .
conversely , suppose that there is a solution @xmath58 to the convex program with support contained within @xmath33
. then steps ( a ) through ( c ) of the primal - dual witness construction succeed .
we provide the proof of lemma [ lemkey ] in appendix [ applemkey ] .
it is convex - analytic in nature , based on exploiting the subgradient optimality conditions associated with both the restricted convex program and the original program , and performing some algebra to characterize when the convex program recovers the correct signed support .
lemma [ lemkey ] lies at the heart of all three of our theorems .
in particular , the positive results of theorem [ thmdetdesign ] , theorem [ thmgauss ] and theorem [ thmphase](a ) are based on claims ( i ) and ( iii ) , which show that it is sufficient to verify that the primal - dual witness construction succeeds with high probability .
the negative result of theorem [ thmphase](b ) , in contrast , is based on part ( ii ) , which can be restated as asserting that if the primal - dual witness construction fails , then no solution has support contained with @xmath33 . before proceeding to the proofs themselves ,
we introduce some additional notation and develop some auxiliary results concerning the primal - dual witness procedure , to be used in subsequent development .
with reference to steps ( a ) and ( b ) , we show in appendix [ applemkey ] that unique solution @xmath59 has the form @xmath60 where the matrix @xmath61 has columns @xmath62 , \qquad \mbox{for $ { \ensuremath{i}}= 1 , \ldots , { \ensuremath{r}}$,}\end{aligned}\ ] ] and @xmath63 is the @xmath64 column of the sub - gradient matrix @xmath65 . with reference to step ( c ) ,
we obtain the candidate dual solution @xmath66 as follows . for each @xmath53 , let @xmath67 denote the orthogonal projection onto the range of @xmath68 . using the sub - matrix @xmath69 obtained from step ( b ) ,
we define column @xmath70 of the matrix @xmath71 as follows : @xmath72 see the end of appendix [ applemkey ] for derivation of this condition .
finally , in order to further simplify notation in our proofs , for each @xmath73 , we define the random variable @xmath74 with this notation , the strict dual feasibility condition is equivalent to the event
we begin by establishing a set of sufficient conditions for deterministic design matrices , as stated in theorem [ thmdetdesign ] .
we begin by obtaining control on the probability of the event @xmath75 , so as to show that step ( c ) of the primal - dual witness construction succeeds . recall that @xmath67 denotes the orthogonal projection onto the range space of @xmath68 , and the definition of the incoherence parameter @xmath76 $ ] . by the mutual incoherence condition
, we have @xmath77 where we have used the fact that @xmath78 for each @xmath79 . recalling that and using the definition
, we have by triangle inequality @xmath80 \ ; \leq \ ; { { \ensuremath{\mathbb{p}}}}[{\ensuremath{\mathbb{a}}}({\ensuremath{\gamma}})],\ ] ] where we have defined the event @xmath81 to analyze this remaining probability , for each index @xmath82 and @xmath73 , define the random variable @xmath83 since the elements of the @xmath84-vector @xmath85 follow a @xmath86 distribution , the variable @xmath87 is zero - mean gaussian with variance @xmath88 . since @xmath89 by assumption and @xmath90 is an orthogonal projection matrix , the variance of each @xmath87 is upper bounded by @xmath91 .
consequently , for any choice of sign vector @xmath92 , the variance of the zero - mean gaussian @xmath93 is upper bounded by @xmath94 .
consequently , by taking the union bound over all sign vectors and over indices @xmath73 , we have @xmath95 \ ; = \ ; { \ensuremath{\mathbb{p}}}\big[\max_{{\ensuremath{k}}\in { \ensuremath{{u^{c } } } } } \max_{{\ensuremath{b}}\in \{-1,+1\}^{\ensuremath{r } } } \sum_{{\ensuremath{i}}=1}^{\ensuremath{r}}{\ensuremath{b}}_{\ensuremath{i}}{\ensuremath{w^{{\ensuremath{i}}}_{{\ensuremath{k } } } } } > { \ensuremath{\gamma}}\big ] & \leq & 2 \ , \exp \big ( -\frac{{\lambda_n}^2 { \ensuremath{n}}{\ensuremath{\gamma}}^2}{4 { \ensuremath{r}}\sigma^2 } + { \ensuremath{r}}+ \log { \ensuremath{p}}\big).\end{aligned}\ ] ] with the choice @xmath96 for some @xmath97 , we conclude that @xmath98 & \geq & 1 - 2 \exp(-({\ensuremath{\xi}}-1 ) [ { \ensuremath{r}}+ \log { \ensuremath{p } } ] ) \ ; \rightarrow \ ; 1.\end{aligned}\ ] ] by lemma [ lemkey](i ) , this event implies the uniqueness of the solution @xmath99 , and moreover the inclusion of the supports @xmath100 , as claimed .
we now turn to establishing the claimed @xmath5-bound on the difference @xmath101 .
we have already shown that this difference is exactly zero for rows in @xmath51 ; it remains to analyze the difference @xmath102 .
it suffices to prove the @xmath5 bound for the columns @xmath103 separately , for each @xmath104 .
we split the analysis of the random variable @xmath105 from equation , one involving the dual variables @xmath63 , and the other involving the observation noise @xmath85 , as follows : @xmath106 the second term is easy to control : from the characterization of the subdifferential ( lemma [ lemsubdiff ] ) , we have @xmath107 , so that @xmath108 . turning to the first term @xmath109
, we note that since @xmath68 is fixed , the @xmath110-dimensional random vector @xmath111 is zero - mean gaussian , with covariance @xmath112 .
therefore , we have @xmath113 , and can use this in standard gaussian tail bounds . by applying the union bound twice , first over @xmath114 , and then over @xmath115 , we obtain @xmath116 & \leq & 2 \exp(-t^2 { \ensuremath{n}}{\ensuremath{c_{\min}}}/(2 ) + \log ( { \ensuremath{r}}{\ensuremath{s } } ) + \log { \ensuremath{r}}),\end{aligned}\ ] ] where we have used the fact that @xmath117 . setting @xmath118 yields that @xmath119 with probability greater than @xmath120 , as claimed .
finally , to establish support recovery , recall that we proved above that @xmath121 is bounded by @xmath122 .
hence , as long as @xmath123 , then we are guaranteed that if @xmath124 , then @xmath125 .
we now turn to the proof of theorem [ thmgauss ] , providing sufficient conditions for general gaussian ensembles .
recall that for @xmath126 , each @xmath127 is a random design matrix , with rows drawn i.i.d . from a zero - mean gaussian with @xmath128 covariance matrix @xmath129 .
recalling that and using the definition , we have the decomposition @xmath130 in order to show that @xmath131 with high probability , we deal with each of these two terms in turn , showing that @xmath132 , and @xmath133 , both with high probability . in order to bound @xmath134
, we require the following condition on the columns of the design matrices : [ lemcolcontrol ] let @xmath135 . for @xmath136 , each column of the design matrices
@xmath137 has controlled @xmath138-norm : @xmath139 & \leq & 2 \exp \big ( -\frac{{\ensuremath{n}}}{2 } + \log ( { \ensuremath{p}}{\ensuremath{r } } ) \big ) \ ; \rightarrow 0.\end{aligned}\ ] ] this claim follows immediately by union bound and concentration results for @xmath140-variates ; in particular , the bound in appendix [ applargedev ] . under the condition of lemma [ lemcolcontrol ] , each variable @xmath141 is zero - gaussian , with variance at most @xmath142 .
consequently , for any choice of signs @xmath92 , the vector @xmath93 is zero - mean gaussian , with variance at most @xmath143 .
therefore , for any @xmath144 , we have @xmath145 & = & { { \ensuremath{\mathbb{p}}}}[\max_{{\ensuremath{k}}\in { \ensuremath{{u^{c } } } } } \max_{{\ensuremath{b}}\in \{-1,+1\}^{\ensuremath{r } } } \sum_{{\ensuremath{i}}=1}^{\ensuremath{r}}{\ensuremath{b}}_{\ensuremath{i}}{\ensuremath{w^{{\ensuremath{i}}}_{{\ensuremath{k } } } } } \geq t ] \\ & \leq & 2 \ ; \exp \big ( -\frac{{\lambda_n}^2 { \ensuremath{n}}}{4 \sigma^2 { \ensuremath{r } } } t^2 + { \ensuremath{r}}+ \log { \ensuremath{p}}\big)\end{aligned}\ ] ] setting @xmath146 yields that @xmath147 & \leq & 2 \ ; \exp \big ( -\frac{{\lambda_n}^2 { \ensuremath{n}}}{16 \sigma^2 { \ensuremath{r } } } { \ensuremath{\gamma}}^2 + { \ensuremath{r}}+ \log { \ensuremath{p}}\big).\end{aligned}\ ] ] [ lempopmutinco ] suppose that the design covariance matrices @xmath148 satisfy the mutual incoherence condition
. then we have @xmath149 where each random vector @xmath150 has i.i.d .
@xmath151 entries , and is independent of @xmath85 and @xmath68 .
see appendix [ applempopmutinco ] for the proof of this claim .
+ it remains to show that the random variable @xmath152 defined in equation is upper bounded by @xmath153 with high probability .
conditioning on @xmath68 and @xmath85 , the scalar random variable @xmath154 is zero - mean gaussian , with variance upper bounded as @xmath155 recalling that @xmath78 , for any choice of signs @xmath92 , the variable @xmath156 is zero - mean gaussian , with variance at most @xmath157 .
therefore , we have @xmath158 & \leq & { { \ensuremath{\mathbb{p}}}}\big[\max_{{\ensuremath{k}}\in { \ensuremath{{u^{c } } } } } \max_{{\ensuremath{b}}\in \{-1,+1 \}^{\ensuremath{r } } } { \ensuremath{\big\langle { \ensuremath{y^{{\ensuremath{i}}}_{{\ensuremath{k } } } } } , \ , { \ensuremath{x^{{\ensuremath{i}}}_{{\ensuremath{u } } } } } ( { \ensuremath{\langle \frac{1}{{\ensuremath{n } } } { \ensuremath{x^{{\ensuremath{i}}}_{{\ensuremath{u } } } } } , \ , { \ensuremath{x^{{\ensuremath{i}}}_{{\ensuremath{u } } } } } \rangle}})^{-1 } { \ensuremath{\widetilde{z}^{\,{\ensuremath{i}}}_{{\ensuremath{u } } } } } \big\rangle}}| \geq { \ensuremath{\gamma}}/2 \big ] \\ & \leq & 2 \exp ( -\frac{{\ensuremath{c_{\min}}}{\ensuremath{n}}}{8 { \ensuremath{r}}{\ensuremath{s } } } { \ensuremath{\gamma}}^2 + { \ensuremath{r}}+ \log { \ensuremath{p}}).\end{aligned}\ ] ] this probability vanishes faster than @xmath159 , as long as @xmath160 we now turn to establishing the claimed @xmath5-bound on the difference @xmath101 . as in the analogous portion of the proof of theorem [ thmdetdesign ] , we use the decomposition @xmath161 in the setting of random design matrices , a bit more work is required to control these terms .
beginning with the second term , by triangle inequality , we have @xmath162 \|_\infty + \big \| \big({\ensuremath{\sigma^{{\ensuremath{i}}}_{{\ensuremath{u}}{\ensuremath{u}}}}})^{-1 } { \lambda_n}{\ensuremath{\widetilde{z}^{\,{\ensuremath{i}}}_{{\ensuremath{u } } } } } \big \|_\infty \\ & \leq & { |\!|\!| \big[\big ( { \ensuremath{\langle \frac{1}{{\ensuremath{n } } } { \ensuremath{x^{{\ensuremath{i}}}_{{\ensuremath{u } } } } } , \ , { \ensuremath{x^{{\ensuremath{i}}}_{{\ensuremath{u } } } } } \rangle } } \big ) ^{-1 } - ( { \ensuremath{\sigma^{{\ensuremath{i}}}_{{\ensuremath{u}}{\ensuremath{u}}}}})^{-1 } \big ) | \ ! | \!|_{{2 } } } { \lambda_n}\sqrt{{\ensuremath{s } } } + { \ensuremath{d_{\max}}}{\lambda_n}\end{aligned}\ ] ] where we have used the facts that @xmath163 , since @xmath63 belongs to the sub - differential of the block @xmath4 norm ( see lemma [ lemsubdiff ] ) so that @xmath164 for all @xmath79 . by , concentration bounds for eigenvalues of gaussian random matrices ( see equation in appendix [ applargedev ] ) ,
we conclude that @xmath165.\end{aligned}\ ] ] now consider the first term @xmath109 : if we condition on @xmath68 , then the @xmath110-dimensional random vector @xmath166 is zero - mean gaussian , with covariance @xmath167 . by concentration bounds for eigenvalues of gaussian random matrices ( see equation in appendix [ applargedev ] )
, we have @xmath168 since @xmath169 .
therefore , we have shown that the variance of each element of @xmath170 is upper bounded by @xmath171 , so that we can apply standard gaussian tail bounds . by applying the union bound twice , first over @xmath114 , and then over @xmath172
, we obtain @xmath116 & \leq & 2 \exp(-t^2 { \ensuremath{n}}{\ensuremath{c_{\min}}}/(50 ) + \log |{\ensuremath{u}}| + \log { \ensuremath{r}}).\end{aligned}\ ] ] setting @xmath173 yields that @xmath116 & \leq & 2 \exp \big \ { -2 { \ensuremath{\xi}}^2 \log ( { \ensuremath{r}}{\ensuremath{s } } ) + \log ( { \ensuremath{r}}{\ensuremath{s } } ) + \log { \ensuremath{r}}\big \ } \\ & \leq & 2 \exp \big \ { -2 ( { \ensuremath{\xi}}^2 -1)\log ( { \ensuremath{r}}{\ensuremath{s } } ) \big \},\end{aligned}\ ] ] where we have used the fact that @xmath117 . combining the pieces ,
we conclude that @xmath174,\end{aligned}\ ] ] with probability greater than @xmath175 as claimed .
we now turn to the proof of the phase transition predicted by theorem [ thmphase ] , which applies to random design matrices @xmath176 and @xmath177 drawn from the standard gaussian ensemble .
this proof requires significantly more technical work than the preceding two proofs , since we need to control all the constants exactly , and to establish both necessary and sufficient conditions on the sample size .
we begin with the achievability result .
our proof parallels that of theorems [ thmdetdesign ] and [ thmgauss ] , in that we first establish strict dual feasibility , and then turn to proving @xmath5 bounds and exact support recovery . recalling that @xmath178
, we have @xmath179 where the random variables @xmath134 and @xmath180 were defined at the start of section [ subsecgaussrand ] . in order to prove that @xmath131 with high probability for the values of @xmath84 , @xmath181 , and @xmath1 , we will first establish that @xmath182 and @xmath183 for an appropriately chosen value of @xmath184 . by the results from the previous section
, we have @xmath185 with probability @xmath186 & \leq & 2 \ ; \exp \big ( -\frac{{\lambda_n}^2 { \ensuremath{n}}\epsilon^2}{32 \sigma^2 } + 2 + \log { \ensuremath{p}}\big)\end{aligned}\ ] ] recall that @xmath187 and that @xmath188 is independent of @xmath68 and @xmath85 .
we will show that @xmath183 with high probability by using results on gaussian extrema .
conditioning on @xmath189 , the random variable @xmath190 is zero - mean with variance upper - bounded as @xmath191 under the given conditioning , the random variables @xmath192 and @xmath193 are independent and for any sign vector @xmath194 , the random variable @xmath195 is gaussian , zero - mean with variance upper bounded as @xmath196 by lemma [ lemrandmat ] , @xmath197 with probability at least @xmath198 for sufficiently large @xmath199 and @xmath6 under the given scaling for each @xmath70 .
hence , @xmath195 is normal , zero - mean , with variance upper bounded as @xmath200 recall that @xmath65 was obtained from step ( b ) of the prima - dual witness construction .
the next lemma provides control over @xmath201 .
[ lemvarbound ] under the assumptions of theorem [ thmphase ] and corollary [ gapcor ] , if @xmath202 and @xmath203 , then @xmath204 is concentrated : for all @xmath11 , we have that for sufficiently large @xmath181 and @xmath84 @xmath205 & \rightarrow & 0 , \qquad \mbox{and } \\
\label{eqnvarupper } { { \ensuremath{\mathbb{p}}}}\big [ \|{{\ensuremath{\widetilde{z}^{\,1}}}}_{\ensuremath{u}}\|^2_2 + \|{{\ensuremath{\widetilde{z}^{\,2}}}}_{\ensuremath{u}}\|^2_2 \geq ( 1+\delta ) \frac{s}{2 } \big \{(4 - 3 { \alpha } ) + \frac{1}{{\lambda_n}^2 { \ensuremath{s } } } { \| { \ensuremath{{\ensuremath{\widebar{b}}}_{\operatorname{diff}}}}\|}_2 ^ 2 \big \ } \big ] & \leq & c_1 \exp ( -c_2 { \ensuremath{n}}),\end{aligned}\ ] ] see appendix [ applemvarbound ] for the proof of this claim . now , by applying the union bound and using gaussian tail bounds , we obtain that the probability @xmath206 $ ] is upper bounded by @xmath207 + \log({\ensuremath{p}}- ( 2-{\alpha } ) { \ensuremath{s } } ) \big),\ ] ] which goes to @xmath14 as @xmath208 under the condition @xmath209/(1-\epsilon)^2 \log({\ensuremath{p}}- ( 2-{\alpha } ) { \ensuremath{s}}).\end{aligned}\ ] ] we now turn to the proof of the converse claim in theorem [ thmphase ] .
we establish the claim by contradiction .
we show that if a solution @xmath99 exists such that @xmath210 , then under the stated upper bound on the sample size @xmath84 , there exists some @xmath211 such that @xmath212 $ ] converges to one . from the definition
, we see that conditioned on @xmath213 , the variables @xmath214 are i.i.d .
zero - mean gaussians , with variance given by @xmath215 by orthogonality , we have @xmath216 , so that ( using the idempotency of projection operators ) , we have @xmath217 note that @xmath218 is a scalar random variable , but fixed under the conditioning . turning to the variables @xmath219 , a similar argument shows that have @xmath220 , where @xmath221 is the analogous random variable . for @xmath73 ,
let @xmath222 and @xmath223 .
we then have @xmath224 & \stackrel{(a)}{\geq } & { \ensuremath{\mathbb{p}}}[\max_{{\ensuremath{k}}\in { \ensuremath{{u^{c } } } } } { |{\ensuremath{\widetilde{z}^{\,1}_{{\ensuremath{k}}}}}| } + { |{\ensuremath{\widetilde{z}^{\,2}_{{\ensuremath{k}}}}}| } > 1 + \epsilon ] \\ & \stackrel{}{\geq } & { \ensuremath{\mathbb{p}}}[\max_{{\ensuremath{k}}\in { \ensuremath{{u^{c } } } } } ( { \ensuremath{\widetilde{z}^{\,1}_{{\ensuremath{k } } } } } + { \ensuremath{\widetilde{z}^{\,2}_{{\ensuremath{k } } } } } ) > 1+\epsilon ] \\ & \stackrel{(b)}{= } & { \ensuremath{\mathbb{p}}}[\max_{{\ensuremath{k}}\in { \ensuremath{{u^{c } } } } } z_{\ensuremath{k } } > 1 + \epsilon],\end{aligned}\ ] ] where @xmath225 .
here inequality ( a ) follows because @xmath226 and @xmath227 are lower bounds on the variances of @xmath228 and @xmath229 respectively , and equality ( b ) follows since @xmath230 and @xmath231 are independent zero - mean gaussians with variances @xmath226 and @xmath227 , respectively . to simplify notation ,
let @xmath232 . by standard results for gaussian maxima @xcite , for any @xmath11
, there exists an integer @xmath233 such that for all @xmath234 , @xmath235 & \geq & ( 1-\delta ) \sqrt{2 ( { \ensuremath{\sigma^2}}+ { \ensuremath{\widetilde{\sigma}^2 } } ) \log n}.\end{aligned}\ ] ] moreover , the maximum function is lipschitz , so that by gaussian concentration for lipschitz functions @xcite , for any @xmath236 , we have @xmath237 -\eta \big ] & \leq & \exp \big(-\frac{\eta^2}{2 ( { \ensuremath{\sigma^2}}+ { \ensuremath{\widetilde{\sigma}^2 } } ) } \big).\end{aligned}\ ] ] combining these two statements yields that for all @xmath234 , we have @xmath238 & \leq & \exp \big(-\frac{\eta^2}{2 ( { \ensuremath{\sigma^2}}+ { \ensuremath{\widetilde{\sigma}^2 } } ) } \big).\end{aligned}\ ] ] it remains to show that there exists some @xmath211 such that converges to zero .
+ * case 1 : * first suppose that @xmath239 . in this case
, we have @xmath240 . with probability greater than @xmath241 ,
this quantity is lower bounded by a constant , using concentration for @xmath140-variates . in this case , @xmath242 w.h.p .
, so that the result follows trivially .
+ * case 2 : * otherwise , we must have @xmath202 . under this condition , we now establish a lower bound on @xmath226 that holds with high probability ; it will be seen that a similar lower bound holds for @xmath227 .
we begin by noting the lower bound @xmath243 . to control the minimum eigenvalue , define the event @xmath244 by standard random matrix concentration arguments ( see appendix [ applargedev ] ) , for some fixed @xmath245 , we are guaranteed that consequently , conditioned on @xmath246 , we have @xmath247 from lemma [ lemvarbound ] , we note that if @xmath248 , then for any @xmath11 , we have the lower bound @xmath249 the following result is the final step in the proof of theorem [ thmphase](b ) . [ lemcaseone ] suppose that @xmath202 . under this condition : 1 . if @xmath250 , then @xmath251 \rightarrow 0 $ ] .
2 . if @xmath252 , then there exists some @xmath211 such that @xmath253 \rightarrow 0 $ ] .
\(a ) if @xmath254 is bounded below by some constant @xmath255 , then we have @xmath256 which implies that @xmath257 .
thus , setting @xmath258 and @xmath259 in equation yields that ( for @xmath260 sufficiently large ) : @xmath261 & = & { \ensuremath{\mathbb{p}}}\big[\max_{k \in { \ensuremath{{u^{c } } } } } z_k \leq \frac{1}{4 } \ ; \sqrt{2 ( { \ensuremath{\sigma^2}}+ { \ensuremath{\widetilde{\sigma}^2 } } ) \log n } \big ] \\ & \leq & \exp
\big(-\frac{\log n}{4 } \big ) \rightarrow 0.\end{aligned}\ ] ] since @xmath262 for @xmath260 large enough , the claim follows .
\(b ) in this case , we may apply the lower bound , so that , for any @xmath11 , we have @xmath263 with high probability .
since @xmath264 { \ensuremath{s}}\log n$ ] by assumption , we have @xmath265 consequently , from equation , for any @xmath236 and @xmath11 , we have for all @xmath234 , @xmath266 & \leq & \exp \big(-\frac{\eta^2}{2 ( { \ensuremath{\sigma^2}}+ { \ensuremath{\widetilde{\sigma}^2 } } ) } \big).\end{aligned}\ ] ] since @xmath267 , we may choose @xmath268 sufficiently small so that for sufficiently large choices of @xmath269 , we have @xmath270 for some @xmath211 . since from lemma [ lemvarbound ] , the condition @xmath271 implies that @xmath272 w.h.p , we thus conclude that , using these choices of @xmath273 and @xmath274 , we have @xmath275 & \leq & o(1 ) + \exp \big(-\frac{\eta^2}{2 ( { \ensuremath{\sigma^2}}+ { \ensuremath{\widetilde{\sigma}^2 } } ) } \big ) \ ; \rightarrow \ ; 0,\end{aligned}\ ] ] as claimed .
in this paper , we provided a number of theoretical results that provide a sharp characterization of when , and if so by how much the use of block @xmath4 regularization actually leads improvements in statistical efficiency in the problem of multivariate regression . as suggested in a body of past work , the use of block @xmath4 regularization is well - motivated in many application contexts .
however , since it involves greater computational cost than more naive approaches , the question of whether this greater computational price yields statistical gains is an important one .
this paper assessed statistical efficiency in terms of the number of samples required to recover the support exactly ; however , one could imagine studying the same issue for related loss functions ( e.g. , @xmath138-loss or prediction loss ) , and it would be interesting to see if the results were qualitatively similar or not .
our results demonstrate that some care needs to be exercised in the application of @xmath4 regularization .
indeed , it can yield improved statistical efficiency when the regression matrix exhibits structured sparsity , with high overlaps among the sets of active coefficients within each column .
however , our analysis shows that these improvements are quite sensitive to the exact structure of the regression matrix , and how well it aligns with the regularizing norm .
when this alignment is not high enough , then the use of @xmath4 can actually impair performance relative to more naive ( and less computationally intensive ) schemes based on @xmath276-regularization , such as the lasso . moreover , whether or not the @xmath4 yields statistical improvements is very sensitive to the actual magnitudes of the different regression problems . in comparison to related results obtained by obozinski et al .
@xcite on block @xmath277 regularization , the block @xmath4 exhibits some fragility , in that the conditions under which it actually improves statistical efficiency are delicate and easily violated .
an interesting open direction is study whether or not it is possible to develop computationally efficient methods that are _ fully adaptive _ to the sparsity overlap namely , methods that behave like ordinary @xmath276-regularization when there is no or little shared sparsity , and behave like block regularization schemes in the presence of shared sparsity .
in this appendix , we discuss some issues associated with recovering individual signed supports .
we begin by observing that once the support union @xmath33 has been recovered , one can restrict the regression problem to this subset @xmath33 , and then apply lasso to each problem separately ( with substantially lower cost , since each problem is now low - dimensional ) in order to recover the individual signed supports .
if one is not willing to perform some extra computation in this way , then the the interpretation of theorems [ thmdetdesign ] and [ thmgauss]in terms of recovering the individual signed supports requires a more delicate treatment , which we discuss in this appendix .
interestingly , the structure of the block @xmath4 norm permits two ways in which to recover the individual signed supports .
[ [ ell_1ell_infty - primal - recovery ] ] @xmath4 primal recovery : + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + solve the block - regularized program , thereby obtaining a ( primal ) optimal solution @xmath278 . estimate the support union via @xmath279 , and and estimate the signed support vectors via @xmath280_{\ensuremath{k } } & { \ensuremath { : \ , = } } & { \operatorname{sign}}({\ensuremath{\widehat{\beta}^{\,{\ensuremath{i}}}}}_{\ensuremath{k}}).\end{aligned}\ ] ] [ [ ell_1ell_infty - dual - recovery ] ] @xmath4 dual recovery : + +
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + solve the block - regularized program , thereby obtaining an primal solution @xmath278 . for each row @xmath41 ,
compute the set @xmath281 . estimate the support union via @xmath282 , and estimate the signed support vectors @xmath283
& = & \begin{cases } { \operatorname{sign}}({\ensuremath{\widehat{\beta}^{\,{\ensuremath{i}}}}}_{\ensuremath{k } } ) & \mbox{if $ { \ensuremath{i}}\in { \ensuremath{\mathbb{m}}}_{\ensuremath{k}}$ } \\ 0 & \mbox{otherwise . }
\end{cases}\end{aligned}\ ] ] the procedure corresponds to estimating the signed support on the basis of a dual optimal solution associated with the optimal primal solution .
the dual signed support recovery method is more conservative in estimating the individual support sets .
in particular , for any given @xmath43 , it only allows an index @xmath284 to enter the signed support estimate @xmath285 when @xmath286 achieves the maximum magnitude ( possibly non - unique ) across all indices @xmath82 .
consequently , unlike the primal estimator , a corollary of theorem [ thmdetdesign ] guarantees that the dual signed support method never suffers from false inclusions in the signed support set . on the other hand , unlike the primal estimator , it may incorrectly exclude indices of some supports that is , it may exhibit false exclusions . to provide a concrete illustration of this distinction ,
suppose that @xmath287 and @xmath288 , and that the true matrix @xmath289 and estimate take the following form : @xmath290 consistent with the claims of theorem [ thmdetdesign ] , the estimate @xmath99 correctly recovers the support union viz . @xmath291 .
the primal and dual methods return the following estimates of the individual signed supports : @xmath292 consequently , the primal estimate includes false non - zeros in positions @xmath293 and @xmath294 , whereas the dual estimate includes false zeros in positions @xmath295 and @xmath296 .
we note that it is possible to ensure that under some conditions that the dual support method will correctly recover each of the individual signed supports , without any incorrect exclusions .
however , as illustrated by theorem [ thmphase ] and corollary [ gapcor ] , doing so requires additional assumptions on the size of the gap @xmath297 for indices @xmath298 .
note that conditioned @xmath299 , the rows of the random matrix @xmath188 are i.i.d .
gaussian random vectors with mean @xmath300 and covariance @xmath301 @xmath302 where @xmath303 . using these expressions and triangle inequality
, we obtain that @xmath304 is upper bounded by @xmath305 applying the mutual incoherence assumption , we obtain @xmath306 as claimed .
recall that @xmath307 , @xmath308 , and that @xmath309 is the set where @xmath310 is concentrated . if @xmath311 , then the claim is trivial , so that we may assume that @xmath312 .
recall that @xmath313^{-1 } { \ensuremath{f}}^1 - { \ensuremath{m}}^1 \big[{\ensuremath{m}}^2 + { \ensuremath{m}}^1 \big]^{-1 } { \ensuremath{f}}^2 \big \ } + { \ensuremath{m}}^1 \big [ { \ensuremath{m}}^1 + { \ensuremath{m}}^2 \big]^{-1 } \vec{1 } - \\ \frac{1}{{\lambda_n } } [ ( { \ensuremath{m}}^1)^{-1 } + ( { \ensuremath{m}}^2)^{-1}]^{-1}{\ensuremath{{\ensuremath{\widebar{b}}}_{\operatorname{diff}}}}.\end{gathered}\ ] ] using @xmath314 to denote the spectral norm , we first claim that as long as @xmath203 , then the following events hold with probability greater than @xmath315 : [ allmatcontrol ] @xmath316^{-1 } - i/2 | \ !
| \!|_{{2 } } } & = & { \ensuremath{{\ensuremath{\mathcal{o}}}(\sqrt{{\ensuremath{s}}/{\ensuremath{n } } } ) } } , \qquad \mbox{and } \\
\label{eqnratcontrol } { |\!|\!| { \ensuremath{m}}^1 \big[{\ensuremath{m}}^1 + { \ensuremath{m}}^2 \big]^{-1 } - i/2 | \
! | \!|_{{2 } } } & = & { \ensuremath{{\ensuremath{\mathcal{o}}}(\sqrt{{\ensuremath{s}}/{\ensuremath{n } } } ) } } , \end{aligned}\ ] ] as well as the analogous events with @xmath317 and @xmath318 interchanged . to verify the bound , we first diagonalize the projection matrix .
all of its eigenvalues are @xmath14 or @xmath12 , and it has rank @xmath319 w.p .
one , so that we may write @xmath320 for some orthogonal matrix @xmath321 , and the diagonal matrix @xmath322 , @xmath323 but the projection @xmath324 is independent of @xmath325 , which implies that the random rotation matrix @xmath321 is independent of @xmath325 , and hence @xmath326 . since @xmath327 is diagonal with @xmath319 ones and @xmath181 zeros , @xmath328 , where @xmath329 is a standard gaussian random matrix .
consequently , we have @xmath330 since @xmath331 , and @xmath332 using concentration arguments for random matrices ( see lemma [ lemrandmat ] in appendix [ applargedev ] ) . for we may use the triangle inequality and the submultiplicativity of the norm so that @xmath333^{-1 } - i/2 | \ !
| \!|_{{2 } } } & = & { |\!|\!| [ { \ensuremath{m}}^{-1 } + { \ensuremath{\widetilde{{\ensuremath{m}}}}}^{-1}]^{-1}(i - [ { \ensuremath{m}}^{-1 } + { \ensuremath{\widetilde{{\ensuremath{m}}}}}^{-1}]/2 ) | \ ! | \!|_{{2 } } } \\ & \leq & { |\!|\!| [ { \ensuremath{m}}^{-1 } + { \ensuremath{\widetilde{{\ensuremath{m}}}}}^{-1}]^{-1 } | \ ! | \!|_{{2 } } } \ ; { |\!|\!| i - [ { \ensuremath{m}}^{-1 } + { \ensuremath{\widetilde{{\ensuremath{m}}}}}^{-1}]/2 | \ ! | \!|_{{2 } } } \\ & \leq & \frac{1}{2 } \big \ { { |\!|\!| i/2 - { \ensuremath{m}}^{-1}/2 | \ !
| \!|_{{2 } } } + { |\!|\!| i/2 - { \ensuremath{\widetilde{{\ensuremath{m}}}}}^{-1}/2 | \ !
| \!|_{{2 } } } \big \ } { |\!|\!| [ { \ensuremath{m}}^{-1 } + { \ensuremath{\widetilde{{\ensuremath{m}}}}}^{-1}]^{-1 } | \ ! | \!|_{{2 } } } \\ & = & { |\!|\!| [ { \ensuremath{m}}^{-1 } + { \ensuremath{\widetilde{{\ensuremath{m}}}}}^{-1}]^{-1 } | \ ! | \!|_{{2 } } } \ ; { \ensuremath{{\ensuremath{\mathcal{o}}}(\sqrt{{\ensuremath{s}}/{\ensuremath{n}}})}},\end{aligned}\ ] ] finally , since @xmath334^{-1 } | \ ! | \!|_{{2 } } } = { \ensuremath{\mathcal{o}}}(1)$ ] , equation is valid . in order to establish the bound
, we have @xmath335^{-1 } - i/2 | \ ! | \!|_{{2 } } } & = & { |\!|\!| ( { \ensuremath{m}}/2 - { \ensuremath{\widetilde{{\ensuremath{m}}}}}/2 ) [ { \ensuremath{m}}+ { \ensuremath{\widetilde{{\ensuremath{m}}}}}]^{-1 } | \ ! | \!|_{{2 } } } \\ & \leq & \frac{1}{2}\big \ { { |\!|\!| { \ensuremath{m}}- i | \ ! | \!|_{{2 } } } \ ; + { |\!|\!| { \ensuremath{\widetilde{{\ensuremath{m}}}}}-i | \ ! | \!|_{{2 } } } \big \ } \ ; { |\!|\!| [ { \ensuremath{m}}+ { \ensuremath{\widetilde{{\ensuremath{m}}}}}]^{-1 } | \ ! | \!|_{{2 } } } \\ & = & { |\!|\!| [ { \ensuremath{m}}+ { \ensuremath{\widetilde{{\ensuremath{m}}}}}]^{-1 } | \ ! | \!|_{{2 } } } \ ; { \ensuremath{{\ensuremath{\mathcal{o}}}(\sqrt{{\ensuremath{s}}/{\ensuremath{n}}})}}.\end{aligned}\ ] ] since @xmath336 - 2i | \ !
| \!|_{{2 } } } = { \ensuremath{{\ensuremath{\mathcal{o}}}(\sqrt{{\ensuremath{s}}/{\ensuremath{n}}})}}\rightarrow 0 $ ] , we have @xmath336^{-1 } | \ ! | \!|_{{2 } } } = { \ensuremath{\mathcal{o}}}(1)$ ] , which establishes the claim .
we are now ready to establish the claims of the lemma . from the representation ,
we apply triangle inequality and our bounds on spectral norms , thereby obtaining @xmath337 with probability greater than @xmath338 , where @xmath339 . by the decomposition of @xmath340 in equation and applying bounds @xmath341 \big \}\ ] ] since @xmath271 , in order to establish the upper bound it suffices to show that @xmath342 w.h.p . similarly , in the other direction , we have @xmath343 following the same line of reasoning , in order to prove the lower bound , it suffices to show that @xmath344 w.h.p .
since @xmath345 and @xmath346 behave similarly , it suffices to show that @xmath347 . from the definition , we see that conditioned on @xmath348 , the random vector @xmath349 is zero - mean gaussian , with i.i.d .
elements with variance @xmath350 recalling that @xmath351 , we have @xmath352 by random matrix concentration ( see the discussion following lemma [ lemrandmat ] in appendix [ applargedev ] ) , we have @xmath353 w.h.p . , and by @xmath140 tail bounds ( see lemma [ lemchitail ] in appendix [ applargedev ] )
, we have @xmath354 w.h.p .
consequently , with high probability , we have @xmath355 .
since the gaussian random vector @xmath349 has length @xmath356 , again by concentration for @xmath140 random variables , we have ( with probability greater than @xmath357 ) , @xmath358 . combining the pieces ,
we conclude that w.h.p .
@xmath359 \big ) \ ; = \ ; o({\lambda_n}^2 { \ensuremath{s}}),\end{aligned}\ ] ] where the final equality follows since @xmath271 and @xmath360 .
this section is devoted to the development of various properties of the optimal solution(s ) of the block @xmath4-regularized problem . by standard conditions for optimality in convex programs @xcite
, the zero - vector must belong to the subdifferential of the objective function in the convex program , or equivalently , we must have for each @xmath361 @xmath362 where @xmath39 must be an element of the subdifferential @xmath363 . substituting the relation @xmath364 , we obtain @xmath365 we begin with the proof of part ( i ) : suppose that steps ( a ) through ( c ) of the primal - witness construction succeed
. by definition , it outputs a primal pair , of the form @xmath366 , along with a candidate dual optimal solution @xmath367 .
note that the conditions defining the @xmath4 subdifferential apply in an elementwise manner , to each index @xmath368 . since the sub - vector @xmath369 was chosen from the subdifferential of the restricted optimal solution , it is dual feasible .
moreover , since the strict dual feasibility condition holds , the matrix @xmath370 constructed in step ( c ) is dual feasible for the zero - solution in the sub - block @xmath51 .
therefore , we conclude that @xmath366 is a primal optimal solution for the full block - regularized program .
it remains to establish uniqueness of this solution .
define the ball @xmath371 and observe that we have the variational representation @xmath372 where @xmath373 denotes the euclidean inner product . with this notation ,
the block - regularized program is equivalent to the saddle - point problem @xmath374 since this saddle - point problem is strictly feasible and convex - concave , it has a value .
moreover , given any dual optimal solution in particular , @xmath30 from the primal - dual construction
any optimal primal solution @xmath99 must satisfy the saddle point condition @xmath375 but this condition can only hold if @xmath376 , @xmath377 for any index @xmath378 such that @xmath379 .
therefore , any optimal primal solution must satisfy @xmath380 , so that solving the original program is equivalent to solving the restricted program .
lastly , if the matrices @xmath381 are invertible for each @xmath382 , then the restricted problem is strictly convex , and so has a unique solution , thereby completing the proof of lemma [ lemkey](i ) .
we now prove part ( ii ) of lemma [ lemkey ] .
suppose that we are given an estimate @xmath99 of the true parameters @xmath289 by solving the convex program such that @xmath383 .
since @xmath99 is an optimal solution to the convex program , the the optimality conditions of equation , must be satified .
we may rewrite those conditions as @xmath384 where @xmath385 . recalling that @xmath386
, we obtain @xmath387 again , by standard conditions for optimality in convex programs @xcite , the first of these two equations is exactly the condition that must be satisfied by an optimal solution of the restricted program .
however , we have already shown that the candidate solution @xmath388 satisfies this condition , so that it must also be an optimal solution of the convex program .
additionally , the value of @xmath65 that satisfies equation for each @xmath389 is an element of @xmath390 .
we have thus shown that steps ( b ) and ( c ) of the primal - witness construction succeed .
it remains to establish uniqueness in part ( a ) .
however , we note that @xmath391 is invertible for each @xmath70 . hence , for any solution @xmath99 such that @xmath380 , @xmath392\ ] ] is well - defined and unique , noting that @xmath393 .
thus , we have established the equality and that @xmath388 is unique .
therefore , @xmath99 gives solutions to steps ( a ) and ( b ) when solving the restricted convex program over the set @xmath33 .
finally , we derive the form of the dual solution @xmath394 , as a function of @xmath391 , @xmath369 , and @xmath395 . recall that
@xmath391 is invertible , @xmath369 is an element of the subdifferential of @xmath396 , and @xmath397 . from equation
, we have @xmath398 the claimed form of the dual solution follows by substituting equation into equation . in this section , we focus on the specific form of the dual variables @xmath399 .
our approach is to construct a candidate set of dual variables , and then show that they are valid .
we begin by defining the sets @xmath400 , corresponding to the intersection of the supports , and the set @xmath401 corresponding to elements in one ( but not both ) of the supports . for @xmath402 ,
we let @xmath403 is a diagonal matrix whose diagonal entries correspond to @xmath404 . in addition
, we define the vectors @xmath405 and matrices @xmath406 via @xmath407\\ { \ensuremath{m}}^{\ensuremath{i } } & { \mathrel{\mathop:}= } & \frac{1}{{\ensuremath{n } } } { \ensuremath{s}}^{\ensuremath{i}}{\ensuremath{\big\langle { \ensuremath{x^{{\ensuremath{i}}}_{{\ensuremath{b } } } } } , \ , ( i - { \pi}_{({\ensuremath{x^{{\ensuremath{i}}}_{{\ensuremath{{\ensuremath{{b^{c}}}}}}}}})}){\ensuremath{x^{{\ensuremath{i}}}_{{\ensuremath{b } } } } } \big\rangle } } { \ensuremath{s}}^{\ensuremath{i}}. \end{aligned}\ ] ] given these definitions , we have the following lemma : [ lemmaboth ] assume that @xmath7 , and that latexmath:[$|{\ensuremath{\widehat{\beta}^{\,1}}}_{\ensuremath{b}}| = @xmath386 , then the dual variable @xmath409 satisfies the relation @xmath410^{-1 } { \ensuremath{f}}^1 - { \ensuremath{m}}^1 \big[{\ensuremath{m}}^2 + { \ensuremath{m}}^1 \big]^{-1 } { \ensuremath{f}}^2 \big \ } + { \ensuremath{m}}^1 \big [ { \ensuremath{m}}^1 + { \ensuremath{m}}^2 \big]^{-1 } \vec{1 } - \\ \frac{1}{{\lambda_n } } [ ( { \ensuremath{m}}^1)^{-1 } + ( { \ensuremath{m}}^2)^{-1}]^{-1 } { \ensuremath{{\ensuremath{\widebar{b}}}_{\operatorname{diff}}}}\end{gathered}\ ] ] and @xmath411 , with analogous results holding for @xmath412 .
given these forms for @xmath413 and @xmath414 , it remains to show that the relation @xmath415 holds under the conditions of theorem [ thmphase](a ) .
intuitively , this condition should hold since under the conditions of theorem [ thmphase](a ) , the matrix @xmath416 is approximately the identity , and the vector @xmath417 is approaching @xmath14 .
finally , we expect that @xmath418 is very small , hence the final term is also very small .
therefore , on the set @xmath309 , both @xmath413 and @xmath419 are approximately equal to @xmath420 .
we formalize this rough intuition in the following lemma : [ lembound ] under the assumptions of theorem [ thmphase](a ) each of the following conditions hold for sufficiently large @xmath84 , @xmath181 , and @xmath1 with probability greater than @xmath421 : @xmath422^{-1}({\ensuremath{{\ensuremath{\widebar{b}}}_{\operatorname{diff } } } } ) \|}_{\infty } & \leq & \epsilon \\
\label{lembound2 } \| \frac{1}{{\lambda_n } } \big \ { { \ensuremath{m}}^2 \big[{\ensuremath{m}}^1 + { \ensuremath{m}}^2 \big]^{-1 } { \ensuremath{f}}^1 - { \ensuremath{m}}^1 \big[{\ensuremath{m}}^2 + { \ensuremath{m}}^1 \big]^{-1 } { \ensuremath{f}}^2 \big \ } \|_{\infty } & \leq & \epsilon \\ \label{lembound3 } \| { \ensuremath{m}}^1 \big [ { \ensuremath{m}}^1 + { \ensuremath{m}}^2 \big]^{-1 } \vec{1 } - \frac{1}{2 } \|_{\infty } & \leq & \frac{1}{2 } - 3 \epsilon .
\end{aligned}\ ] ] given lemmas [ lemmaboth ] and [ lembound ] , we can conclude that the definition for the dual variables on the support is valid .
the remaining subsections in this appendix are dedicated to verifying the above results : in particular , we prove lemma [ lemmaboth ] in appendix [ applemboth ] and lemma [ lembound ] in appendix [ applembound ] .
we now proceed to establish the validity of the closed form expressions for @xmath423 and @xmath424 . from equation
we have that @xmath425 + ( \frac{1}{{\ensuremath{n } } } { \ensuremath{\big\langle { \ensuremath{x^{{\ensuremath{i}}}_{{\ensuremath{{\ensuremath{{b^{c } } } } } } } } } , \ , { \ensuremath{x^{{\ensuremath{i}}}_{{\ensuremath{{\ensuremath{{b^{c } } } } } } } } } \big\rangle}})^{-1 } { \ensuremath{(x^{1}_{{\ensuremath{{\ensuremath{{b^{c}}}}}}})^t } } { \ensuremath{w^{1}}}\ ] ] substituting back into @xmath426 so that we obtain @xmath427 recall that by assumption that @xmath428 , and @xmath429 . subtracting @xmath430 and @xmath431 from equations and @xmath432 applying the fact that @xmath433 .
@xmath434 where @xmath435 . then solving for @xmath436 letting @xmath437 and substituting back into equation @xmath438 ^{-1 } { \lambda_n}{\ensuremath{\vec{1}}}- [ ( { \ensuremath{m}}^1)^{-1 } + ( { \ensuremath{m}}^2)^{-1}]^{-1}({\ensuremath{{\ensuremath{\widebar{b}}}_{\operatorname{diff } } } } ) \\ + { \ensuremath{m}}^2 \big [ { \ensuremath{m}}^1 + { \ensuremath{m}}^2 \big ] ^{-1 } { \ensuremath{f}}^1 - { \ensuremath{m}}^1 \big [ { \ensuremath{m}}^1 + { \ensuremath{m}}^2 \big ] ^{-1 } { \ensuremath{f}}^2.\end{gathered}\ ] ] the first term @xmath439^{-1 } \ , { \ensuremath{{\ensuremath{\widebar{b}}}_{\operatorname{diff}}}}$ ] can be decomposed as @xmath440^{-1}{\ensuremath{{\ensuremath{\widebar{b}}}_{\operatorname{diff } } } } & = & \underbrace{\frac{1}{{\lambda_n } } ( [ ( { \ensuremath{m}}^1)^{-1 } + ( { \ensuremath{m}}^2)^{-1}]^{-1 } - i/2 ) { \ensuremath{{\ensuremath{\widebar{b}}}_{\operatorname{diff}}}}}_{\mbox{\large $ { t}_1 $ } } + \underbrace{\frac{{\ensuremath{{\ensuremath{\widebar{b}}}_{\operatorname{diff}}}}}{2 { \lambda_n}}}_{\mbox{\large $ { t}_2$}}\end{aligned}\ ] ] under the assumptions of theorem [ thmphase](a ) , we have @xmath441 , hence , for @xmath181 large enough , @xmath442 . in order to bound @xmath443
, we note that with probability greater than @xmath241 , the spectral norm of @xmath444^{-1 } - i/2)$ ] is @xmath445 ( see the bound from appendix [ applemvarbound ] ) .
consequently , we may decompose @xmath446^{-1 } - i/2)$ ] as @xmath447 where @xmath448 and @xmath327 are independent and @xmath448 is distributed uniformly over all orthogonal matrices , and @xmath449 . using this decomposition
, the following lemma , proved in appendix [ appbiglemmaone ] , allows us to obtain the necessary control on the quantity @xmath450 : [ biglemma1 ] let @xmath451 be a matrix chosen uniformly at random from the space of all orthogonal matrices . consider a second random matrix @xmath452 , independent of @xmath453 .
if @xmath271 , then for any fixed vector @xmath454 and fixed @xmath211 , we have : 1 .
[ lem9:parta ] if @xmath455 , then @xmath456 & \leq & c_1 \exp \big ( -c_2 \epsilon^2 \frac{{\ensuremath{n}}}{{\ensuremath{s}}\|x\|_\infty^2 } + \log({\ensuremath{s } } ) \big ) .
\end{aligned}\ ] ] 2 .
[ lem9:partb ] if @xmath457 , then @xmath456 & \leq & c_1 \exp \big ( -c_2 \epsilon^2 \frac{{\ensuremath{n}}^2}{{\ensuremath{s}}^2 \|x\|_\infty^2 } + \log({\ensuremath{s } } ) \big ) .
\end{aligned}\ ] ] with reference to the problem of bounding @xmath458 , we may apply part ( a ) of this lemma with @xmath459 and @xmath460 to conclude that @xmath461 with high probability , thereby establishing the bound .
we now turn the proving the bound .
we begin by decomposing the terms involved in this equation as @xmath462^{-1 } { \ensuremath{f}}^1 & = & \frac{1}{{\lambda_n } } \biggr [ { \ensuremath{m}}^2 \big [ { \ensuremath{m}}^1 + { \ensuremath{m}}^2 \big]^{-1 } - \frac{i}{2 } \biggr ] { \ensuremath{f}}^1 + \frac{{\ensuremath{f}}^1}{2 { \lambda_n } } \\ \frac{1}{{\lambda_n } } { \ensuremath{m}}^1 \big[{\ensuremath{m}}^1 + { \ensuremath{m}}^2 \big]^{-1 } { \ensuremath{f}}^2 & = & \frac{1}{{\lambda_n } } \biggr [ { \ensuremath{m}}^1 \big [ { \ensuremath{m}}^1 + { \ensuremath{m}}^2 \big]^{-1 } - \frac{i}{2 } \biggr ] { \ensuremath{f}}^2 + \frac{{\ensuremath{f}}^2}{2 { \lambda_n}}\end{aligned}\ ] ] recalling the form of @xmath405 , conditioned on @xmath463 and @xmath85 , we have @xmath464 however , by lemmas [ lemchitail ] and [ lemrandmat ] ( see appendix [ applargedev ] ) , as well as the fact that @xmath465 , for @xmath84 and @xmath181 large enough , the variance term is bounded by @xmath466 with probability greater than @xmath467 . hence , by standard gaussian tail bounds , the inequalities @xmath468 and @xmath469 both hold with probability greater than @xmath470 . now to bound the first term in the decomposition we begin by diagonalizing @xmath471 .
note that @xmath448 is independent of @xmath176 and @xmath472 and by symmetry @xmath473 .
following some algebra , we find that @xmath474^{-1 } - \frac{i}{2 } \biggr ] { \ensuremath{f}}^1 = \frac{1}{{\lambda_n } } { \ensuremath{q}}^t \biggr [ d \big [ { \ensuremath{q}}{\ensuremath{m}}^1 { \ensuremath{q}}^t + d \big]^{-1 } - \frac{i}{2 } \biggr ] { \ensuremath{q}}{\ensuremath{f}}^1\ ] ] the random vector @xmath475 is independent of @xmath453 and @xmath476 is independent of @xmath453 by symmetry .
hence , the vector @xmath477 { \ensuremath{q}}\frac{1}{{\lambda_n } } { \ensuremath{f}}^1 $ ] is independent of @xmath453 . for
a given constant @xmath478 , let us define the event @xmath479 \big \}.\ ] ] we can then write @xmath480 & \leq & { \ensuremath{\mathbb{p}}}\big [ { \| { \ensuremath{q}}^t v \|}_{\infty } \geq \epsilon \ , \mid \ , { \ensuremath{\mathcal{s}}}\big ] + { \ensuremath{\mathbb{p}}}[{\ensuremath{\mathcal{s}}}^c].\end{aligned}\ ] ] note that we may consider the event that @xmath481 and @xmath482 = { \ensuremath{{\ensuremath{\mathcal{o}}}(\sqrt{{\ensuremath{s}}/{\ensuremath{n}}})}}$ ] . we claim that each of these events happens with high probability .
note that the former event occurs with high probability by lemma [ lemrandmat ] .
the latter event holds with high probability since , @xmath483 = [ 2d((d + q { \ensuremath{m}}^1 q^t)^{-1}-i/2 ) + d - i].\ ] ] and , both @xmath484 and @xmath485 by equation .
thus , the sum of the two random matrices is also @xmath445 .
recall the bound on the variance of each component of @xmath475 from equation and note that each component is independent . applying the concentration results from lemma [ lemchitail ] for @xmath486-squared random variables
yields that @xmath487 with high probability .
hence , under the above conditions @xmath488 { \ensuremath{q}}\frac{1}{{\lambda_n } } { \ensuremath{f}}^1\|_2 ^ 2 & \leq & { |\!|\!| \| \frac{1}{2 } [ 2 { \ensuremath{d}}({\ensuremath{d}}+ { \ensuremath{q}}{\ensuremath{m}}^1 { \ensuremath{q}}^t)^{-1 } - i ] | \ ! | \!|_{{2}}}^2 \| { \ensuremath{q}}\frac{1}{{\lambda_n } } { \ensuremath{f}}^1\|_2 ^ 2 \\ & \leq & c_3 ^ 2 \ ; \frac{{\ensuremath{s}}^2}{{\ensuremath{n } } } \big [ \frac{{\ensuremath{s}}}{{\ensuremath{n } } } + \frac{1}{{\lambda_n}^2 { \ensuremath{n } } } \big ] , \end{aligned}\ ] ] with high probabilty , which implies that @xmath489 holds with high probability as well .
therefore , it immediately follows then that @xmath490 \leq c_1 \exp(-c_2 { \ensuremath{s}})$ ] .
it remains to control the first term .
we do so using the following lemma , which is proved in appendix [ appminilemmaone ] : [ minilemma1 ] let @xmath491 be a matrix chosen uniformly at random from the space of orthogonal matrices .
let @xmath492 be a random vector independent of @xmath453 , such that @xmath493 with probability one .
then we have @xmath494 \ ; = \ ; o(1).\ ] ] we now apply this lemma to the random vector @xmath495 with @xmath496 , and @xmath497 .
note that @xmath498 from which the second claim in lemma [ lembound ] follows .
finally , we turn to proving the third claim in lemma [ lembound ] . following some algebra
, we obtain @xmath499^{-1 } \vec{1 } - \frac{1}{2 } \|_{\infty } \ , = \ , \frac{1}{4 } ( { \ensuremath{m}}^1 - { \ensuremath{m}}^2 ) { \ensuremath{\vec{1}}}+ \frac{1}{2 } ( { \ensuremath{m}}^1 - { \ensuremath{m}}^2)(i/2 - ( { \ensuremath{m}}^1 + { \ensuremath{m}}^2)^{-1 } ) { \ensuremath{\vec{1}}}.\ ] ] we diagonalize the matrix @xmath500 , where @xmath472 is diagonal . since the random matrix @xmath317 has a spherically symmetric distribution , the matrix @xmath453 has a uniform distribution over the space of orthogonal matrices and is independent of @xmath472 . using this decomposition , we can rewrite the second term in equation as @xmath501 where @xmath502 .
we note that @xmath503 is independent of @xmath453 , because @xmath472 and @xmath318 are independent of @xmath453 .
this independence follows from the spherical symmetry of @xmath318 and the fact that @xmath504 . defining the event
, we claim that @xmath505 \ ; \leq \ ; c_1 \exp(-c_2 { \ensuremath{n } } ) \ ; \rightarrow \ ; 0.\ ] ] in order to establish this claim , we note that sub - multiplicativity and triangle inequality imply that @xmath506 since @xmath507 with probability greater than @xmath508 , from the discussion following lemma [ lemrandmat ] .
similarly , from this same result , we have @xmath509 , so that the claim follows .
using the decomposition and the tail bound , we have @xmath510 & = & { { \ensuremath{\mathbb{p } } } } [ { \| { \ensuremath{q}}^t { \ensuremath{r}}{\ensuremath{q}}1 \|}_{\infty } \geq \epsilon \ ; \mid \ ; { \mathcal{t } } ] + { { \ensuremath{\mathbb{p}}}}[{\mathcal{t}}^c ] \\ & \leq & { \ensuremath{\mathcal{o}}}\big ( \ , \frac{1}{s } \ , \big ) + { \ensuremath{\mathcal{o}}}(\exp(-c(\epsilon ) { \ensuremath{n}})),\end{aligned}\ ] ] where lemma [ biglemma1 ] ( proved in appendix [ appbiglemmaone ] ) provides control on the first term in the inequality .
we provide the proof for part ( a ) of the lemma and note that part ( b ) is analogous . by union bound
, we have @xmath511 & \leq & { \ensuremath{s}}\ ; \max_{i=1 , \ldots , { \ensuremath{s } } } { { \ensuremath{\mathbb{p } } } } [ | e_i^t { \ensuremath{q}}^t { \ensuremath{a}}{\ensuremath{q}}x| \geq \epsilon].\end{aligned}\ ] ] we will derive a bound on the probability @xmath512 $ ] that holds for all @xmath513 . we write @xmath514 , where @xmath515 denotes the first column of @xmath453 , and @xmath516 denotes the weighted sum of the remaining @xmath517 columns of @xmath453 . since @xmath453 is orthogonal , the vector @xmath515 has unit norm @xmath518 , the vector @xmath519 is orthogonal to @xmath515 , and moreover @xmath520 .
owing to the bound on the spectral norm of @xmath452 , we have @xmath521 which is less than @xmath522 for @xmath269 sufficiently large , since @xmath271 .
we now turn to the second term .
note that conditioned on @xmath519 , the vector @xmath515 is uniformly distributed over an @xmath523-dimensional unit sphere , contained within the subspace orthogonal to @xmath519 . still conditioning on @xmath519 ,
consider the function @xmath524 .
for any pair of vectors @xmath525 on the unit sphere , we have @xmath526,\end{aligned}\ ] ] where @xmath527 is the geodesic distance . using the inequality @xmath528 , valid for @xmath529 $ ] , and the assumption @xmath530 , and taking square roots , we obtain @xmath531 so that @xmath532 is a lipschitz constant on the unit sphere ( with dimension @xmath533 ) with constant @xmath534 .
consequently , by levy s theorem @xcite , for any @xmath211 , we have @xmath535 & \leq & 2 \exp(- ( { \ensuremath{s}}-2 ) \ ; { \frac{{\ensuremath{n}}}{\|x\|_\infty^2 \ ; { \ensuremath{s}}({\ensuremath{s}}-1 ) } } \ ; \epsilon^2 ) \ ; \leq \ ; 2 \ , \exp \big ( -c_1 \ , \frac{{\ensuremath{n}}}{\|x\|_\infty^2 \ ; { \ensuremath{s } } } \epsilon^2 \big).\end{aligned}\ ] ] as a final side remark , we note that under the scaling of theorem [ thmphase](b ) , we have as @xmath208 , so that the probability in question vanishes . by union bound and symmetry of the distribution @xmath453 , for any @xmath536 , we have @xmath537 & \leq & { \ensuremath{m}}\ ; { \ensuremath{\mathbb{p}}}\big [ { |e_1^t { \ensuremath{q}}^t { \ensuremath{v}}| } \ , \geq \ , t \big ] \\ & = & { \ensuremath{m}}\ ; { \ensuremath{\mathbb{p}}}\big [ { |{\ensuremath{q}}_1^t { \ensuremath{v}}| } \ , \geq \ , t \big],\end{aligned}\ ] ] where @xmath538 is the first column of @xmath453 .
note that @xmath538 is a random vector distributed uniformly over the unit sphere @xmath539 in @xmath540 dimensions .
viewing the vector @xmath541 as fixed , consider the function @xmath542 defined over @xmath539 .
as in lemma [ biglemma1 ] , some calculation shows the lipschitz constant of @xmath543 over @xmath539 is at most applying levy s theorem @xcite , we conclude that for any @xmath211 , @xmath544 & \leq & 2 \exp \big ( -({\ensuremath{m}}- 1 ) \
, \frac{t^2 } { 2 \|v\|_2 ^ 2 } + \log { \ensuremath{m}}\big).\end{aligned}\ ] ] since @xmath493 by assumption , it suffices to set @xmath545 .
in this appendix , we state some known large deviation bounds for the gausssian variates , @xmath140-variates , as well as the eigenvalues of random matrices .
the following gaussian tail bound is standard : letting @xmath553 in equation , we have @xmath554 \ ; \geq \ ; { \ensuremath{\mathbb{p } } } [ \frac{x}{d } \geq ( 1 + t)^2],\end{aligned}\ ] ] thereby establishing . with the same choice of @xmath555 , equation implies the bound immediately .
[ lemrandmat ] let @xmath556 be a random matrix from the standard gaussian ensemble ( i.e. , @xmath557 , i.i.d ) .
then with probability greater than @xmath241 , for any @xmath11 , its eigenspectrum satisfies the bounds @xmath558 ^ 2 \ ; \leq \ ; \lambda_{\min } \big ( \frac{x^t x}{{\ensuremath{n } } } \big ) \leq
\lambda_{\max } \big ( \frac{x^t x}{{\ensuremath{n } } } \big ) \ ; \leq \ ; ( 1+\delta ) \ ; \big[1 + \sqrt{\frac{{\ensuremath{s}}}{{\ensuremath{n}}}}\big]^2.\ ] ] note that this lemma implies similar bounds for eigenvalues of the inverse : @xmath559 ^ 2}\ ; \leq \ ; \lambda_{\min } \big ( ( \frac{x^t x}{{\ensuremath{n}}})^{-1 } \big ) \leq \lambda_{\max } \big ( ( \frac{x^t x}{{\ensuremath{n}}})^{-1 } \big ) \ ; \leq \ ; \frac{1}{(1-\delta ) \ ; \big[1- \sqrt{\frac{{\ensuremath{s}}}{{\ensuremath{n}}}}\big]^2}.\ ] ] d. l. donoho and j. m. tanner . counting faces of randomly - projected polytopes when the projection radically lowers dimension .
technical report , stanford university , 2006 . submitted to journal of the ams .
e. p. simoncelli .
denoising of visual images in the wavelet domain . in p.
mller and b. vidakovic , editors , _ bayesian inference in wavelet based models _ , chapter 18 , pages 291308 .
springer - verlag , new york , june 1999 .
lecture notes in statistics , vol .
141 .
m. j. wainwright .
information - theoretic bounds for sparsity recovery in the high - dimensional and noisy setting .
technical report 725 , department of statistics , uc berkeley , january 2007 . posted as arxiv : math.st/0702301 ; presented at international symposium on information theory ,
june 2007 .
m. j. wainwright .
sharp thresholds for high - dimensional and noisy recovery of sparsity using using @xmath276-constrained quadratic programs .
. appeared as tech .
report 709 , department of statistics , uc berkeley .
may 2006 .
w. wang , m. j. wainwright , and k. ramchandran .
information - theoretic limits on sparse signal recovery : dense versus sparse measurement matrices
. technical report arxiv:0806.0604 , uc berkeley , june 2008 . presented at isit 2008 , toronto , canada .
zhao , g. rocha , and b. yu . grouped and hierarchical model selection through composite absolute penalties .
technical report , statistics department , uc berkeley , 2007 . to appear in annals of statistics .
|
given a collection of @xmath0 linear regression problems in @xmath1 dimensions , suppose that the regression coefficients share partially common supports .
this set - up suggests the use of @xmath2-regularized regression for joint estimation of the @xmath3 matrix of regression coefficients .
we analyze the high - dimensional scaling of @xmath4-regularized quadratic programming , considering both consistency rates in @xmath5-norm , and also how the minimal sample size @xmath6 required for performing variable selection grows as a function of the model dimension , sparsity , and overlap between the supports .
we begin by establishing bounds on the @xmath5-error as well sufficient conditions for exact variable selection for fixed design matrices , as well as designs drawn randomly from general gaussian matrices .
our second set of results applies to @xmath7 linear regression problems with standard gaussian designs whose supports overlap in a fraction @xmath8 $ ] of their entries : for this problem class , we prove that the @xmath2-regularized method undergoes a phase transition that is , a sharp change from failure to success characterized by the rescaled sample size @xmath9 .
more precisely , given sequences of problems specified by @xmath10 , for any @xmath11 , the probability of successfully recovering both supports converges to @xmath12 if @xmath13 , and converges to @xmath14 for problem sequences for which @xmath15 .
an implication of this threshold is that use of @xmath16-regularization yields improved statistical efficiency if the overlap parameter is large enough ( @xmath17 ) , but has _
worse _ statistical efficiency than a naive lasso - based approach for moderate to small overlap ( @xmath18 ) .
empirical simulations illustrate the close agreement between these theoretical predictions , and the actual behavior in practice .
these results indicate that some caution needs to be exercised in the application of @xmath4 block regularization : if the data does not match its structure closely enough , it can impair statistical performance relative to computationally less expensive schemes . * * [ cols="^ " , ] figure [ figreleff ] provides an alternative perspective on the data , where we have plotted how the sample size required by block regression changes as a function of the overlap parameter @xmath19 $ ] .
each set of data points plots a scaled form of the sample size required to hit @xmath20 success , for a range of overlaps , and the straight line @xmath21 that is predicted by theorem [ thmphase ] note the excellent agreement between the experimental results , for all three problem sizes for @xmath22 , and the full range of overlaps .
the line @xmath21 also characterizes the relative efficiency @xmath23 of block regularization versus the naive lasso - based method , as described in corollary [ corscaling ] . for overlaps
@xmath24 , this parameter @xmath23 drops below @xmath12 . on the other hand , for overlaps @xmath25 , we have @xmath26 , so that applying the joint optimization problem actually decreases statistical efficiency . intuitively , although there is still some fraction of overlap , the regularization is misleading , in that it tries to enforce a higher degree of shared sparsity than is actually present in the data .
|
the startrack population synthesis code was initially developed for the study of double compact object mergers in the context of gamma - ray burst ( grb ) progenitors ( belczynski , bulik & rudak 2002b ) and gravitational radiation inspiral sources ( belczynski , kalogera & bulik 2002c , hereafter bkb02 ) .
startrack has undergone major updates and revisions in the last few years . with this code
we are able to evolve isolated ( not dynamically interacting ) single stars and binaries for a wide range of initial conditions .
the input physics incorporates our latest knowledge of processes governing stellar evolution , while the most uncertain aspects are parameterized to allow for systematic error analysis . during the code development ,
special emphasis was placed on the compact object populations : white dwarfs ( wds ) , neutron stars ( nss ) , and black holes ( bhs ) .
the input physics currently includes all major processes important for the formation and evolution of compact objects . among other things
we have developed fast procedures to treat and diagnose various types of mass transfer episodes ( including phases of thermal timescale and dynamically unstable mass transfer leading to common envelopes ) .
we also compute tidal effects on orbital evolution , angular momentum losses due to magnetic braking and gravitational radiation , as well as mass loss from stellar winds and during mass transfer phases .
rejuvenation of binary components is taken into account .
the full orbital evolution of binaries is also computed , including angular momentum and mass loss .
supernovae ( sne ) and compact object formation are also treated in detail .
the new version of startrack presented here has already been tested and used in many applications .
belczynski & taam ( 2004a ) studied the formation of ultrashort period x - ray binaries and they also demonstrated that the faint x - ray galactic center population can neither be explained by quiescent ns / bh transients nor by hard / faint wind - fed sources ( belczynski & taam 2004b ) .
belczynski , sadowski & rasio ( 2004b ) and belczynski et al .
( 2006 ) developed a comprehensive description of young bh populations , which can also provide realistic initial conditions for the dynamical modeling of bhs in star clusters .
belczynski et al .
( 2004a ) derived for the first time a synthetic x - ray luminosity function which agrees with _ chandra _ observations of ngc 1659 , and sepinsky , kalogera , & belczynski ( 2005 ) explored the numbers and spatial distribution of x - ray binaries formed in young star clusters .
belczynski , bulik & ruiter ( 2005b ) tested different models of type ia sn progenitors , arriving at the conclusion that the double degenerate scenario most easily reproduces the observed delay times between star formation and type ia sne .
belczynski et al .
( 2005a ) used startrack to study the gravitational radiation signal from the galactic population of double wds .
nutzman et al .
( 2004 ) , oshaughnessy et al .
( 2005a , b , c ) , and ihm , kalogera , & belczynski ( 2005 ) studied binary compact object populations and derived merger rates and detection rates by ground - based interferometers ; they also examined bh spin magnitudes and studied the eccentricities of double neutron stars .
startrack was also incorporated into a simple stellar dynamics code , allowing the study of the effects of dynamical interactions on binary populations in dense star clusters . in that form
it has been used for the study of binary fractions in globular clusters ( ivanova et al .
2005 ) and an investigation of intermediate - mass bhs in clusters and their connection to ultra - luminous x - ray sources ( blecha et al .
. among other things startrack has been adapted for the study of accretion powered x - ray binaries ( xrbs ) . in forthcoming papers we will present the synthetic populations of xrbs formed in different stellar environments
. we will start with young starburst galaxies , and move on to spiral and , eventually , old elliptical galaxies . in the next stage it will be possible to compare the models with rapidly improving observations of various x - ray point source populations .
this will offer a new perspective to the study of several uncertain aspects of binary evolution leading to the formation of xrbs .
it may also result in an independent diagnostic of star formation rates for nearby galaxies , since both the numbers and properties of xrbs are directly connected to the star formation history ( see e.g. , grimm , gilfanov & sunyaev 2003 ; gilfanov 2004 ; kim & fabbiano 2004 ; belczynski et al .
2006 , in preparation ) . in this paper we provide a detailed description of the current version of startrack , and we present the results of a number of tests .
we describe the implementations of single star evolution in 2 , binary orbit evolution in 3 , stellar wind mass loss / accretion in 4 , roche lobe overflow calculations in 5 , spatial velocities in 6 , and the assumed distributions of initial parameters in 7 . in 8
we discuss the validity of various input physics assumptions , and we compare startrack calculations with detailed evolutionary models and with various observations . section 9
is dedicated to the discussion of x - ray binary modeling . in 10
we conclude with a short summary .
in all subsequent sections we use units of @xmath0 for mass , @xmath1 for orbital separations and stellar radii , myr for time , @xmath2 for bolometric luminosity , unless specified otherwise .
we use @xmath3 and @xmath4 to denote stellar radius and mass , while @xmath5 represent the binary orbital parameters : semi - major axis and eccentricity , respectively .
index @xmath6 is used to mark the binary components ( or single stars for consistency ) , or to denote an accretor and a donor in mass transfer calculations : @xmath7 .
roche lobe parameters are indexed with lob " .
the initially more massive ( at zero age main sequence ) binary component is referred to as primary , while its companion as secondary . the evolution of single stars and non - interacting binary components have remained mostly unchanged since the last published description of the code ( bkb02 ) and therefore we only give a brief outline here .
however , we do point out the new additions and reiterate the modifications to the original formulas which were used as the base for the implementation of single star evolution in startrack . to evolve single stars from the zero age main sequence ( zams ) until remnant formation ( wd , ns , bh , or a remnant - less supernova ) we employ the analytic formulas of hurley , pols & tout ( 2000 ) .
each star is followed along an evolutionary track specific for its initial mass and metallicity . various wind mass loss rates that vary with the stellar evolutionary stage
are incorporated into the code and their effect on stellar evolution is taken into account .
once the remnant is formed , we terminate the calculations but keep track of the numbers , properties and formation times of a given type of remnant .
additionally , for white dwarf remnants we take into account their subsequent luminosity evolution , and follow cooling tracks adopted from hurley et al .
( 2000 ) .
we follow hurley et al .
( 2000 ) to denote different stages of stellar evolution with an integer @xmath8 , where + 0 main sequence ( ms ) @xmath9 + 1 ms @xmath10 + 2 hertzsprung gap ( hg ) + 3 red giant branch ( rg ) + 4 core helium burning ( cheb ) + 5 early asymptotic giant branch ( eagb ) + 6 thermally pulsing agb ( tpagb ) + 7 helium main sequence ( hems ) + 8 helium hertzsprung gap ( hehg ) + 9 helium giant branch ( hegb ) + 10 helium white dwarf ( he wd ) + 11 carbon / oxygen white dwarf ( co wd ) + 12 oxygen / neon white dwarf ( one wd ) + 13 neutron star ( ns ) + 14 black hole ( bh ) + 15 massless remnant ( after sn ia explosion ) + 16 hydrogen white dwarf ( h wd ) + 17 hybrid white dwarf ( hyb wd ) + in addition to the star types introduced and coded by the numbers @xmath11 in the original hurley et al .
( 2000 ) formulas , we have introduced two new stellar types @xmath12 .
@xmath13 denotes a h - rich white dwarf
. only main sequence stars less massive than about @xmath14 can produce such a h - rich remnant through mass loss in a close binary system .
these low - mass stars do not process a significant amount of hydrogen into helium in their cores ( even in a hubble time ) and once their mass is stripped below the hydrogen burning limit ( close to @xmath15 ) they become degenerate h - rich white dwarfs .
these stars , although not frequently encountered in population synthesis , may become donors in the shortest - period interacting binaries .
@xmath16 denotes a hybrid white dwarf , with a carbon - oxygen - helium mixture in the core and a helium envelope .
these objects are the remnants of naked helium main sequence stars ( @xmath17 ) which are stripped of mass below @xmath18 during roche lobe overflow ( rlof ) . at that point ,
thermonuclear reactions stop and the star becomes degenerate ( eg . , savonije , de kool & van den heuvel 1986 ) .
several major changes to the original hurley et al .
( 2000 ) formulas have been implemented within startrack .
the remnant masses of neutron stars and black holes are calculated in a different way than originally suggested by hurley et al .
( 2000 ) . in the present version of the code
we have further revised our prescription presented in belczynski et al .
( 2002c ) to include the more recent calculations of feni core masses and allow for the possibility of ns formation through electron capture supernovae ( ecs ) .
white dwarfs masses are calculated with the original formulas of hurley et al .
( 2000 ) , although one wds are formed in a slightly narrower range since we allow for ecs ns formation ( see below ) .
we determine the mass of a ns / bh remnant using information on the final co and feni core masses , combined with the knowledge of the pre - supernova mass of the star . for a given initial zams mass
, the final co core mass is obtained from the original hurley et al .
( 2000 ) formulas , while we use the models of timmes , woosley & weaver ( 1996 ) to estimate final feni core mass .
the results of timmes et al .
( 1996 , see their fig.2 ) show two distinctive feni core masses ( we use models with the addition of si shell mass ) , below and above initial masses of @xmath19 .
the dichotomy arises from different carbon burning ( convective versus radiative ) the in pre - supernova stellar core . for higher mass progenitors ( @xmath20 )
there is a slow rise in a feni core mass that we approximate with a linear relation .
the final feni core mass for a given co core mass ( @xmath21 ) is obtained from : + @xmath22 + @xmath23 where all masses are expressed in @xmath24 .
the above dependence was obtained for solar metallicity models , but we use it for all compositions considered here ( @xmath25 ; the metallicity range covered by hurley et al .
2000 in fits to the stellar evolution models ) . timmes et al . ( 1996 ) results for metallicity are very similar for zero metallicity .
for example , in the mass range most important for ns formation the core mass changes from @xmath26 ( solar metallicity ) to @xmath27 ( zero metallicity ) .
the differences can be larger for bh progenitors , but then the exact core mass does not play such an important role on the final bh mass , as most bhs form through fall back or direct collapse ( see below ) .
the effects of material fallback ( ejected initially in the sn explosion ) during the star s final collapse are included . for the most massive stars we also allow for the possibility of a silent collapse ( no supernova explosion ) and direct bh formation . for solar metallicity and standard wind mass loss
the compact object masses are obtained from @xmath28 where m is the pre - supernova mass of the star , and @xmath29 is the _ fall - back _ factor , i.e. the fraction ( from 0 to 1 ) of the stellar envelope that falls back .
the value of @xmath29 is interpolated linearly between @xmath30 ( @xmath31 ) and @xmath32 ( @xmath33 ) .
the regimes of no fall - back ( @xmath34 ) , partial fall - back ( @xmath35 ) and direct collapse ( @xmath36 ) are estimated from core collapse models of fryer , woosley & hartmann ( 1999 ) and the analysis of fryer & kalogera ( 2001 ) .
we also allow for ns formation through ecs ( e.g. , podsiadlowski et al . 2004 ) . following hurley et al .
( 2000 ) we use the he core mass at the agb base to set the limits for the formation of various co cores . if the he core mass is smaller than @xmath37 the star forms a degenerate co core , and ends up forming a co wd .
if the core is more massive than @xmath38 the star forms a non - degenerate co core with subsequent burning of elements until the formation of a feni core which ultimately collapses to a ns or a bh . stars with cores between @xmath37 and @xmath39 may form partially degenerate co cores . if such a core reaches a critical mass ( @xmath40 , hurley et al .
2000 ) , it ignites co off - center and non - explosively burns co into one , forming a degenerate one core .
if in subsequent evolution the one core increases its mass to @xmath41 the core collapses due to electron capture on mg , and forms ns ( we will refer to a ns formed in this way as ecs ns , as opposed to a regular feni core collapse compact object formation ) . the ecs nss are assumed to have unique masses of @xmath42 . if the one core mass remains below @xmath43 the star forms a one wd .
hurley et al . ( 2000 ) suggested @xmath44 corresponding to @xmath45 for @xmath46 .
later calculations with the same , but updated evolutionary code ( eldridge & tout 2004a , b ) indicated that ecs may occur for higher initial masses ( @xmath47 ) . for our standard model
, we adopt @xmath48 ( @xmath49 ) , and this results in ecs ns formation above @xmath50 ( for masses @xmath51 the one core does not reach @xmath43 and a one wd is formed ) .
it is noted that binary evolution through rlof , may either decrease the initial the mass of the zams star required to form the core of mass @xmath37 ( due to rejuvenation ) or increase it ( due to mass loss ) .
therefore , binary evolution effectively leads to wider initial progenitor mass for ecs ns formation .
also metallicity , and wind mass loss influences the ecs ns formation range .
the remnant masses calculated above express the mass of baryons , and for ns / bh remnants we convert them to gravitational masses ( @xmath52 ) .
we use the quadratic relation @xmath53 for neutron stars ( lattimer & yahil 1989 ; see also timmes et al .
1996 ) , while for black holes we simply approximate the gravitational mass with @xmath54 the resulting remnant mass spectrum covers a wide range of masses and is presented in figure [ mfin ] . in summary ,
helium wds form from progenitors of initial masses in range @xmath55 , co wds from @xmath56 , and one wd are formed in range @xmath57 .
neutron stars formed through ecs ( @xmath58 )
have mass of @xmath59 . regular core - collapse ( of feni cores ) ns are formed with mass @xmath60 ( @xmath61 ) ; @xmath62 ( @xmath63 ) and for higher initial progenitor masses nss form up to adopted maximum ns mass : @xmath64 . in our standard model
we adopt @xmath65 , and above @xmath66 bhs are are formed .
it is found that single stars may form bhs up to @xmath67 for solar metallicity ( @xmath46 ) and @xmath68 for lower metallicities ( @xmath69 ) , which is consistent with the current observations of the most massive bhs in galactic x - ray transients .
the initial - final mass relation described above and presented in figure [ mfin ] holds only for single stellar evolution and for a specific metallicity ( @xmath46 ) .
effects of binary evolution , in particular mass loss / gain in rlof , may alter the initial - final mass relation in two ways .
first , a compact object may have a different mass ; higher if a progenitor ( or a compact object itself ) accreted mass or smaller if a progenitor lost mass in rlof .
second , the initial mass limits for formation of a given type of compact object are not sharp , since with binary mass loss / gain stars of various initial masses may form compact objects of a given mass and type .
these limits are blurred by binary evolution .
the mass estimates of neutron stars in relativistic double neutron star binaries point to a ns formation mass of @xmath70 ( thorsett & chakrabarty 1999 ) .
it is also suggestive that the ns mass in vela x-1 is high @xmath71 .
since vela x-1 is a high - mass x - ray binary , system with wind - fed accretion ( small mass capture efficiency ) and a massive star donor ( short lifetime ) , the ns probably has not accumulated much mass , and the measured mass is close to its formation mass . our adopted model for ns formation masses ( timmes et al .
1996 ) falls in qualitative agreement with these observations .
finally we allow compact object masses to increase through accretion in binary systems .
accretion and mass accumulation onto wds is described in detail in 5.7 . for ns
we need to adopt a maximum ns mass , over which ns collapses to bh .
such a collapse may lead to a short - hard gamma - ray burst event .
depending on the preferred equation of state the maximum ns mass may vary in a wide range ( @xmath72 ) , and in particular may reach @xmath73 if rotation is included ( morrison , baumgarte & shapiro 2004 ) . at the moment the highest measured ns mass is @xmath74 for a millisecond pulsar in psr j0751 + 1807 ; a relativistic binary with helium white dwarf secondary ( nice et .
al . 2005 ) .
as stated above we adopt @xmath65 for our standard model , but we relax this assumption in parameter studies . the compilation of stellar wind mass loss rates presented in hurley et al.(2000 ) has been expanded to include mass loss from low- and intermediate - mass main sequence stars .
we have adopted the formulas of nieuwenhuijzen & de jager ( 1990 ) to calculate the mass loss rates for main sequence stars below @xmath75 .
although the mass loss from these stars is not large enough to significantly alter the evolution of a mass losing star , it may play an important role in the formation and evolution of wind - accreting close binaries .
even with small mass transfer rates characteristic for the low- and intermediate - mass main sequence stars , the x - ray luminosities for accreting bhs and nss are high enough to be detected in deep chandra exposures .
a number of faint point x - ray sources were discovered in the galactic center with deep exposures ( wang , gotthelf & lang 2002 ; muno et al .
2003 ) , some of which may be explained in terms of wind - fed close binaries ( pfahl , rappaport & podsiadlowski 2002a ; bleach 2002 ; willems & kolb 2003 ; belczynski & taam 2004b ) . a compilation of updated observational data on rotational velocities is used to initiate the stellar spins on the zams .
the spin evolution is followed as detailed here for single stars and in 3 for binary components . in order to obtain a functional form of the relation of the equatorial rotational velocity and stellar mass
, we used the compilation of rotational velocities of stauffer & hartmann ( 1986 ) for stars in open clusters .
the difference between cluster and field stars is quite small for massive stars ( with a maximum difference of @xmath76 for intermediate b - type stars ) , but can be as high as 40% for stars later than f - type , with field stars having systematically lower rotational velocities .
the mean rotational velocity @xmath77 was determined from the projected velocity ( @xmath78 ) assuming a random distribution of angles with @xmath79 .
we fitted @xmath77 as a function of stellar mass , and we obtained the following empirical functional form @xmath80 where , @xmath81 , @xmath82 , @xmath83 and @xmath84 ( errors are at the @xmath85 level ) .
we stress that this is only an empirical functional form of the equatorial rotational velocity as a function of stellar mass . in fig [ fig03 ]
we present the observational data from stauffer & hartmann ( 1986 ) , together with the best fit function . in the bottom panel of this figure
we also show the ratio of the stauffer & hartmann ( 1986 ) data and the model .
the spin angular momentum of a star may be expressed as @xmath86 where , @xmath87 is the angular rotational velocity and the coefficient @xmath88 varies as the star evolves and its internal structure changes ( e.g. , it is 2/5 for a solid sphere and 2/3 for a spherical shell ) . following hurley et al .
( 2000 ) we consider two structural components for each star : a core and an envelope .
the spins of these two components may decouple in the course of evolution , although we keep them coupled in our standard model calculations .
the spin angular momentum of a star is then @xmath89 \omega_{\rm i } \label{sin02}\ ] ] we use different values than hurley at al .
( 2000 ) for the internal structure coefficient @xmath88 . for stars with no clear core - envelope structure ( @xmath90 ) we use simple polytropic models ( e.g. , lai , rasio & shapiro 1993 ) with @xmath91 and @xmath92 for low - mass and high - mass objects , respectively , giving @xmath93 for giants with a clear separation between core and envelope ( @xmath94 ) and additionally for stars in the hertzsprung gap ( @xmath95 ) we use detailed models of giant envelopes ( hurley et al .
2000 ) and for the core we apply a polytropic model with @xmath91 to obtain @xmath96 conservation of the spin angular momentum of a star is used then to determine its rotational velocity .
additional angular momentum losses from magnetic braking ( see 3.2 ) are also taken into account .
stars with convective and radiative envelopes respond differently to various physical processes ( e.g. , magnetic braking , tidal interactions or mass loss ) .
stars that have a significant convective envelope are : low - mass h - rich ms stars ( @xmath97 ) within the mass range of @xmath98 , where @xmath99 is the maximum mass for a ms star to develop a convective envelope ; giant - like stars ( @xmath100 ) independent of their mass and evolved low - mass helium stars ( @xmath101 ) below @xmath102 . for stars crossing h - rich hertzsprung gap ( @xmath103 ) and core helium burning stars ( @xmath104 we obtained detailed models using the code described in ivanova & taam ( 2004 ) to check how far from hayashi line stars cross the border between radiative and convective envelopes ; stars cooler than @xmath105 have convective envelopes , while hotter stars have radiative envelopes .
we have also examined the models presented by schaller et al . ( 1992 ) for both solar metallicity and @xmath106 and have found that the above temperature cut works rather well for these two classes of stars for both metallicities .
ms stars with masses below @xmath107 are fully convective .
the value of @xmath99 ) depends strongly on metallicity @xmath108 values of @xmath99 in metallicity range @xmath109 are obtained from a fit to detailed evolutionary calculations ( ivanova 2006 ) .
all other stars ( e.g. , @xmath110 helium stars in the hertzsprung gap ) are assumed to have radiative envelopes .
we use the original hurley et al .
( 2002 ) formulas to calculate the mass and depth of convective envelopes .
we assume that low - mass evolved helium stars ( @xmath101 ) below @xmath102 ( as opposed to @xmath111 in hurley et al .
2000 ) expand and form deep convective envelopes in their late stages of evolution ( e.g. , ivanova et al . 2003 ; dewi & pols 2003 ) .
helium stars with convective envelopes are subject to strong tidal interactions ( convective tides as opposed to radiative damping , see 3.3 ) , and if found in an interacting binary , they may alter significantly the fate of a given system .
all helium stars ( @xmath112 ) may be subject to stable rlof .
however , in dynamically unstable cases we assume a binary component merger in the case of a hems donor ( @xmath17 ) and a helium hertzsprung gap donor ( @xmath113 ; e.g. , ivanova et al . 2003 ) or we follow a given system through a ce phase for evolved he star donors ( @xmath101 and test whether the system survives or merges .
the examination of rlof stability and development of dynamical instability are described in detail in 5 ) .
the treatment of helium stars is important , for example , in later stages of evolution leading to double neutron star formation .
the immediate consequences , leading to the formation of a new class of close double neutron stars , were discussed in belczynski & kalogera ( 2001 ) and belczynski , bulik & kalogera ( 2002a ) .
the formation of the new class of ( ultracompact ) double neutron stars involves helium hertzsprung gap stars initiating ce phase .
now , guided by the better understanding of ce phase survival ( ivanova et al .
2003 ) , we do not allow survival in such cases , and in our reference model both double neutron star merger rates and double neutron star properties have changed ( belczynski et al . 2007 ) .
however , we note that it is still predicted that a significant fraction ( @xmath114 ) of close double neutron star binaries form in very close orbits ( merger times smaller than 100 myr ) . due to significant updates of the code and new observational results on short grbs with double neutron stars suggested as their progenitors ( e.g. , fox et al .
2005 ) new startrack calculations relevant to the double neutron star formation have been performed ( belczynski et al .
2007 ; belczynski et al .
, in prep . )
throughout the course of binary evolution we track the changes in orbital properties . a number of physical processes may be responsible for these changes . in the general case of eccentric orbits
we numerically integrate a set of four differential equations describing the evolution of orbital separation , eccentricity and component spins , which depend on tidal interactions as well as angular momentum losses associated with magnetic braking , gravitational radiation and stellar wind mass losses .
for circular orbits with synchronized components , we can obtain an exact solution for the change of orbital separation using conservation of angular momentum .
losses of angular momentum and/or mass associated with rlof events , magnetic braking and gravitational radiation are taken into account .
we assume that any system entering rlof becomes circularized and synchronized ( if it had not already reached this equilibrium state before rlof ) .
in such a case we circularize to the periastron distance @xmath115 where and @xmath116 mark initial and final values ) , and both components are synchronized to the new mean angular orbital velocity .
non circular orbits and non synchronous rlof cases are important for massive binaries , in which massive stars may remain eccentric / unsynchronized as tidal interactions are not as effective as for low mass stars ( see 3.3 ) . in particular , the vast majority of close ( coalescing ) ns - ns binary progenitors may evolve through such configuration more than once . it should be noted that our prescription ( eq . [ eccrlof ] ) does not conserve total angular momentum ( i.e. , sum of orbital and spin angular momenta of the components ) .
for most of the cases in double compact object progenitor evolution , our prescription leads to a moderate ( @xmath117 ) loss of angular momentum . for systems which have not been circularized and synchronized before entering rlof there might be substantial mass loss ( e.g. , hut & paczynski 1984 ) , and although this is not taken into account in our calculations , it may also lead to some angular momentum loss . for some systems , consisting of low mass object ( e.g. ,
a neutron star ) and a massive star ( e.g. , a massive cheb star ) eccentricity is induced by the supernova explosion and once the massive star overfills its roche lobe we circularize / synchronize the system , and this may lead to a slight increase of total angular momentum ( @xmath76 ) .
however , it needs to be stressed that there is no solution that conserves total angular momentum in such cases , as systems evolve toward a darwinian unstable state ( see 3.3 ) and eventually evolve towards a common envelope phase . since the orbital shrinkage of a system with a massive star donor during the common envelope phase is usually dramatic ( @xmath118 orders of magnitude ) , the moderate change of orbital separation prior ce phase ( eq . [ eccrlof ] ) and its specific magnitude ( in the prescription used ) does not play an important role .
violent processes like sn explosions or common envelope phases are taken into account in binary orbital evolution .
also nuclear evolution of components ( expansion / contraction affecting stellar spins ) is considered . in what follows
we describe the elements used to calculate the orbital evolution .
the orbital angular momentum of the binary and its mean angular velocity are expressed as @xmath119 @xmath120 where @xmath121 is the gravitational constant .
binary angular momentum loss due to gravitational radiation is estimated for any type of binary following peters ( 1964 ) @xmath122 where , @xmath123 is the speed of light .
emission of gravitational radiation causes orbital decay as well as circularization , both taken into account during the evolution of a binary system . for any given system , the merger time
may be easily estimated ( e.g. , see eq.14 in bkb02 ) .
each binary component may be subject to magnetic braking , causing the decrease of the component s rotation . in the case of a detached binary configuration magnetic braking
is applied directly to the component spins , while during rlof the effects of magnetic braking are applied to the orbit , since the components are then kept in synchronism .
three different prescriptions for magnetic braking are incorporated within the startrack code and may be used interchangeably for parameter studies . in what follows
we provide a detailed description of the specific braking laws adopted .
magnetic braking is applied to stars with a significant convective envelope , i.e. , for low - mass h - rich ms stars , h - rich giant - like stars and cool hg and cheb stars ( see 2.3.4 for details ) with the exception of low - mass evolved helium stars for which there is not much known about magnetic fields . for fully convective ms stars ( @xmath124 , @xmath125 ) magnetic braking may also operate , although it has been hypothesized that the braking is suppressed ( rappaport , verbunt & joss 1983 ; zangrilli , tout & bianchini 1997 ) in order to provide an explanation of the observed period gap for cataclysmic variables .
therefore we assume that magnetic braking is not operative for fully convective stars , independent of the prescription used . since massive core helium burning stars , more massive h - rich ms stars , and he - rich ms stars have radiative envelopes , we assume that magnetic braking does not operate in these stars .
the prescription for the loss of angular momentum associated with magnetic braking @xmath126 takes several forms .
historically , most studies have adopted the form suggested by rappaport et al .
( 1983 ) where @xmath127 with parameter @xmath128 in our model calculations .
studies based on the observations of rapidly rotating stars show that the skumanich relation ( @xmath129 ) is inadequate in this regime and point to a weakening of magnetic braking due to saturation of the dynamo ( andronov , pinsonneault & sills 2003 ) . in this case
, the angular momentum loss rate takes the form @xmath130 where , @xmath131 denotes the component for which magnetic braking is operating , @xmath132 [ @xmath133 is angular velocity , and @xmath134 stands for a critical value of angular velocity above which the angular momentum loss rate enters the saturated regime . if the latter law is used , the saturation is applied only for ms stars and @xmath134 is interpolated from table 1 of andronov et al .
( 2003 ) .
in addition , we also include the form of magnetic braking from the results of a study by ivanova & taam ( 2003 ) . in this latter study , an intermediate form of the angular momentum loss rate
was derived ( @xmath135 ) based on a two component coronal model as applied to the observational data relating stellar activity to stellar rotation .
specifically , we adopt @xmath136 with @xmath137 .
this law is used for the startrack standard model calculations .
the evolution of the orbital parameters ( @xmath138 ) as well as component spins ( @xmath132 , @xmath139 ) driven by tidal interactions of binary components is computed in the standard equilibrium - tide , weak- friction approximation ( zahn 1977 , 1989 ) , following the formalism of hut ( 1981 ) .
this formalism allows us to treat binaries with arbitrarily large eccentricities .
we assume that the only sources of dissipation are eddy viscosity in convective envelopes and radiative damping in radiative envelopes .
specifically , we integrate numerically the following differential equations in parallel with the stellar evolution @xmath140 @xmath141 @xmath142 where + @xmath143 + @xmath144 + @xmath145 + @xmath146 + @xmath147 + and @xmath148 is the gyration radius and is defined by @xmath149 , with @xmath150 denoting the moment of inertia of a given binary component . here
the mass ratio is defined as follows , @xmath151 the quantity @xmath152 is the ratio of the apsidal motion constant @xmath153 ( which depends on the interior structure of the star ) over the timescale @xmath154 of tidal dissipation .
following hurley , tout & pols ( 2002 ) , we calculate that constant for either the equilibrium tide with convective damping ( @xmath155 ) or the dynamical tide with radiative damping ( @xmath156 ) .
radiative damping is applied to stars with radiative envelopes : ms stars with mass above @xmath99 , cheb stars with mass above @xmath157 , massive evolved he stars and all he ms stars . for all other stars ,
convective damping is applied ( see 2.3.4 for details on convective / radiative envelopes ) .
we do not calculate tides on stellar remnants , e.g. , on wds ( @xmath158 ) .
the constant for convective damping is obtained from @xmath159 where @xmath160 is the mass contained in the convective envelope of component @xmath131 .
the eddy turnover time @xmath161 is calculated as @xmath162^{1/3 } { \rm yr } \label{tid6}\ ] ] with @xmath163 denoting the depth of the convective envelope and @xmath164 the bolometric luminosity of a given component ( rasio et al .
1996 ) . the numerical factor @xmath165 is defined as @xmath166 \label{tid7}\ ] ] with the tidal pumping timescale @xmath167 defined as @xmath168 where @xmath169 and @xmath170 are the binary orbital period and the spin period of component @xmath131 , respectively .
this factor represents the reduction in the effectiveness of eddy viscosity when the forcing period is less than the turnover period of the largest eddies ( goldreich & keeley 1977 ) the constant for radiative damping is calculated from @xmath171 where a second - order tidal coefficient @xmath172 was fitted ( hurley et al .
however , their binary code utilizes the proper formula .
jarrod hurley , private communication . ] ) to values given by zahn ( 1975 ) .
finally , we have introduced an additional scaling factor @xmath173 in the evolution equations ( eq .
[ tid1 ] , [ tid2 ] , [ tid3 ] ) which we normally set to : @xmath174 and the distinction between the stars with convective and radiative envelopes is given in 2.3.4 .
this factor makes tidal forces ( both in case of convective and radiative damping ) more effective than predicted by the standard zahn theory .
the choice of this specific value of @xmath173 is a result of our calibration against the cutoff period for circularization of binaries in m67 and from the orbital decay of the high mass x - ray binary lmc x-4 ( for details see 8.2 ) .
the orbital angular momentum change associated with tides is calculated from @xmath175 and the change of binary parameters is calculated with eqs .
[ tid1 ] , [ tid2 ] , [ tid3 ] .
_ pre main sequence tidal synchronization and circularization .
_ we also allow for pre - ms tidal interactions . since we do not follow pre - ms evolution , all binaries with orbital periods shorter than @xmath176d ( mathieu et al .
1992 ) are simply assumed to have circularized and all binary components to have synchronized by the time they reach the zams .
for binaries with longer orbital periods we apply our assumed distribution of initial eccentricities ( see 7 ) and initial rotational velocities for binary components ( see 2.3.3 ) . _ darwin instability .
_ one important consequence of tidal interactions in massive binaries is the possible occurrence of the darwin instability ( e.g. , lai et al.1993 ) .
when the more massive component is spinning slowly compared to the orbital rate of its companion , tidal forces will tend to spin it up , leading to loss of orbital angular momentum ( orbital decay ) .
usually this orbital decay will stop when synchronization is established .
however , if , in the synchronized state , more than a third of the total binary angular momentum would be in the component spins , then synchronization can never be reached and the components will continue to spiral in .
we follow this process until one of the binary components overflows its roche lobe . in all cases rlof is found to be dynamically unstable ( 5.1 and 5.2 ) and the system goes through a ce phase leading either to a merger , or further orbital decay with envelope ejection ( 5.4 ) .
mass lost from the binary components in stellar winds carries angular momentum , in turn affecting the orbit through tidal coupling .
similarly , during rlof , some of the transferred material and its associated angular momentum may be lost from the system . in this section
we consider the amount of angular momentum loss associated both with stellar winds and rlof phases .
however , for rlof we only consider here dynamically stable phases , while the change of the orbit following unstable rlof ( common envelope events ) is described in 5.4 . for stellar winds we assume spherically symmetric mass loss , which carries away the specific angular momentum of the mass losing component ( jeans mode mass loss ) .
the corresponding change of the orbit ( jeans - mode mass loss ) is calculated from @xmath177 the above approach holds for circular orbits , however the change in binary separation @xmath178 is similar for eccentric orbits ( vanbeveren , van rensbergen & de loore 1998 , see p. 124 ) . in the case of stable rlof with compact accretors ( wd , ns , bh ;
@xmath179 ) we limit ( although this assumption may be relaxed ) accretion to the eddington critical rate @xmath180 and the corresponding critical eddington luminosity may be expressed as @xmath181 where @xmath182 denotes the radius at which the accretion onto compact object takes place ( a ns or a wd radius , and three schwarzschild radii for a bh ) , @xmath183 denotes the composition of accreted material ( 0.7 for the h - rich material , and 0.0 for all other compositions ) , and @xmath184 gives the conversion efficiency of gravitational binding energy to radiation associated with accretion onto a wd / ns ( surface accretion @xmath185 ) and onto a bh ( disk accretion @xmath186 ) .
we also note that above some critical ( very high ) accretion rate , nuclear burning will start on the wd surface
. this can be much more radiatively efficient than the gravitational energy release and the above relations break down .
if the mass transfer rate is higher than @xmath187 we expect the excess material to leave the system from the vicinity of the accreting object and thus to carry away the specific angular momentum of the accretor .
the angular momentum loss associated with a given systemic mass loss in a rlof phase is obtained from @xmath188 where @xmath189 is the distance between the accretor and binary center of mass , and @xmath190 is the mass transfer rate ( donor rlof rate , see eq .
[ mt10a ] ) .
the @xmath191 fraction of material transferred from the donor is accreted on the compact object . if mass transfer is sub - eddington then @xmath192 ( conservative ) , otherwise it is @xmath193 ( non - conservative evolution ) .
here we assume that the radiative efficiency is not a function of the mass transfer rate .
some work has suggested that at high transfer rates , flows onto black holes may become radiatively inefficient as photons are trapped in the flow and advected into the black hole ( see e.g. , abramowicz et al .
1988 ) , or that substantially super - eddington accretion may be possible in non - spherical accretion flows ( e.g. , begelman 2002 ) . in the current version of the code , we do not consider these possibilities .
for all other , non - degenerate accretors ( @xmath194 ) we assume a non - conservative evolution through stable rlof , with part of the mass lost by the donor accreted onto the companion ( @xmath191 ) , and the rest ( @xmath195 ) leaving the system with a specific angular momentum @xmath196 expressed in units of @xmath197 ( see podsiadlowski , joss & hsu 1992 ) .
the angular momentum loss is then estimated from @xmath198 for our standard model calculation we adopt @xmath199 , and @xmath200 ( half of the transferred mass lost from system , e.g. meurs & van den heuvel 1989 )
. however , we note that the amount of mass loss ( as well as the specific angular momentum with which the mass is lost from the binary ) may change from case to case .
ideally , one would like to know the details of mass transfer / loss for all potential binary configurations , and change @xmath196 and @xmath191 according to the types of interacting stars as well as binary properties .
since , such predictions or understanding are not available , we treat @xmath196 and @xmath191 as parameters , which are applied evenly to all the stars in a given simulation .
we adopt the compilation of mass loss rates from hurley at al .
we have further extended the original formulas to include winds from low- and intermediate - mass ms stars .
the structure of the star ( and its subsequent evolution ) in response to stellar wind mass loss is self - consistently taken into account with the hurley et al .
( 2000 ) evolutionary formulas .
the most important effects include possible removal of the h - rich envelope of a massive star or a more gradual nuclear evolution with decreasing mass .
the effects of wind mass loss from binary components on the orbital parameters are also accounted for ( see 3.4 ) .
the effects of mass increase of binary components due to accretion from the companion winds are neglected . either the wind accretion rates are very low or the high wind accretion phases do not last for long , which does not translate into significant mass increase of a companion star .
however , we estimate the wind accretion rates onto nss and bhs since it may give rise to bright x - ray emission ( see 9 ) . the wind accretion rate is calculated in the general case of eccentric orbits ,
i.e. we obtain accretion rate ( and accretion luminosity ) for a specified position on the orbit , or we integrate over a specific part of the orbit ( e.g. , corresponding to the exposure time of given observations ) .
this may be of importance for eccentric wind - fed binaries , e.g. , high mass x - ray binaries ( see 9.2 ) .
we have also implemented an orbital - averaged solutions .
the two solutions may be adopted as required for a given project or analysis .
we follow the bondi & hoyle ( 1944 ) accretion model to calculate the accretion from stellar wind . as an approximation we may express ( boffin & jorissen 1988 ) the accretion rate as @xmath201 where @xmath202 is the accretion efficiency in the bondi hoyle model , although it may be as low as 0.05 in some specific cases ( e.g. , see hydrodynamical simulations of theuns , boffin & jorissen 1996 for barium star formation )
, @xmath203 is the wind sound speed , and @xmath204 is the relative velocity of the wind with respect to the accreting star
. the local ( undisturbed ) density of the wind matter @xmath205 in the vicinity of the accreting object may be calculated in a steady spherically symmetric case from @xmath206 where @xmath207 is the wind mass loss rate form the donor , @xmath208 is the instantaneous distance between the two stars , and @xmath209 is the wind velocity .
we assume that the wind flow is supersonic ( @xmath210 ) so that @xmath211 may be dropped from eq .
[ wind01 ] .
we introduce @xmath205 ( expressed through eq .
[ wind02 ] ) into eq .
[ wind01 ] to obtain @xmath212 the accretion rate calculated with eq .
[ wind03 ] varies as the accreting object moves in its orbit around the mass losing star .
the relative distance @xmath208 of the two stars is obtained through the kepler equation for a given orbit .
obviously @xmath208 is a function of orbital position .
the vector of the relative velocity @xmath213 is defined as @xmath214 where @xmath215 denotes the instantaneous velocity of the accretor on the orbit relative to the mass losing star , and is readily obtained for a given position through the kepler equation .
the direction of the wind velocity vector @xmath216 follows the vector pointing toward the accretor on its relative orbit around the mass losing star .
we set the wind velocity proportional to the escape velocity from the surface of the mass losing star @xmath217 and vary @xmath218 with the spectral type of the mass losing star .
for extended ( @xmath219 ) h - rich giants ( @xmath220 ) slow winds are assumed @xmath221 .
for the most massive ms stars ( @xmath222 ) @xmath223 , for low mass ms stars ( @xmath224 ) @xmath225 and the value of @xmath226 is interpolated in - between . for he - rich stars ( @xmath227 ) ; @xmath223 for @xmath228 , @xmath221 for @xmath229 , and is interpolated in - between .
the values of @xmath218 follow from the observations of wind velocities for different type of stars ( lamers , snow & lindholm 1995 ; kucinskas 1999 ) and are adopted from the discussion of wind properties in hurley et al .
( 2002 ) .
we use eq .
[ wind03 ] to obtain the orbit - averaged accretion rate .
the wind velocity vector is assumed to be perpendicular to the orbital speed vector ( as on a circular orbit ) , i.e. , @xmath230 .
the wind velocity is taken from eq .
[ wind05 ] .
the orbital velocity of the accretor is taken to be constant and is obtained from the circular orbit approximation @xmath231 .
finally , @xmath232 is substituted in eq .
[ wind03 ] with its mean value over one orbital revolution , i.e. , @xmath233 to obtain @xmath234 where @xmath235 is a parameter ( see below ) and @xmath236 . for highly eccentric orbits , the averaged ( over one orbit ) accretion rate calculated with the eq .
[ wind06 ] may exceed the companion mass loss rate .
this is a direct result of the orbital averaging used above . to avoid this
we follow hurley et al .
( 2002 ; 2.1 ) and adopt @xmath235 such that @xmath237 never exceeds @xmath238 .
different physical processes may be responsible for driving rlof . in the following we describe the treatment of mass loss and mass accretion in our model . for any binary system during rlof phases with a non - degenerate donor ( @xmath239 )
we calculate the radius mass exponents for the donor and its roche lobe @xmath240 @xmath241 and we estimate the change of donor radius with time due to its nuclear evolution as @xmath242 the above derivatives are calculated numerically with the use of the analytic single star formulas of hurley et al .
the time derivative of the stellar radius ( @xmath243 ) is obtained directly from the single star formulas , since the radius of a given star is tracked in time . to obtain response of the donor to mass loss ( @xmath244 )
we calculate the response of the star to an instantaneous ( over a timestep of only 1yr , a time interval unimportant for stellar evolution ) mass loss through ( artificially ) increased wind mass loss .
finally , the roche lobe exponent is obtained by removing 1% of the donor mass , part of which is transferred to the accretor and the rest is lost with the specific angular momentum of the accretor from the binary ( see 3.4 ) ; a new roche lobe radius and the numerical derivative are then readily calculated .
rlof may be driven by different physical processes ; angular momentum losses connected to magnetic braking ( applied directly to the orbit , given the assumption of synchronism during rlof ) and gravitational radiation or expansion due to nuclear evolution .
we do not include tides as we assume that binary is circular and synchronized during rlof .
the timescales for magnetic braking , and gravitational radiation are calculated from @xmath245 @xmath246 where expressions for @xmath247 , are given in 3.1 , and 3.2 respectively .
if rlof is driven by the combination of angular momentum losses changing the orbit and nuclear evolution of the donor we then calculate the mass transfer rate from @xmath248 and the corresponding mass transfer timescale @xmath249 additionally we estimate the thermal timescale for the donor following kalogera & webbink ( 1996 ) from @xmath250 and the mass transfer rate on the thermal timescale @xmath251 in the case of stable rlof , @xmath252 , a donor is in thermal equilibrium , and we use eq .
[ mt7 ] to calculate the mass transfer rate . otherwise , for @xmath253 , rlof proceeds on the thermal timescale and we evolve a given system calculating the mass transfer rate from eq . [ mt10 ] .
we follow the timescales of the donor as it evolves through rlof , and apply the appropriate mass loss rate . for example , a massive donor may be transferring mass on the thermal timescale at first , but once it loses a fraction of its mass , the mass transfer becomes stable and rlof proceeds on the timescale defined by @xmath254 .
however , in some cases the rlof is so rapid that it may eventually lead to a dynamical instability .
once @xmath255 changes sign and becomes positive , the donor loses its equilibrium , and the system evolves either on the thermal or dynamical timescale . in this case a special _ diagnostic diagram _ is used ( see below ) to decide which of the two timescales is relevant .
we also allow for the development of a delayed dynamical instability , which may occur for stars with a radiative envelope , but with a deep convective layer .
dynamical instability during rlof leads to a spiral - in of the binary components and common envelope evolution ( ce ) .
we follow the ce phase to determine whether the binary survives ( ejection of the envelope at the expense of orbital energy ) or if a merger of the binary components ( single star formation ) occurs .
the following summarizes the calculation of the rlof mass transfer rates @xmath256 where we additionally assume that above some critical mass ratio ( @xmath257 ) the binary system will evolve toward delayed dynamical instability ( hjellming & webbink 1987 ) , leading to rapid inspiral and ce evolution .
for h - rich stars hjellming ( 1989 ) gives a range @xmath258 depending on the evolutionary state of a donor , while ivanova & taam ( 2004 ) obtain @xmath259 . in our standard model calculations
we adopt @xmath260 for h - rich stars ( @xmath261 ) . for he - rich stars we adopt critical mass ratios from ivanova et al .
( 2003 ) ; @xmath262 for hems stars ( @xmath17 ) , while @xmath263 for evolved he stars ( @xmath264 ) .
we note that the study of ivanova et al .
( 2003 ) was targeted for he stars with ns accretors only
. however , we adopt their values for systems with he star donors and arbitrary accretors , since detailed models for arbitrary accretors are not available .
also , dynamical instability may be encountered if the trapping radius of the accretion flow exceeds the roche lobe radius of the accretor ( 5.4 ) .
additionally , we consider the case of spiral in in the case of the darwin instability , where the components spin angular momentum is comparable to the orbital angular momentum ( 3.3 ) .
for the donor stars without a well defined core - envelope structure ( @xmath265 ) we assume that dynamical instability during rlof _ always _ leads to a merger .
the same is assumed for the donors in the hertzsprung gap ( @xmath266 ) as there is no clear entropy jump at the core - envelope transition ( ivanova & taam 2004 ; belczynski et al .
2007 ) . in the case of a merger
a single stellar object is formed .
however , we do not follow its evolution here , as the chemical composition and structure of merged remnants is not well understood and certainly is different than normal stars . this may lead to an underestimate of our synthetic supernovae rate , since potentially some merger products are massive enough to evolve and explode as type ii or ib / c sne . for h - rich and he - rich giant - like donors ( @xmath267 )
we follow ce evolution , and assuming ejection of the entire donor envelope , we calculate the most probable outcome with conservation of energy ( see 5.4 ) .
if rlof is encountered for a system with an evolved helium star donor ( @xmath264 ) , then it is found that for low donor masses ( @xmath268 ) rlof is stable ( although it may proceed at very high rates ) while for higher donor masses it leads to a ce phase ( e.g. , see ivanova et al .
the survival of the binary then depends on the donor properties ( e.g. , stellar structure , envelope binding energy , its mass , binary separation ) and in particular for he stars in the hertzsprung gap ( @xmath269 ) ce phase always leads to merger .
the mass accretion rate in a dynamically stable rlof is calculated from @xmath270 where @xmath190 is the donor rlof mass transfer rate ( see eq . [ mt10a ] ) .
the parameter @xmath271 denotes the fraction of the transferred mass which is accreted , while the rest ( @xmath272 ) is ejected from the system ( see 3.4 ) .
mass accretion in dynamically unstable cases ( ce events ) is calculated only for ns and bh accretors , since only then significant accretor mass gain may be expected in spite of the very short timescales ( for details see bkb02 ) .
the aforementioned diagnostic diagram is shown in fig [ fig01 ] .
once rlof proceeds on the thermal timescale , and the donor is no longer in thermal equilibrium , we do not have proper stellar models to use and calculate the donor properties ( e.g. , rlof rate ) . therefore
, we use an approximate method and calibrate it based on the results from detailed stellar evolutionary and mass transfer calculations , which are not limited to stars in thermal equilibrium .
when the donor loses its equilibrium , we use the stellar and binary properties to predict whether the system will evolve through the phase of thermal mass transfer and the donor will regain its equilibrium , or the rlof will become dynamically unstable and will eventually lead to ce evolution .
we plot the donor roche lobe radius versus decreasing donor mass under the assumption that mass transfer is non - conservative and proceeds on the thermal timescale ( see eqs . [ mt9 ] and [ mt10 ] ) . for ns / bh accretors the accretion rate
is limited by the eddington rate , while for all other accretors , a fraction @xmath191 of transferred material is accreted .
the associated specific angular momentum loss is described in 3.4 . as the mass of the donor decreases with mass transfer
the roche lobe first shrinks and then at some critical mass ratio ( @xmath273 ) , it starts expanding again ( see the solid line on the top panel , fig . 2 ) .
if the mass ratio at the moment the star loses its equilibrium @xmath274 is not greatly different than @xmath273 we expect that the donor may regain the equilibrium when the system is expanding .
the dashed line arrow in figure 2 shows the expected behavior of the donor when it loses its equilibrium .
if the system does not evolve into a ce phase then we expect the donor to regain its equilibrium at the position indicated by the arrow .
of course this is just an approximation , since , as the donor evolves , the radius - mass exponent changes .
we use a number of published ( tauris & savonije 1999 ; wellstein & langer 1999 ; wellstein , langer & braun 2001 ; dewi & pols 2003 ) and unpublished ( n. ivanova 2004 , private communication ) detailed calculations to calibrate the diagnostic diagram . based on these studies
we find that a ce phase ensues if @xmath275 otherwise the system is evolved through rlof on the donor s thermal timescale .
once a binary is identified as a thermal timescale rlof system , we assume that the mass transfer rate remains constant throughout the entire episode .
we calculate the rate using eq .
[ mt10 ] where we use properties corresponding to the time the donor loses its thermal equilibrium .
this may be justified by the following : _
( i ) _ thermal mass transfer rates have been shown to be rather constant within a factor of @xmath276 ( paczynski 1971 ) , _ ( ii ) _ since the rates are calculated at the time the star loses equilibrium , it is a good approximation ( and the best possible with only equilibrium stellar models being available ) for the short lived phase of thermal mass transfer that follows . in the bottom panel of figure [ fig01 ]
we show an example calculation through a thermal rlof phase , followed with a slower ( driven by nuclear evolution ) rlof period after the donor has regained its thermal equilibrium .
the specific system was chosen to match the rlof calculation of wellstein et al .
( 2001 ) for a @xmath277 and @xmath278 binary with an initial period of 8 days .
the rlof starts when the primary evolves off the main sequence and crosses the hertzsprung gap .
mass transfer initially proceeds on a thermal timescale at a very high rate ( @xmath279 ) , then the star regains its equilibrium and the rlof rate decreases with time by more than order of magnitude ( @xmath280 ) . our calculation
can be directly compared to wellstein et al .
( 2001 ) : see their figure 4 , left panel .
their detailed stellar evolution calculation shows a thermal rlof rate of @xmath281 , followed by a slower rlof phase characterized by rates of @xmath280 , very similar to what we find with our simplified prescription .
our rlof phase lasts about twice as long as that of wellstein et al .
( 2001 ) , who in contrast to our calculation assumed conservative evolution and did not include effects of tidal spin - orbit interactions .
we choose not to modify our standard model assumptions ( e.g. , neglect tidal interactions ) for comparisons , and therefore emphasize some differences with previous calculations .
more comparisons of rlof sequences are presented in 8.1 .
dynamically unstable mass transfer may be encountered in a number of ways .
most often it is the direct consequence of stellar expansion during rapid nuclear evolution phases .
however , loss of orbital angular momentum ( e.g. , via magnetic braking , gravitational radiation , or tides ) may also lead to dynamical instability .
additionally , we allow a system to evolve into a ce phase if the trapping radius of the accretion flow exceeds the roche lobe radius of the accretor
. the trapping radius is defined as ( begelman 1979 ) @xmath282 following king & begelman ( 1999 ) and ivanova et al . ( 2003 ) we check whether the mass transfer rate exceeds a critical value above which the system is engulfed in a ce @xmath283 where @xmath284 is the accretor roche lobe radius , and @xmath285 is the eddington critical accretion rate ( see eq .
[ dmedd ] ) .
below we present two different implementations of the orbital contraction calculation during ce that are incorporated in startrack . _
standard energy balance prescription .
_ if dynamical instability is encountered a binary may enter a ce phase .
we use the standard energy equation ( webbink 1984 ) to calculate the outcome of the ce phase @xmath286 where , @xmath287 is the mass of the donor envelope ejected from the binary , @xmath288 is the roche lobe radius of the donor at the onset of rlof , and the indices @xmath289 denote the initial and final values , respectively .
the parameter @xmath290 is a measure of the central concentration of the donor ( de kool 1990 ; dewi & tauris 2000 ) .
the right hand side of equation ( [ ce ] ) expresses the binding energy of the donor s envelope , the left hand side represents the difference between the final and initial orbital energy , and @xmath291 is the ce efficiency with which orbital energy is used to unbind the stellar envelope .
if the calculated final binary orbit is too small to accommodate the two post - ce binary components then a merger occurs . in our calculations , we combine @xmath291 and @xmath290 into one ce parameter , and for our standard model , we assume that @xmath292 .
this is for all but evolved naked helium stars ( @xmath101 ) for which we adopt @xmath293 and @xmath294 , where @xmath295 is radius of helium star in solar radii .
the relation for @xmath290 was obtained with ivanova s ( 2003 ) evolutionary code .
if a compact object spirals in the common envelope it may accrete significant amounts of material because of hyper - critical accretion ( blondin 1986 ; chevalier 1989 , 1993 ; brown 1995 ) .
we have incorporated a numerical scheme to include the effects of hyper - critical accretion on nss and bhs in our standard ce prescription ( for details see bkb02 ) .
compact objects gain , on average , several tenths of solar mass in ce if hyper - critical accretion is allowed .
however , we also allow for evolution with no hyper - critical accretion following recent results of accretion flow calculations with geometry specific for compact object moving through common envelope .
these calculations indicate that accretion can be limited to only @xmath296 ( e.ramirez-ruiz , private communication ) .
_ alternative angular momentum prescription _ in addition to the standard prescription for common envelope evolution based on comparing the binding and orbital energies ( see above ) , we investigate the alternative approach proposed by nelemans & tout ( 2005 ) , based on the non - conservative mass transfer analysis by paczynski & ziolkowski ( 1967 ) , with the assumption that the mass loss reduces the angular momentum in a linear way .
this leads to reduction of the orbital separation @xmath297 where @xmath287 is the mass of the donor envelope lost by the system , @xmath298 , @xmath299 are the total masses of the system before and after ce , and @xmath300 is a scaling factor .
following nelemans & tout ( 2005 ) we use @xmath301 and note that hyper - critical accretion is not included in this prescription .
the two above prescriptions are extended ( e.g. , bkb02 ) to the case where both stars lose their envelopes , which happens if the stars have giant - like structure ( @xmath94 ) at the onset of ce phases ( see bethe & brown 1998 ) .
degenerate donors ( @xmath302 ) , are also considered .
the rlof is assumed to be driven by gravitational radiation only @xmath303 with @xmath304 where the mass ratio is defined as @xmath305 , @xmath191 denotes the fraction of transferred material that is accreted by the companion ( defined and evaluated in 3.4 ) , and @xmath306 .
mass loss / gain changes the subsequent evolution of stars .
we implement rlof mass loss / gain by adding an extra term in the original hurley et al .
( 2000 ) stellar evolution formulae .
in case of mass loss we increase the wind mass loss rate to match the combined effects of wind and rlof mass loss . to treat mass gain and potential accretor rejuvenation ,
we add the rlof mass accretion rate , as calculated in 5.6 , to the accretor wind mass loss rate ( they have opposite signs ) .
additionally , in case of rejuvenation evolutionary timescales and stellar ages are modified as suggested by tout et al .
( 1997 ) and hurley et al .
( 2002 ) by the relative mass change ( see the following equation ) to calculate the net effect on the star . for main sequence stars we can calculate the change of the age of a given star ( due to accretion or mass loss ) from @xmath307 where @xmath308 is the main sequence lifetime , and indices @xmath116 mark the state before and after some amount of mass is transferred , respectively .
the factor @xmath309 is unity for all mass losing stars and for hydrogen ms stars ( @xmath310 ) with radiative cores ( @xmath311 ) , while it is @xmath312 for hydrogen ms stars with convective cores and helium ms stars ( @xmath17 ; they have convective cores ) and it reflects the effects of additional fuel in the core . for hg stars ( @xmath103 )
we change the timescales using @xmath313 as for ms stars with radiative cores . in this way
we ensure that the subsequent evolution of the donor is followed consistently , i.e. , evolutionary timescales and physical properties of mass losing / gaining stars ( e.g. , core masses ) are changed in agreement with stellar models .
our wind mass loss formulae are implemented the same way as in the original hurley et al .
( 2000 , see their 7.1 ) and the above scheme allows for appropriate changes of evolutionary timescales and core masses , both in cases of mass loss and gain . for simplicity , we assume that the composition of the accreted material matches that of the accretor , although this may not always be the case . only in the case of accretion onto white dwarfs we take into account the composition of accreted material ( see 5.7 ) . a number of important phenomena , e.g. , novae and type ia sn explosions or accretion - induced collapses , are associated with mass accretion onto wds
. we incorporate the most recent results to estimate the accumulation efficiencies on wds .
in particular we consider accretion of matter of various compositions onto different wd types .
we also include the possibility that ns formation can occur via accretion induced collapse ( aic ) of a massive one white dwarf ( e.g. , bailyn & grindlay 1990 ; belczynski & taam 2004a ) . in this section
we discuss the accumulation of material and growth of the wd mass in binary systems . only during
dynamically stable rlof phases can the mass accretion onto wds be sustained for a prolonged period of time and hence affect the evolution of accreting wds . during dynamically unstable cases ( i.e. , ce evolution ) we assume that the wds do not accrete any material . if dynamical instability is encountered for a binary with two white dwarfs we assume that a merger occurs .
based on the results of saio & nomoto ( 1998 ) mergers of a one wd with any type of wd companion and two co wds lead to either aic and ns formation ( if total merger mass @xmath314 is above @xmath41 ) or the formation of the single one wd ( with new mass equal to @xmath314 ) . for mergers of co wd and he wd , we assume a type ia sna explosion ; either sub - chandrasekhar ( @xmath315 ) ( see woosley , taam , & weaver 1986 ; woosley & weaver 1994 ) or chandrasekhar mass sn ia ( @xmath316 .
mergers of other types of wds have total mass below the chandrasekhar mass , and in particular for co wd and h wd we assume formation of single co wd , while for he wd and h wd we assume formation of single he wd with masses equal to @xmath314 . during a phase of sustained mass accumulation the massive one wd ( @xmath317 ) may eventually collapse to a ns .
we include aic in our standard model calculations since it naturally follows from the adopted accumulation physics ( see below ) .
since little is known about potential asymmetries of the collapse , we either apply no natal kick ( standard model ) or a full natal kick ( parameter studies ) obtained from arzoumanian , chernoff , & cordes ( 2002 ) or hobbs et al .
( 2005 , see also 6.2 ) .
however , we also allow for the possibility of sn ia explosion instead of aic in parameter studies .
it is also worth noting the difference between accretion and accumulation .
the calculation of accretion rate during stable rlof was described in 5.1 , and this rate could be used to calculate , for example , the accretion luminosity of the system ( mostly in the uv part of spectrum for wd accretors ) . however , it is believed that in many cases ( see below ) not all of the accreted material remains on the surface of the accreting wd .
mass is lost either in shell flashes ( nova - like explosions ) or through optically thick winds from the surface of accreting wds .
to calculate the actual wd mass growth through the rlof phase the accumulation efficiency , @xmath318 , which is defined as @xmath319 must be known . here , @xmath320 is the mass accumulation rate on the surface of wd and the mass accretion rate ( @xmath321 ) is given by eq .
( [ mt10b ] ) . in what follows
we discuss the accumulation efficiency in various evolutionary scenarios .
_ accretion onto helium and hybrid white dwarfs .
_ it is assumed that if the mass accretion rate @xmath321 from the h - rich donor ( @xmath322 ) is smaller than some critical value @xmath323 , an unstable hydrogen shell flash will occur in the accreted layer . in response
, the envelope will expand beyond the roche lobe of the white dwarf .
we shall assume no material is accumulated , and that the accumulation efficiency is @xmath324 , i.e. the entire accreted material is lost from the binary . if @xmath325 then the material piles up on the wd leading to mass loss from the system .
assuming that a contact binary configuration is not formed in this case , the system will eventually undergo an inspiral . for giant - like donors we assume the system evolves through ce to examine if the system survives
; for all other donors we call it a merger and halt binary evolution .
the critical accretion rate is calculated as @xmath326 where , @xmath327 is an energy yield of hydrogen burning , @xmath183 is the hydrogen content of accreted material .
for population i stars ( metallicity @xmath328 ) we use @xmath329 , while for population ii stars ( @xmath330 ) we use @xmath331 ( ritter 1999 , see his eq .
10,12 and table 2 ) .
if the mass accretion rate from the he - rich donor ( @xmath332 ) is higher than @xmath333 all the material is accumulated ( @xmath334 ) until the accreted layer of material ignites in a helium shell flash . at this point degeneracy
is lifted , a main sequence helium star ( @xmath335 ) is formed and further accretion on the helium star is then taken into account . following the calculations of saio & nomoto ( 1998 ) we estimate the maximum mass of the accreted shell at which the flash occurs as @xmath336 where @xmath321 is expressed in @xmath337 .
the newly formed helium star may overfill its roche lobe , in which case either a single helium star is formed ( he or hyb wd companion , @xmath338 ) , a helium contact binary is formed ( hems companion , @xmath339 ) which we assume leads to a merger or the system goes through ce evolution ( evolved helium star companion , @xmath340 merger is assumed in such a case . ] ) . for accretion rates lower than @xmath341 , accumulation is also fully efficient ( @xmath334 )
. however , the sn ia occurs at a sub - chandrasekhar mass @xmath342 where @xmath321 is expressed in @xmath337 . for mass accretion rates close to @xmath341 , the above extrapolations from the results of hashimoto et al .
( 1986 ) yield masses smaller than the current mass of the accretor , and we assume an instantaneous sn ia explosion .
we note that above explosions disrupt the accreting wd , and although possibly subluminous , they appear as type ia sne ( no hydrogen ) .
we do not consider the accumulation of heavier elements since they could only originate from more massive wds ( e.g. , co or one wds ) , which would have smaller radii and could not be donors to lighter he or hyb wds .
_ accretion onto carbon / oxygen white dwarfs .
_ in this case we adopt the prescription from ivanova & taam ( 2004 ) . for h - rich donors and mass accretion rates lower than @xmath343
there are strong nova explosions and no material is accumulated ( @xmath324 ) . in the range
@xmath344 we interpolate for @xmath318 from prialnik & kovetz ( 1995 , see their table 1 ) . for rates higher than @xmath345
all accreted material burns into helium ( @xmath334 )
. additionally we account for the effects of strong optically thick winds ( hachisu , kato & nomoto 1999 ) , which eject any material accreted over the critical rate @xmath346 this corresponds to @xmath347 for @xmath348 .
the accretor is allowed to increase in mass up to @xmath349 , and then explodes as a chandrasekhar mass sn ia . in the case of
he - rich donors , if the mass accretion rate is higher than @xmath350 helium burning is stable and contributes to the accretor mass ( @xmath334 ) . for rates in the range @xmath351 accumulation is calculated from + @xmath352 $ ] + @xmath353 $ ] + @xmath354 $ ] + @xmath355 \\
-0.54 ( \log\mdot_{\rm acc}+5.6)^2 + 1.01\ [ -5.95\div-5.76 ] \\ \end{array } \right.\ ] ] @xmath356 $ ] + @xmath357 \label{wdacc10}\ ] ] and represents the amount of material that is left on the surface of the accreting wd of a specific mass ( denoted by a superscript on @xmath318 in @xmath24 ) after the helium shell flash cycle ( kato & hachisu 1999 , 2004 ) .
logarithms of critical mass accretion rates for a given specific wd mass are given in square brackets : @xmath358 $ ] . to obtain the accumulation rate for co wd within the mass range @xmath359 we incorporate results of the closest ( by mass ) model from the set of eqs .
[ wdacc10 ] .
if the wd mass drops below @xmath14 we use @xmath334 and we set @xmath360 ( see kato & hachisu 2004 ) .
the mass of the co wd accretor is allowed to increase up to @xmath349 , and then a chandrasekhar mass sn ia takes place in the two above he - rich accretion regimes . if mass accretion rates drop below @xmath361 , the helium accumulates ( @xmath334 ) on top of the co wd and once the accumulated mass reaches @xmath362 ( kato & hachisu 1999 ) , a detonation follows and ignites the co core leading to the disruption of the accretor in a sub - chandrasekhar mass sn ia ( e.g. , taam 1980 ; garcia - senz , bravo & woosley 1999 ) . if the mass of the accreting wd has reached @xmath349 before the accretion layer has reached @xmath362 then the accretor explodes in a chandrasekhar mass sn ia .
carbon / oxygen accumulation takes place without mass loss ( @xmath334 ) and leads to sn ia if chandrasekhar mass is reached . _ accretion onto oxygen / neon / magnesium white dwarf . _ accumulation onto one wds is treated the same way as for co wd accretors .
the only difference arises when an accretor reaches the chandrasekhar mass . in the case of one wd
this leads to an aic and ns formation , and binary evolution continues ( see belczynski & taam 2004a , 2004b ) .
all stars ( single and binary systems ) may be initialized with arbitrary velocities appropriate for a given environment . for example
, a galactic rotation curve may be used for a field population of a given galaxy , or a velocity dispersion can be applied for a cluster population .
the velocities of stars are then followed throughout their evolution .
single stars and binary systems are subject to recoil ( change of spatial velocity ) in sn explosions .
additionally , binary systems may be disrupted as a result of an especially violent explosion .
we account for both mass / angular momentum losses as well as for sn asymmetries ( through natal kicks that nss and bhs receive at their formation ; see below ) .
the detailed description of sn explosion treatment is given in bkb02 .
here , we only list the new additions to startrack .
the most important modification allows us to trace velocities of disrupted binary components after a sn explosion . for the first time , a full general approach with explosions taking place on orbits of arbitrary eccentricity ( in contrast to circular orbits only ) is applied to follow the trajectories of disrupted components .
first population synthesis results are presented in belczynski et al .
( 2006 ) . at the time of birth , nss and
possibly bhs receive a natal kick , which is connected to asymmetries in sn explosions .
we use the distributions inferred from observed velocities of radio pulsars .
we have replaced the natal kick distribution used in bkb02 ( cordes & chernoff 1998 ) with two more recent alternatives .
one presented by arzoumanian et al .
( 2002 ) is a bimodal distribution with a weighted sum of two gaussians , one with @xmath363 km sec@xmath364 ( 40% ) and another with @xmath365 km sec@xmath364 ( 60% ) .
the other was derived by hobbs et al .
( 2005 ) and is a single maxwellian with @xmath366 km sec@xmath364 . according to this most recent study
there is no indication of a bimodal ( low- and high - velocity ) kick distribution claimed in earlier studies ( e.g. , fryer , burrows & benz 1998 ; cordes & chernoff 1998 ; arzoumanian et al .
if this is indeed true , then some theoretical models built in support of the bimodal kick distribution ( e.g. , pfahl et al .
2002b and podsiadlowski et al . 2004 model of high mass x - ray binaries ) may need to be revised .
motions of many hundreds of pulsars are expected to be measured in the next few years .
these measurements will provide better constraints on the natal kick distribution ( hobbs et al .
until then we will use both distributions to assess the associated uncertainties in startrack calculations .
compact objects formed without any fall back receive full kicks drawn from one of the above distributions ; this case includes most of nss ( see 2.3.1 ) . the only exception are ns formed through ecs , for which we adopt either no natal kicks ( standard model ) or full kicks ( as part of parameter studies ) . during accretion induced collapse of wd to ns in an accreting binary system , the ns is formed through the same ecs process and the same prescription is adopted .
the recent numerical simulations of aic and ns formation through ecs point towards significantly lower energies of explosion than for regular core collapse sne ( dessart et al .
2006 ) , and although the kick mechanism is not yet identified , these results may be also indicative of lower kicks in ecs ns formation . for compact objects formed with partial fallback , heavy nss and light bhs ,
kicks are lowered proportionally to the amount of fallback associated with ns / bh formation @xmath367 where @xmath368 is the kick magnitude drawn from either arzoumanian et al .
( 2002 ) or hobbs et al .
( 2005 ) distribution , and @xmath29 is a fallback parameter , i.e. , the fraction ( from 0 to 1 ) of the stellar envelope that falls back ( see also 2.3.1 ) . for the most massive bhs , formed silently ( no sn explosion ) in a direct collapse ( @xmath33 ) of a massive star to a bh , we assume that no natal kick is imparted .
the adopted natal kick distribution and kick scaling for nss and bhs can be readily changed for parameter studies ( e.g. , full bh kicks ) . just prior to the sn explosion , the two components of the binary move with velocities @xmath369 and @xmath370 , which , in the center of mass ( cm ) system of coordinates , denoted here with the superscript @xmath371 , satisfy @xmath372 where @xmath373 denotes the sn component and @xmath374 its companion .
subscripts @xmath289 stand for initial and final values .
we make no assumptions about the orbit ; it can have an arbitrary eccentricity , in contrast to the derivation by tauris & takens ( 1998 ) , who assumed that the orbit is circular prior to the explosion . at the moment of a supernova explosion
the orbital separation is @xmath375 .
the exploding star loses its envelope , its mass becomes @xmath376 and receives a kick @xmath377 , so now its velocity in the coordinate system @xmath371 is @xmath378 the secondary star may be affected by the expanding shell and may receive an additional velocity @xmath379 , however it has been shown ( kalogera 1996 ) that the effect of this velocity is small , unless the pre - supernova orbital separation is smaller than @xmath380r@xmath381 .
we assume that the velocity of the companion is not affected by the impact of supernova ejecta .
we also assume that the velocity of the shell is large and the shell leaves the system quickly , i.e. @xmath382 , where @xmath383 is the orbital period of the system prior to the explosion . in order to calculate the final velocity of the two stars we first transform the velocities to the cm system of the two post - sn stars .
the velocity of this system , denoted as @xmath384 , in relation to system @xmath371 is @xmath385 the relative velocity of the two stars in this system is @xmath386 while the initial direction between the two stars remains the same as in the coordinate system @xmath371 , @xmath387 . in this new system
the relative motion of the stars is a hyperbola in the plane perpendicular to the angular momentum vector : @xmath388 where @xmath389 is the reduced mass of the system .
it is convenient now to introduce a third coordinate system @xmath390 in which the angular momentum @xmath391 lies along the z - axis .
the transformation from @xmath384 to @xmath390 is a rotation @xmath392 : @xmath393,@xmath394 .
the orbit in @xmath390 is described by @xmath395 where @xmath396 with @xmath397 is the ( positive ) energy of the system , and @xmath398 .
the final velocity , at @xmath399 , follows from energy conservation : @xmath400 in order to find the direction of the final velocity we note that conservation of angular momentum implies that at infinity ( @xmath399 ) : the final relative @xmath401 is parallel to the direction between the stars @xmath402 .
the initial position of the two stars on the trajectory described by eq .
[ traj ] is @xmath403 the sign of the angle @xmath404 is negative if the two stars initially lie on the descending branch of the hyperbola @xmath405 and positive if they are on the ascending one @xmath406 . in the first case ,
when the two stars are initially on the descending branch , we need to compare the distance of closest approach on the orbit @xmath407 with the radius of the companion star to examine whether the two stars collide instead of escaping to infinity .
we obtain the final position on the trajectory from @xmath408 and @xmath409 . thus the final direction between the two stars at @xmath410
is @xmath411 , where @xmath412 is the matrix of rotation around the z - axis , and their relative velocity is : @xmath413 we now have to transform quantities from system @xmath390 back to system @xmath371 to obtain the final velocities of the two disrupted binary components in the initial ( pre - sn ) cm system : @xmath414
each binary system is initialized by four parameters , which are assumed to be independent : the primary mass @xmath415 ( the initially more massive component ) , the mass ratio @xmath416 , where @xmath417 is the mass of the secondary , the semi - major axis @xmath178 of the orbit , and the orbital eccentricity @xmath418 . for both single stars and binary system primaries
, we use the initial mass function adopted from kroupa , tout & gilmore ( 1993 ) and kroupa & weidner ( 2003 ) , @xmath419 where parameter @xmath420 , with our standard choice being 2.7 for field populations and 2.35 for cluster populations .
stars are generated within an initial mass range : @xmath421 , and the range is chosen accordingly based on the targeted stellar population .
for example , ns studies would require evolution of single stars within range @xmath422 while the formation of wds would require an initial range @xmath423 .
binary evolution , due to mass transfer events ( both mass accretion and mass loss ) may significantly broaden any of the ranges mentioned above .
we assume a flat mass ratio distribution , @xmath424 in the range @xmath425 in agreement with recent observational results of kobulnicki , fryer & kiminki ( 2006 ) .
however , it should be noted that massive binaries may form with components of comparable mass ( pinsonneault & stanek 2006 ) , and we will test this alternative mass ratio distribution in our parameter studies .
given the value of the primary mass and the mass ratio , we obtain the mass of the secondary @xmath426 . the distribution of initial binary separations is assumed to be flat in the logarithm ( abt 1983 ) , @xmath427 where @xmath178 ranges from a minimum value , such that the primary fills at most 50% of its roche lobe at zams , up to @xmath428 . finally , we adopt the thermal - equilibrium eccentricity distribution for initial binaries , @xmath429 in the range @xmath430 ( e.g. , heggie 1975 ; duquennoy & mayor 1991 ) .
in the following subsections we present startrack mass transfer calculations and compare them to published and unpublished results based on the use of stellar evolution and mass transfer codes .
we choose this rlof sequence from wellstein et al .
( 2001 , their model b ) and start with a @xmath277 + @xmath278 zams binary in a 8 day circular orbit .
rlof starts after the primary evolves off the ms .
the system at the onset of rlof ( @xmath431 myr since zams ) is characterized by : @xmath432 .
the evolution of the system during the rlof phase is shown in figure [ case1 ] .
the rlof phase proceeds on the thermal timescale of the donor , which is rapidly expanding while crossing the hertzsprung gap .
first , there is a phase characterized by very high mass transfer rates ( @xmath433 ) , until the mass ratio is reversed and the donor becomes the less massive binary component .
shortly thereafter , the transfer rate slowly decreases ( @xmath434 ) .
the mass ratio is reversed just right after the onset of rlof and the orbit expands .
our assumption is that all ( 100% ) of the transferred material is accreted by the companion ( conservative evolution is adopted for a better comparison with wellstein et al .
2001 calculation ) .
rlof terminates when the envelope of the donor is nearly exhausted and its radius contracts below the roche lobe radius , thereby , causing the system to become detached .
the primary loses most of its mass and becomes a core helium burning star ( @xmath435 ) , while the secondary gains mass and is rejuvenated ( @xmath436 ) .
the orbital period increases to reach @xmath437 days at the end of the rlof phase .
both stars continue to evolve in a detached configuration .
eventually , the primary becomes a naked helium star ( @xmath438 ) that evolves and loses some mass ( @xmath439 ) until the final explosion and forms a neutron star ( @xmath440 ) .
since the mass of the primary is quite large during the helium burning stage it does not significantly expand and in particular it does not initiate another rlof .
the helium star primary reaches a maximum radius of @xmath441 just prior the explosion , while its roche lobe radius is @xmath442 . after
the explosion the system is disrupted due to a large natal kick the neutron star receives ( e.g. , hobbs et al . 2005 ) . at this point the secondary is still on the main sequence ( @xmath443 ) , and will eventually ( in @xmath118 myr ) form a single black hole ( @xmath444 ) . the calculation of wellstein et al .
( 2001 ) shows similar behavior during the first rlof phase in terms of the duration , mass transfer rate , and orbital period .
the final donor and accretor masses in both simulations are almost the same . therefore , our calibration and rates used for thermal and then nuclear timescale rlof are in very good agreement with the detailed evolutionary calculation of wellstein et al .
( 2001 ) .
however , wellstein et al .
( 2001 ) find a second rlof phase , when the primary expands again , once it becomes a helium giant ( @xmath445 ) after it loses significant amount of mass in a stellar wind : @xmath446 at helium star formation and at the start of the second ( case bb ) mass transfer phase , respectively .
their system evolves through the , so called , case bb mass transfer phase .
the rlof stops when the primary loses a significant part of its helium envelope .
the system becomes wider , and eventually the primary explodes in type ib / c supernova forming a neutron star .
the difference in the evolution into a second rlof phase is explained by the difference in modeling the helium star evolution , and in particular the stellar winds .
the helium star primaries in both calculations start with about the same mass ( @xmath447 ) and their lifetimes are similar ( @xmath448 myr ) .
however , in the case of wellstein et al . (
2001 ) the helium star loses significant amount of mass : @xmath449 , while our star loses only half of that : @xmath450 . therefore
, our helium star remains massive and does not significantly expand to initiate the second rlof as it is found by wellstein et al .
the reason for this difference in wind mass loss from helium stars is that we use the downward revised empirical wind mass loss rates ( hamann & koesterke 1998 ; see also hurley et al .
2000 ) that take into account wind clumping and predict rates lower by factor of @xmath118 than previously estimated ( hamann schonberner , & heber 1982 ) .
this rlof sequence is selected from wellstein et al .
( 2001 , their model a ) .
we start with a @xmath451 + @xmath452 zams binary in a 2.5 day circular orbit .
rlof starts while the primary still evolves through the ms phase .
the system at the onset of rlof ( @xmath453 myr since zams ) is characterized by : @xmath454 .
the evolution of the system during the rlof phase is shown in figure [ case2 ] . in this calculation
we invoke conservative evolution ( @xmath455 ; all mass lost from donor is accreted by the companion ) to match the assumption in wellstein et al .
( 2001 ) .
_ first phase .
_ at first , the rlof proceeds on the thermal timescale with a mass transfer rate of @xmath456 , through the so called rapid case a transfer phase .
the transfer rate then rapidly decreases by more than 2 orders of magnitude until the component masses are nearly equal .
subsequent evolution proceeds on the much slower nuclear timescale of the donor with transfer rates below @xmath345 .
rlof continues until the final stages of the donor ms lifetime , when the primary contracts and detaches from its roche lobe .
the evolution of the orbital period is characterized by an initial small decrease and then ( after the thermal timescale phase has ended ) a slow but rather constant increase up to 3.5 days . at that point
the primary mass is @xmath457 and the secondary mass @xmath458 .
_ second phase .
_ after @xmath459 myr the primary starts expanding as it enters the hertzsprung gap and rlof restarts .
this mass transfer phase is much more rapid and is driven by expansion of the primary .
this phase is characterized by high mass transfer rates ( @xmath460 ) and the envelope of the primary is soon ( @xmath461 myr ) exhausted , ending the second rlof phase . during this relatively short phase ,
the orbit expands significantly ( final orbital period @xmath462 days ) , while the primary loses most of its mass ( @xmath463 ) while the secondary , still on its ms , gains mass ( @xmath464 ) and is rejuvenated .
the dramatic orbit expansion is an effect of the rather extreme mass ratio for this system at the time of the second rlof . for both rlof phases
conservative evolution was applied .
the evolution of this system ends when the massive secondary evolves off ms , initiating a ce phase while crossing the hertzsprung gap .
this phase leads to inspiral and merger .
the calculation of wellstein et al .
( 2001 ) shows a qualitatively similar system behavior during both rlof phases ; initial high mass transfer phase , then a slower one , short break in rlof followed by another rapid phase while the donor evolves off the ms .
also the mass transfer rates are comparable as it is duration of the second rlof phase . however , there is a difference in the duration of the first rlof phase , our calculation showing factor of @xmath465 longer rlof phases than that of wellstein et al .
the initial rapid ( on a thermal timescale ) phase of rlof lasts longer in the wellstein et al .
( 2001 ) calculation , and the mass transfer rates are slightly different than in our approximations . these result in a somewhat different mass evolution of the binary components ( e.g. , our secondary has reached @xmath466 at the end of first rlof phase while in wellstein et al . (
2001 ) calculation it ends up with @xmath467 ) , and therefore different mass ratio of the system that in turn influences the subsequent mass transfer calculations and period evolution .
additionally , we include spin - orbit coupling in our calculations . the period of our system after the second rlof is longer ( 260 days ) than @xmath468 days found by wellstein et al .
( 2001 ) .
we note that wellstein et al .
( 2001 ) account for the rejuvenation effects in detail , given that they use a stellar structure code and do not have to assume full rejuvenation as it is adopted by tout et al .
( 1997 ) , hurley et al .
( 2002 ) , and our code implementation .
wellstein et al .
( 2001 ) remark that had full rejuvenation been assumed in their calculation the system would have ended in a ce merger during the expansion phase of the secondary after ms evolution in agreement with our findings .
this calculation starts with a @xmath469 bh + @xmath470 zams star .
we let the secondary evolve through about half of its ms lifetime before bringing the system into contact at @xmath471 myr ( counted from the secondary zams ) .
the system at the onset of rlof is characterized by : @xmath472 .
the evolution of the system during the rlof phase is shown in figure [ case3 ] . _
first phase .
_ rlof is stable and proceeds on the nuclear timescale of the secondary with a mass transfer rate of @xmath473 .
since this rate is sub - eddington we allow all the transferred material to be accreted onto the primary bh , which increases its mass to @xmath474 , while the secondary mass decreases to @xmath475 . during this phase , the period increases from 1 to 2 days .
the phase ends when the secondary begins contraction at the end of its ms life .
_ second phase .
_ rlof restarts when the secondary crosses the hertzsprung gap with mass transfer proceeding at the high rate ( @xmath476 ) corresponding to rapid expansion of the star on its thermal timescale during that phase . at some point
the donor starts ascending along the red giant branch , and the transfer rate drops by about an order of magnitude to @xmath477 . since the transfer rate is super - eddington throughout this entire phase we limit accretion onto the bh to the eddington rate , allowing the rest of the material to leave the binary with the specific orbital angular momentum of the bh . in the end
the bh has increased its mass to @xmath478 and the mass of the donor has decreased to @xmath479 .
the orbit expands significantly ( @xmath480 days ) during this rapid rlof phase .
the rlof phase ends at the point when the donor , due to the loss of its almost entire h - rich envelope , stops its expansion .
the system ends its life as a wide bh - wd binary .
the same rlof sequence was calculated with the detailed stellar evolution code of ivanova et al .
( 2003 ; also see ivanova & taam 2004 ) .
the comparison of the two phases of rlof shows overall qualitative agreement with the startrack calculation .
the mass transfer rates are virtually the same : @xmath481 , for first and second phase , respectively .
however , the detailed calculation with the evolution code shows a longer duration ( by a factor @xmath118 ) for the first rlof phase .
this calculation starts with a @xmath157 bh + @xmath482 zams star .
we let the secondary evolve through about one third of its red giant lifetime before bringing the system into contact at @xmath483 myr ( counted from the secondary zams ) .
the system at the onset of rlof is characterized by : @xmath484 .
the evolution of the system during rlof phase is shown in figure [ case4 ] .
rlof is stable and proceeds through the entire rg phase ( @xmath485 ) on the nuclear timescale of the donor .
the mass transfer rate is sub - eddington and thus the material transferred to the bh is entirely accreted . in the end
the mass of the bh is increased to @xmath486 while the mass of the donor is decreased to @xmath487 .
as the donor expands , ascending the rg branch , the orbit expands as well , and finally the rlof phase terminates at an orbital period of @xmath488 days .
the phase ends when the donor contracts upon igniting helium in its core .
the system eventually forms a wide bh - wd binary .
this rlof sequence was also calculated with the detailed stellar evolution code of ivanova et al .
the mass transfer rates found in both cases are similar ( @xmath489 ) and in this case the startrack timescales are shorter , but do not differ by more than 50% .
this rlof sequence is chosen from tauris & savonije ( 1999 , their example 2b ) .
we start with a @xmath490 ns + @xmath491 zams star in a 3 day circular orbit .
rlof starts while the secondary is on the rg branch ( @xmath492 myr since secondary zams ) and the binary is described by : @xmath493 .
the evolution of the system during the rlof phase is shown in figure [ case5 ] . at first
the rlof proceeds on a thermal timescale with a highly super - eddington mass transfer rate ( @xmath476 ) .
after the donor becomes less massive than its accretor , the mass transfer is driven by the expansion of the red giant donor ( on its nuclear timescale ) at a much smaller rate of @xmath494 .
as the mass transfer rate decreases , the ns starts to accrete efficiently and its mass increases to @xmath495 . eventually , after @xmath496 myr of rlof , the rg secondary loses most of its mass ( @xmath497 ) and contracts , leaving a remnant helium wd . at this point
the rlof phase ends ( orbital period 60 days ) , and further evolution leads to the formation of wide binary , with a recycled pulsar .
comparison with the detailed evolutionary calculation of tauris & savonije ( 1999 ) shows good agreement between the results .
the detailed calculations show an initial rapid rlof phase followed by a sub - eddington mass transfer phase , eventually leading to the formation of ns - he wd binary .
final component masses ( ns and donor : @xmath498 and @xmath499 , respectively ) are very similar to the ones obtained with startrack . the final orbital period of 42 days obtained by tauris & savonije ( 1999 ) is shorter than in our calculation ( 60 days ) .
in addition , there is a difference in the duration of rlof phase , lasting 123 myr in the tauris & savonije ( 1999 ) model , as compared to 60 myr in our calculations .
this may be understood in terms of a different treatment of binary interactions ( tides , magnetic braking , winds ) as well as the difference in stellar models which may lead to a different starting point of rlof .
this rlof sequence is taken from tauris & savonije ( 1999 , their example 2c ) .
we start with a @xmath490 ns + @xmath500 zams star in a 60 day circular orbit .
rlof starts while the secondary is on the rg branch ( @xmath501 myr since secondary zams ) and the binary is described by : @xmath502 .
the evolution of the system during rlof phase is shown in figure [ case6 ] .
rlof is highly super - eddington and driven by the expansion of the donor on a nuclear timescale .
only shortly before the system detaches as a result of the exhaustion of the donor s envelope , the transfer rate becomes sub - eddington . as a result
, the donor loses most of its mass ( @xmath503 ) while the ns hardly accretes any material ( @xmath504 ) .
the orbit expands throughout this phase with the orbital period increasing to over 300 days .
the system eventually forms a wide ns - he wd binary , with a potential recycled pulsar ( the ns has accreted @xmath505 ) .
the above results are very similar to the calculations of tauris & savonije ( 1999 ) , who obtain a @xmath506 ns with a @xmath507 ns - he wd binary in a 382 day orbit .
the mass transfer rates and duration of the rlof phases are similar in both calculations .
this rlof sequence follows from dewi & pols ( 2003 , see their fig . 1 ) .
we start with a @xmath349 ns + @xmath508 zams he star in a 10 day circular orbit .
rlof starts while the secondary is already an evolved he star ( @xmath509 myr ) and the binary is described by : @xmath510 .
the evolution of the system during rlof phase is shown in figure [ case7 ] .
rlof proceeds on the donor s thermal timescale throughout the entire phase .
the very high mass transfer rate ( @xmath511 ) makes this phase very short and rlof stops after the envelope of he star is exhausted .
since the mass transfer rate is highly super - eddington , the ns hardly accretes any material while the donor loses its entire he - rich envelope ( @xmath512 ) . the orbital period at first decreases to reach a minimum at 8.9 days , and then increases to 9.5 days at the end of rlof phase .
after the phase of rlof the secondary core explodes in sn ic leading to double neutron star formation ( provided that a natal kick does not disrupt the binary ) .
this result was presented also in ivanova et al .
( 2003 ) .
dewi & pols ( 2003 ) calculated mass transfer rates spanning the range : @xmath513 .
our rate is constant and close to the high end of the dewi & pols ( 2003 ) range .
we have adopted a constant mass transfer rate following paczynski ( 1971 ) who pointed out that thermal timescale mass transfer rates do not vary by more than factor of 2 - 3 ( for details see 5.3 ) .
this system may appear as an x - ray binary during this phase .
however , the chances of catching it at this phase are very small , since the thermal timescale mass transfer is very short . besides , in this case the x - rays may be significantly degraded because of high optical depths ( material shed out of the system ) . on the other hand , some of these sources might appear to be soft @xmath300-ray emitters ( i.e. @xmath514 kev range , tail of x - ray emission ) with high intrinsic absorption , and the discovery of objects with these broad characteristics ( see e.g. , dean et al .
2005 ) lends some hope for detecting this phase of binary evolution .
the results from dewi & pols ( 2003 ) reveal a different period evolution than in our simulation ; rlof starts at higher value ( 10.46 days ) , and then decreases to 10.37 days .
however , the period changes in both calculations are rather small , and are probably related to our consistently high mass transfer rate throughout the rlof phase .
this leads to higher mass and angular momentum loss from the binary which determines the orbit evolution .
additionally , we include tidal interactions between binary components ( see 3.3 and 8.2 ) .
these differences between models for low mass helium stars were already noted by dewi & pols ( 2003 ) .
we choose this rlof sequence from dewi & pols ( 2003 , see their fig .
we start with a @xmath349 ns + @xmath515 zams he star in a 0.6 day circular orbit .
rlof starts while the secondary is already an evolved he star ( @xmath516 myr ) and the binary is described by : @xmath517 .
the evolution of the system during rlof phase is shown in figure [ case8 ] .
the rlof proceeds on the donor s thermal timescale with a mass transfer rate of @xmath518 until the envelope of he star is almost exhausted . since the mass transfer rate is highly super - eddington , the ns does not accrete much material while the donor loses most of its he - rich envelope ( @xmath519 ) .
the orbital period decreases from 0.6 to @xmath520 days at the end of rlof phase .
after the rlof phase ceases the secondary explodes in sn ib / ic leading to a close double neutron star system ( again provided that a natal kick does not disrupt the binary ) .
the dewi & pols ( 2003 ) rlof sequence for this case is very similar to our calculation .
they find a period decrease ( from 0.65 to 0.47 days ) and a high constant mass transfer rate of a few @xmath521 . the inspiral phase and ce is not expected in this case , and therefore further evolution may lead to a close double neutron star formation .
whenever coeval binary populations in nearby clusters are observed to constrain the circularization rate , it is found that standard tidal dissipation theories do not match the data ( see meibom & mathieu 2005 for a recent review ) . in all cases
an increase in the tidal dissipation rate appears necessary ( claret & cunha 1997 ; terquem et al .
depending on which theory is used , the increase needed in the overall efficiency of tidal dissipation is by a factor @xmath522 .
we have used startrack models to calibrate our theoretical treatment by comparing them against observations of _ ( i ) _ the cutoff period for circularization in a population of ms binaries ( in m67 ) , and _
( ii ) _ the orbital decay accompanying tidal synchronization in a high mass x - ray binary ( lmc x-4 ) .
the results , presented in two following subsections , confirm that tidal dissipation , at least in case of convective stars , is more effective than predicted by our simple theory .
therefore , in all our standard model calculations , we will use an increased rate of tidal dissipation for convective stars , corresponding to @xmath523 , while using standard dissipation for radiative stars @xmath524 , but we will also allow for even more effective tidal dissipation rates in our parameter studies ( all the way to @xmath525 and @xmath526 ) .
see 3.3 for our implementation of tidal dissipation theory and the definition of @xmath173 .
open star clusters have often been used to test tidal interaction theories ( mathieu et al .
1992 ; meibom & mathieu 2005 ) . observations of single- and double - line spectroscopic binaries allow estimates of the periods and eccentricities for a number of systems within clusters .
it was expected and then confirmed that the cutoff period ( @xmath527 , the longest period of a circular binary ) should increase with the age of the cluster .
the tidal dissipation depends strongly on the orbital separation and therefore the wider , longer - period binaries will take a greater time to circularize . in principle , with knowledge of the initial conditions in a given cluster , the observed value of the cutoff period may be used to calibrate the efficiency of tidal interactions . in practice , binaries within a given cluster form with eccentricities , separations and angular momenta which are not precisely known .
in addition , the observed samples may suffer from small number statistics ( the observed cutoff periods are only lower limits ) , rendering such a calibration quite uncertain .
however , we can use the cutoff - period observations to provide at least an order of magnitude estimate for the factor by which any standard theoretical estimate must be increased .
m67 is an old open cluster with an age of @xmath528gyr and observed cutoff period of @xmath529d ( mathieu , latham & griffin 1990 ; mathieu et al .
1992 ) and a solar metallicity stellar population ( janes & phels 1994 ) .
the period was estimated for a sample of ms binaries with components close to the cluster turnoff mass .
recently meibom & mathieu ( 2005 ) proposed a new way to estimate the point of transition from circular to eccentric systems . instead of a simple cutoff period , they use a new estimator called the `` tidal circularization period . ''
this period is found from fitting a special function which mimics the tidal circularization isochrone of the most frequently occurring eccentric binary orbits for a given cluster .
they find that the tidal circularization period for m67 is @xmath530d .
several calculations , with different efficiencies of tidal dissipation , were performed to try to reproduce the binary population of the open cluster m67 . in each calculation
we have evolved @xmath531 binaries at solar metallicity with component masses in the @xmath359 range , requiring that the mass ratio be greater than 0.5 .
the limits are somewhat arbitrary , but chosen to include the population of bright ms stars observed in m67 .
most of these stars have convective envelopes , and therefore we try to calibrate the scaling factor for convective envelopes ( @xmath532 ) while keeping the one for radiative stars constant ( @xmath524 ) .
the initial distributions were chosen as in our standard evolutionary model ( see 5.7 ) , but with imf exponent @xmath533 , which is more appropriate for clusters ( kroupa & weidner 2003 ) . in figure [ m67 ]
we show synthetic binary ms populations in the period eccentricity plane corresponding to an evolution with different efficiencies for the tidal interaction .
as expected we see that the cutoff period increases for more efficient tidal interactions , @xmath534d for @xmath535 , respectively .
it is found that only for significantly increased dissipation ( @xmath536 ) the the predicted cutoff period approach the observed value of 10 - 12 days .
an additional calculation with @xmath537 results in a cutoff period of @xmath538 days , now clearly higher than the observed value .
levine , rappaport & zojcheski ( 2000 ) measured an orbital period decay for the high mass x - ray binary ( hmxb ) lmc x-4 .
the system consists of a @xmath490 ns and a massive @xmath539 companion in a 1.4-day circular or almost circular orbit ( woo et al .
1996 ; van der meer et al .
the x - ray emission in hmxbs is believed to arise from wind accretion onto the compact object ; however it was also suggested that some systems may be in an atmospheric rlof phase ( e.g. , kaper 2001 ) . for wind - fed detached systems
, the orbital decay may be directly connected to the tidal interaction of the hmxb components .
the secondary is a massive star , and the source of tidal dissipation is radiative damping .
therefore , we use lmc x-4 to check the efficiency of tidal interactions for radiative envelopes ( @xmath540 ) . the rotation of the massive component decreases with time as it expands during its evolution . on the other hand , the tidal forces act to synchronize the massive component , resulting in loss of orbital energy and angular momentum , i.e. , decay of the orbit .
if , in fact , lmc x-4 is a wind - fed system and not in rlof , then the massive star must be smaller than its roche lobe @xmath541 .
a @xmath539 star exceeds that size , while still on ms , after about @xmath542myr of evolution ( from the zams ) .
subsequent rlof is dynamically unstable ( extreme mass ratio ) and leads to a rapid merger of the binary components , terminating the hmxb phase .
we perform a set of calculations for a synthetic binary similar to the lmc x-4 using our standard model parameters , with a metallicity appropriate for the lmc ( @xmath543 ) .
we _ assume _ that the binary configuration is detached and we calculate the rate of orbital decay .
the orbital decay rate depends crucially on the current relative radius of the massive component of lmc x-4 ( @xmath544 , see eq .
[ tid1 ] ) .
the radius of the @xmath539 star ( @xmath543 ) increases from @xmath545 on zams to @xmath546 at the end of ms phase , which takes @xmath547 myr .
based on the roche lobe radius of secondary for 1.4 day orbit the secondary fills its roche lobe at @xmath548 myr , which is close to the end of ms phase . since the primary has already evolved and has formed a ns , a significant amount of time must have elapsed since the binary formation .
for example a @xmath549 star takes @xmath550 myr to form compact object , and such a massive star would have formed a ns only if stripped of a significant part of its mass in rlof episode . for more massive primaries ,
the evolution would be slightly faster ( @xmath551 myr ) , but they would more likely have formed bhs .
so on one hand the secondary can not be older than @xmath548 myr ( @xmath552 and @xmath553 ) and most likely it is not younger than @xmath550 myr ( @xmath554 and @xmath555 ) .
we conclude that the secondary is in the late stage of its evolution on the main sequence and probably close to filling its roche lobe ( see also levine et al .
2000 ) .
we perform the set of calculations for different radii of the massive component of lmc x-4 ( @xmath556 ) for various efficiencies of tidal interactions ( @xmath557 ) .
the results are shown in figure [ lmcx4 ] .
the orbital decay rate increases with time as the massive component expands along the ms and approaches its roche lobe . the time to reach contact ( at which point calculations are stopped ) decreases with increasing effectiveness of tidal forces . for comparison
we show the observed orbital decay for lmc x-4 , which falls within the model with standard tidal interactions efficiency ( @xmath524 ) .
we conclude that in the case of lmc x-4 there is no need for the increased efficiency of tidal interactions , and therefore we adopt @xmath524 for massive stars with radiative envelopes for our standard model value .
in our study we consider only accreting binaries with ns and bh primaries , which are brighter than some x - ray luminosity cut @xmath558 .
this cut may correspond to a detection limit of a particular set of observations .
typical @xmath558 values for most current _ chandra _ observations are in the range @xmath559 erg s@xmath364 . at these high luminosities in the chandra sensitivity range ( @xmath560 kev )
the only wd accretors will be supersoft sources , which are easily identifiable from their x - ray spectra and are thought to have most of their x - ray emission coming from nuclear burning rather than gravitational energy release ( see kuulkers et al .
2003 for a review of the x - ray properties of wd accretors ) .
although , for some deep galactic exposures _
chandra _ has reached levels of @xmath561 erg s@xmath364 ( e.g. , galactic center image of muno et al .
2003 ) and a contribution from cataclysmic variables may also become important . the calculation of x - ray luminosities of systems with wd accretors is described in a separate study ( ruiter , belczynski & harrison 2006 ) .
binary companions to ns / bhs may lose material either through a stellar wind or via rlof . in the latter case ,
the donors transfer all the material toward the accretor , whereas for the wind - fed systems only a fraction of the material is captured by the compact object .
we calculate the bolometric luminosity ( @xmath562 ) arising from the accretion onto a compact object .
the accretion rate is based on the secular averaged mass accretion rate .
if a system is detached then we use the wind mass accretion rate ( eq .
[ wind02 ] ) , and if system is semi - detached the rlof accretion rate is used ( eq . [ mt10b ] ) .
we do not calculate x - ray luminosities arising from the accretion in dynamically unstable phases , since the timescales are very short and additionally x - ray emission would be highly absorbed due to large optical depths in the ce .
the @xmath562 is calculated from @xmath563 where the radius of the accretor is 10 km for a ns and 3 schwarzschild radii for a bh , and @xmath184 gives a conversion efficiency of gravitational binding energy to radiation associated with accretion onto a ns ( surface accretion @xmath185 ) and onto a bh ( disk accretion @xmath186 ) . for rlof - fed systems we make a distinction between persistent and transient x - ray sources .
all wind - fed systems are considered as persistent x - ray sources .
the issue of the wind - fed xrbs with massive b@xmath418 companions and their outburst behavior is discussed in 9.2 .
rlof - fed systems are subject to a thermal disk instability and may appear either as persistent or transient x - ray sources depending on the mass transfer rate .
a system becomes a transient x - ray source when the rlof rate falls below a certain critical value @xmath564 .
we use the work of dubus et al .
( 1999 ) for h - rich disks ( see their eq.30 ) and the study of menou , perna , & hernquist ( 2002 ) for disks with heavier elements ( see their eqs.14 ) @xmath565 where @xmath566 is accretor mass in @xmath0 , @xmath567 is a maximum disk radius ( 2/3 of accretor roche lobe radius ) in @xmath568 cm .
constants are : @xmath569 , with c being radiation parameter of typical value @xmath570 ; @xmath571 , with @xmath572 being a viscosity parameter . following menou et al . ( 1999 ) we adopt @xmath573 for all types of donors since there is empirical evidence from dwarf nova outbursts that this is the right order of magnitude for the viscosity parameter .
the same value of @xmath572 is used to derive the critical mass transfer rate for h - rich disks ( dubus et al .
h - rich donors are the stars with types @xmath574 , he - rich donors are @xmath575 , co - rich donors are @xmath576 , while we apply formulas for o - rich type donors to one wds ( @xmath577 ) .
we adopt a semi - empirical approach to calculate quiescent x - ray luminosities of transient ns rlof - fed sources , since little is known about the emission mechanism during quiescence .
it is not certain if the emission arises from a low level accretion or a deep crustal heating ( for a detailed discussion see belczynski & taam 2004b , and references therein )
. using x - ray studies of galactic transient systems with ns accretors ( e.g. , tavani & arons 1997 ; rutledge et al .
2001 ; campana & stella 2003 ; jonker , wijnands & van der klis 2004 ; tomsick et al . 2004
; campana 2004 ) , we adopt @xmath578 erg s@xmath364 as a lower limit for the hard x - ray luminosity , above 2 kev .
however , it was shown that the average luminosity level can be higher @xmath579 erg s@xmath364 ( e.g. , rutledge et al .
2002 ; jonker et al .
we adopt an x - ray luminosity level of @xmath580 erg s@xmath364 above 2 kev .
furthermore , we assume that the quiescent ns transient x - ray luminosities are evenly distributed ( in @xmath581 ) in the above range .
the quiescent emission from bh transient systems is likely related to a low level of mass accretion .
recent observations of bh transients in their quiescent states ( tomsick et al .
2003 ) reveal rather hard spectra that are not well described by a black body .
the observed luminosities are found in the range @xmath582 erg s@xmath364 with a median luminosity @xmath583 erg s@xmath364 . for bh systems we also use a semi - empirical approach , and we assume that most ( 80% ) of the quiescent bh transient x - ray luminosities above 2 kev are evenly distributed in the @xmath584 erg s@xmath364 range , while the rest ( 20% ) of the systems are slightly brighter : luminosities evenly distributed in the @xmath585 erg s@xmath364 range ( see fig.3 of tomsick et al .
2003 ) . both of the above distributions are uniform in @xmath581 .
there are some indications that the highest quiescent luminosities are found in the longest period systems ( e.g. garcia et al .
2001 ) , but we do not implement this effect until confirmed by more observations .
rlof - fed transient systems in outburst reach high ( close to eddington ) x - ray luminosities .
we introduce a factor @xmath586 describing the fraction of the critical eddington luminosity a given system has reached .
the long period systems , with orbits that are sufficiently extensive for a large accretion disk to be formed , are usually found to emit at the eddington luminosity ( @xmath587 ) during outburst , while the outburst luminosities of short period systems are lower by about an order of magnitude .
the correction factor to an x - ray luminosity at outburst corresponding to @xmath588 and @xmath589 for the short and long period systems is applied respectively .
the critical periods , above which the eddington luminosity is adopted , are taken to be 1 day and 10 hrs for ns and bh transients in outburst , respectively ( chen , shrader & livio 1997 ; garcia et al .
2003 ; see also appendix a1 in portegies zwart , dewi & maccarone 2004 ) . in order to decide if a given transient system is in an active ( outburst ) state or inactive ( quiescent ) state the disk duty cycle ( @xmath590 ; the fraction of a time a given system spends in the outburst ) must be known .
however , the disk instability theory can not provide a reliable estimate of @xmath590 .
empirically it is thought that @xmath591 ( e.g. , taam , king & ritter 2000 )
. we adopt @xmath592 ( probability of finding a system in outburst ) in our calculations and use monte carlo to decide the state of a transient system . in practice when we study a stellar population the information for all x - ray binaries is extracted at some specified time ( time slice ) .
once a given system is identified as a transient ( see eq .
[ lx03 ] ) a random number ( flat probability distribution ) is drawn from the range 01 .
if the number is smaller than 0.01 ( 1% probability ) the system is then in outburst , otherwise it is in quiescence .
the appropriate x - ray luminosity is then assigned to the system ( see eq .
[ lx05 ] ) .
alternatively , we use a phenomenological model for the duty cycle developed by portegies zwart et al .
the model is based on the observations available for the galactic bh transient systems . in particular comparison of the recurrence time and the decay time combined with the observed peak outburst energy
allows to calculate the time in which system is brighter than a certain critical x - ray luminosity .
specific application of that model will be discussed in the forthcoming paper on the evolution of x - ray luminosity function in starburst galaxies ( belczynski et al .
2006 , in preparation ) .
finally , the bolometric accretion luminosity is converted to an x - ray luminosity in a specific energy range .
we perform the conversion to the 0.3 7 kev range , which may be used directly for comparison with _ chandra _ observations .
for all the persistent rlof - fed sources , all wind - fed sources and the transients in the outburst stage , where accretion is the dominant contributor to the observed luminosity , we apply a bolometric correction ( @xmath593 ) . for all quiescent transients the bolometric correction is not needed since we adopted their x - ray luminosities directly from observations . for different types of systems we estimate the correction to be : @xmath594 corrections were obtained from : la barbera et al .
( 2001 ) for wind - fed ns systems ; from di salvo et al .
( 2002 ) and maccarone & coppi ( 2003 ) for rlof - fed ns systems ; and from miller et al . (
2001 ) for bh systems .
these bolometric corrections will be applicable for the typical _ chandra _ observations of external galaxies .
for deeper observations , where the lower luminosity cutoffs are below a few percent of the eddington limit , the objects make spectral state transitions
( see maccarone 2003 and references within ) , and the bolometric corrections are much larger
. combining all of the above information , we can calculate the x - ray luminosity of synthetic x - ray binaries from @xmath595 where @xmath596 is expressed in @xmath597 and @xmath587 represents the eddington luminosity .
note that the x - ray luminosity is calculated directly from the mass transfer rate only for persistent sources . on the other hand ,
we adopt the above empirical description for transient sources since the relation between the quiescent , outburst , bolometric luminosities , and duty cycle are uncertain due to the mass loss from the system during the outburst state .
evidence for such mass loss in the form of jets and/or wind have been observed in , for example , a galactic bh transient grs 1915 - 105 ( dhawan , muno & remillard 2005 ; truss & done 2006 ) .
high mass x - ray binaries consist of a compact object ( either a ns or a bh ) orbiting a massive star .
both galactic and extra - galactic populations of hmxbs are known ( liu , van paradijs & van den heuvel 2000 , 2005 ) .
the majority of hmxbs ( about @xmath598 ; see liu et al .
2000 , 2005 ; hayasaki & okazaki 2005 ) are so - called be / x - ray binaries , in which the primary is a be star , orbiting a magnetized ns .
orbits are generally wide with a moderate eccentricity .
the compact star accretes from the wind of a massive main sequence or subgiant be ( spectral types b3-o with balmer emission lines ; zorec & briot 1997 ) companion .
many of these systems show transient behavior ( see below ) .
the remaining hmxbs are those in which the primary is a supergiant , so called sg / x - ray binaries ( e.g. , liu et al .
2000 ) . for these systems
the compact object either accretes from the wind of the supergiant , or in brighter systems through rlof ( possibly atmospheric but not always ) via an accretion disk .
if indeed some hmxbs are confirmed to be evolving through stable rlof , it should pose a useful constraint on the development of a delayed dynamical instability .
in general , it is expected that mass transfer from a much more massive donor to a low mass companion is dynamically unstable and leads to the formation of a ce ( see 5 ) that ends hmxb phase .
it has been shown that if a h - rich donor is @xmath465 times more massive than a compact star accretor ( see 5.1 ) the rlof will lead to ce phase . for
adopted the maximum ns mass adopted here ( @xmath599 ) we predict that only stars of spectral type later than b3 ( masses smaller than @xmath452 ) could be in dynamically stable rlof with ns accretors . if a higher mass donor is found in a hmxb with a solid case for ongoing rlof , then either _
( i ) _ compact object mass is higher ( e.g. bh ) , or _
( ii ) _ the system is in the phase of short - lived atmospheric rlof and will soon end up in ce phase , or _
( iii ) _ the understanding of development of dynamical instability is incomplete and the observations could be used to set new limits . some be / x - ray binaries ( b@xmath418 xrbs ) are persistent sources ( varying by less than a factor of @xmath600 ) observed at low luminosity levels @xmath601 erg sec@xmath364 ( e.g. , van bever & vanbeveren 2000 ; okazaki & negueruela 2001 ) .
however , most b@xmath418 xrbs show periodic outbursts and are called transient b@xmath418 xrbs .
transient b@xmath418 xrbs exhibit two different types of outbursts ( e.g. , bildsten et al .
1997 ; okazaki & negueruela 2001 ; hayasaki & okazaki 2005 ; baykal et al .
2005 ) : + type i ( normal ) outbursts , which are of moderate intensity ( @xmath602 erg sec@xmath364 ) and they appear to be related to the orbital period .
it is generally accepted that these outbursts are associated with the periastron passage of a ns , and are explained by the increased accretion from the b@xmath418 star wind at periastron .
+ type ii ( giant ) outbursts , with luminosities reaching @xmath603 erg sec@xmath364 , are irregular , and although they seem to appear shortly after the periastron passage , they do not exhibit any other correlations with the orbital period .
although the origin of the type ii outbursts remains unknown , it was suggested that the outflow from the b@xmath418 star may lead to the formation of a transient accretion disk around the ns .
disk accretion results in higher x - ray luminosities than direct surface wind accretion ( see bildsten et al .
1997 for a discussion and references ) .
some systems show both types of outbursts , e.g. , a 0535 + 262 ( motch et al .
1991 ; finger , wilson & harmon 1996 ) , v0332 + 53 ( stella , white & rosner 1986 ) or 4u 0115 + 634 ( baykal et al . 2005 ) .
type i outbursts are averaged out of our calculations if we use the orbit - averaged wind accretion model ( see 4.2 ) . in the general ( arbitrary eccentricity )
wind accretion model ( see 4.1 ) type i outbursts are a natural outcome . however , it was noted ( avni & goldman 1980 ) that the transient phenomenon may be difficult to explain .
we construct a simple phenomenological model for type ii outbursts in order to be able to assess the influence of this transient activity on xrb population characteristics . for a system to be a potential type ii b@xmath418 xrb outburster
we require : + a binary with a ns or a bh accretor and a massive ms ( @xmath310 ) or subgiant ( @xmath604 ) donor ( @xmath605 , spectral type earlier than b3 ) , + that the system is tight enough so it appears as a hmxb with a persistent ( outside outbursts ) wind accretion leading to an x - ray luminosity greater than @xmath606 .
we allow @xmath606 to change within the range @xmath607 erg sec@xmath364 .
furthermore , only a fraction ( @xmath608 ) of donors in the above binaries are b@xmath418 stars ( as opposed to a regular b stars ) , and can potentially trigger the type ii outbursts . to provide an upper limit on the contribution of bursting hmxbs to the xrb population one may choose @xmath609 . for detail studies ,
the value of @xmath608 may be constrained based on the age of a massive star ( mcswain & gies 2005 ) or its spectral type and luminosity class ( zorec & briot 1997 ) . since little
is known about the duty cycle of type ii outbursts , we allow the duty cycle to change within a wide range @xmath610 and use monte carlo to decide whether the system is in outburst or in quiescence . here , @xmath611 gives the fraction of a time a given system spends in the outburst .
an orbit averaged x - ray luminosity ( direct wind accretion ) is used for quiescence ( @xmath612 9.1 ) , although thermal emission from a ns is also observed in some systems . for systems in the type ii outburst
the x - ray luminosity is taken to be uniformly distributed in the range @xmath613 erg sec@xmath364 .
we adopt bolometric correction factors : @xmath612 for ns and bh accretors , respectively ( see 9.1 )
. the x - ray modeling will be further developed as we proceed with the studies of the galactic and extragalactic x - ray binary populations ( e.g. , belczynski et al .
2006 , in preparation ) .
we have presented a detailed description of the updated startrack population synthesis code .
the code is being used to study populations of different varieties of binaries hosting compact objects .
the code has been calibrated and tested against different sets of observations and detailed evolutionary calculations and the results are presented here . the updated version of startrack was already used in several studies of compact object binaries and xrbs .
startrack allows for evolution of stellar systems with a wide variety of different initial conditions ( imf , metallicity , star formation history ) and for a number of different evolutionary models , subject to the parametrization of the input physics .
the startrack code can be compared to the bse population synthesis code ( hurley et al .
startrack incorporates the same single star evolutionary formulas ( hurley et al . 2000 ) as the bse code , however we extend the original formulas to _ ( i ) _ include wind mass loss rates from low- and intermediate - mass main sequence stars ( formation of pre - lmxbs , belczynski & taam 2004b ) ; _ ( ii ) _
account for the late evolution of low - mass helium stars ( important for formation of double neutron star systems , see belczynski et al .
2007 and references therein ) ; and _ ( iii ) _ calculate final masses of neutron stars and black holes , based on recent hydrodynamical calculations ( e.g. , belczynski et al . 2004b ) . for the treatment of tidal interactions we use the same equations ( odes ) as in hurley s code , but we employ a fifth - order runge - kutta scheme with truncation error monitoring and adaptive step - size control integration of the odes instead of simple multiplication of the derivatives by the evolutionary timestep ( euler method ) .
we also adopt convective tides that are more efficient ( by a factor of 10 - 100 ) based on the observational calibration discussed in section 8.2 .
this will have a significant effect on the evolution of close binaries with low mass ( convective ) stars .
the calculation of x - ray luminosities for transient systems with ns and bh accretors is much more comprehensive in startrack .
we use both the recent theoretical work and observations of low- and high - luminosity x - ray sources , to calibrate and test our approach ( belczynski & taam 2004b ; belczynski et al .
the compact object masses formed in core collapse are calculated differently , and in particular we account for possibility of direct bh formation ( no natal kick , no mass loss ) , with maximum bh masses formed reaching @xmath614 depending on metallicity and adopted wind mass loss rates ( e.g. , belczynski et al .
2002 ; belczynski et al .
2004b ) , a result that is consistent with maximum bh mass estimates in galactic bh binaries ( e.g. , @xmath615 for grs 1915 ; @xmath616 for cyg x-1 ; see orosz 2003 ) in contrast , in the bse code all compact objects ( including bh ) have masses below @xmath617 for the entire spectrum of initial progenitor masses and different metallicities ( see fig.20 of hurley et al .
2000 ) , a result that can not be reconciled with the current estimates of bh masses .
also more recent ( arzoumanian et al . 2002 ; hobbs et al . 2005 ) natal kick distributions are used here as compared to lyne & lorimer ( 1994 ) in bse .
the above will affect the post - sna binary orbit , and subsequent evolution of massive binaries .
for example , the effect of natal kicks on population of double compact objects is rather dramatic and was quantified in belczynski et al .
( 2002c ) .
the treatment of nuclear mass transfer rate is different in the two codes . in bse
it is calculated using a formula that keeps the donor star within its roche lobe .
the formula is calibrated to keep the mass transfer steady . in startrack , we use the radius - mass exponents for the donor and its roche lobe along with an estimate of the evolutionary donor radius change with time to calculate the mass transfer rate ( see sec .
the calibration of the bse prescription is not discussed in detail by hurley et al .
ruiter et al . ( 2006 ) find that for intermediate polars ( low mass main sequence donors with wd accretors ) the bse code results in mass transfer rates of about 2 orders magnitude lower ( as calculated by liu & li 2006 ) than the rates predicted by startrack and the observations of intermediate polars may indicate ( muno et al .
2006 ) . for low mass binaries ,
we use a different ( less efficient ) magnetic braking law in our standard model . as a result , binary orbits in our model will tend to take longer time to decay and initiate mass transfer , as compared to bse models .
the startrack prescription of mass accumulation onto white dwarfs is quite unique ( see 5.7 ) .
related results of calculations for accretion induced collapse and ns formation were presented by belczynski & taam ( 2004a ) and progenitor models of sn ia by belczynski et al .
( 2005b ) .
on the other hand , the bse code is more fitted to work with dynamical codes , following in detail merger products ( and their evolution ) of various types of binary components . also , the bse code is much faster than the startrack code , and therefore may be used for simulations of larger stellar populations . in a series of papers that will follow we will address the issues of modeling of xrbs , and will focus on the comparison of synthetic xrb populations with the observed x - ray point source populations in nearby galaxies .
the code is also being used to study populations of binaries with nss and bhs as potential source candidates for ground based interferometric gravitational radiation observatories ( e.g. , geo , ligo , virgo ) as well as populations of less - massive wd binaries for space - based projects ( e.g. , lisa ) .
although a number of physical processes governing single and binary evolution remain highly uncertain , the advances in observational techniques and new results of massive surveys allow now various aspects of stellar evolution to be explored .
we have incorporated several different evolutionary models within startrack ( e.g. , different magnetic braking laws or ce prescriptions ) making possible tests of their validity .
for example , one such test may be based on a comparison of synthetic and observed x - ray luminosity functions for nearby starburst galaxies .
the startrack code described in this paper may be used only for the evolution of isolated stars and binaries , i.e. , in stellar systems in which dynamical interactions are not important ( e.g. , field populations , open clusters ) . however
, a number of interesting studies may be carried out for dense stellar environments , in which both stellar evolution and dynamical interactions play an important role in the formation of compact object binaries . in particular ,
startrack was integrated with a dynamical code for these types of studies ( for details see ivanova et al .
2005 ) .
we would like to thank the anonymous referee for very detailed and insightful reports that have helped us to improve our paper considerably .
also , we would like to thank r.webbink , g.nelemans , t. di salvo , j. sepinsky and a.ruiter for useful discussions and comments on the manuscript .
kb and tb acknowledge partial support through kbn grant 1p03d02228 and 1p03d00530 , and kb acknowledges support through nasa chandra theory grant tm6 - 7006x .
vk acknowledges support through a david and lucile packard foundation fellowship in science and engineering and through nasa grants nag5 - 13056 and nas8 - 03060 .
in addition , this research was supported in part by the nsf under grant no .
ast-0200876 to rt .
abramowicz , m.a . ,
czerny , b. , lasota , j.p .
, & szuszkiewicz , e. 1988 , , 332 , 646 abt , h. a. 1983 , ara&a , 21 , 343 andronov , n. , pinsonneault , m. , & sills , a. 2003 , , 582 , 358 arzoumanian , z. , chernoff , d. f. , & cordes , j. m. 2002 , , 568 , 289 avni , y. , & goldman , i. 1980 , , 90 , 44 bailyn , c.d . , & grindlay , j.e . 1990 , , 353 , 159 baykal , a. , kiziloglu , u , kiziloglu , n. , balman , s. , & inam , s.c .
2005 , , 439 , 1131 begelman , m.c . , 2002 , , 568 , l97 belczynski , k. , benacquista , m. , larson , s. , & ruiter , a. 2005a , , submitted ( astro - ph/0510718 ) belczynski , k. , bulik , t. , & kalogera , v. 2002a , 571 , l147 belczynski , k. , bulik , t. , & rudak , b. 2002b , , 571 , 394 belczynski , k. , bulik , t. , & ruiter , a. 2005b , , 629 , 915 belczynski , k. , & kalogera , v. 2001 , , 550 , l183 belczynski , k. , kalogera , v. , & bulik , t. 2002c , , 572 , 407 ( bkb02 ) belczynski , k. , kalogera , v. , zezas , a. , & fabbiano , p. 2004a
, , 601 , l147 belczynski , k. , sadowski , a. , & rasio , f. 2004b , , 611 , 1068 belczynski , k. , sadowski , a. , rasio , f. , & bulik , t. 2006 , , in press ( astro - ph/0508005 ) belczynski , k. , & taam , r. e. 2004a , , 603 , 690 belczynski , k. , & taam , r. e. 2004b , , 616 , 1159 belczynski , k. , taam , r.e . ,
kalogera , v. , rasio , f. , & bulik , t. 2007 , , accepted ( astro - ph/0612032 ) bethe , h.a . , & brown , g.e .
1998 , , 506 , 780 bildsten , l. , et al .
1997 , , 113 , 367 bleach , j.n . 2002 , , 332 , 689 blecha , l. , ivanova , n. , kalogera , v. , belczynski , k. , fregeau , j. , & rasio , f. 2005 , , submitted ( astro - ph/0508597 ) blondin , j. m. 1986 , , 308 , 755 boffin , h.m.j . , & jorissen , a. 1988 , , 205 , 155 bondi , h. , & hoyle , f. 1944 , , 104 , 273 brown , g.e .
1995 , , 440 , 270 campana , s. 2005 , in binary radio pulsars , asp conference series vol . 328 ( eds . f. a. rasio & i. stairs ) , 432 campana , s. , & stella , l. 2003 , astro - ph/0309811 chevalier , r. a. 1989 , , 346 , 847 chevalier , r. a. 1993 , , 411 , l33 chen , w. , shrader , c.r . , & livio , m. 1997 , , 491 , 312 claret , a. , & cunha , n.c.s . 1997 , , 318 , 187 cordes , j. , & chernoff , d.f .
1998 , , 505 , 315 dean , a.j . , et al . , 2005 ,
, in press , ( astro - ph/0508291 ) dessart , l. , burrows , a. , ott , c.d . , livne , e. , yoon , s.y . , & langer , n. 2006 , , submitted ( astro - ph/0601603 ) dewi , j.d.m .
, & pols , o.r .
2003 , , 344 , 629 dewi , j. d. m. , & tauris , t. m. 2000 , , 360 , 1043 dhawan , v. , muno , m. , & remillard , r. 2005 , in future directions in high resolution astronomy : the 10th anniversary of the vlba , asp conference proceedings , vol .
340 . edited by j. romney and m. reid .
san francisco : astronomical society of the pacific , p.276 di salvo , t. , et al . 2002 , , 386 , 535 dubus , g. et al . 1999 , , 303 , 139 duquennoy , a. , & mayor , m. 1991 , , 248 , 485 eldridge , j. , & tout , c. 2004a , , 348 , 201 eldridge , j. , & tout , c. 2004a , , 353 , 87 finger , m.h . , wilson , r.b . , & harmon , b.a . 1996 , , 459 , 288 fox , d.b . , et al . 2005 , nature , 437 , 845 fryer , c. , burrows , a. , & benz , w. 1998 , , 496 , 333 garcia , m.r . , mcclintock , j.e . , narayan , r. , callanan , p. , barret , d. , & murray , s.s . 2001 , , 553 , l47 garcia , m.r . , miller , j.m . , mcclintock , j.e . , king , a.r
. , & orosz , j. 2003 , , 591 , 388 garcia - senz , d. , bravo , e. , & woosley , s.w . 1999 , , 349 , 177 gilfanov , m. , 2004 , , 349 , 146 goldreich , p. , & keeley , d.a .
1977 , , 211 , 934 grimm , h .- j . , gilfanov , m. , sunyaev , r. 2003 , , 339 , 793 hachisu , i. , kato , m. , & nomoto , k. 1999 , , 522 , 487 hamann , w. , & koesterke , l. 1998 , , 335 , 1003 hamann , w. , schonberner , d. , & heber , u. 1982 , , 116 , 273 hashimoto , m.a . , nomoto , k.i . , arai , k. , & kaminisi , k. 1986 , , 307 , 687 hayasaki , k. , & okazaki , a.t .
2005 , , submitted ( astro - ph/0503066 ) heggie , d. c. 1975 , , 173 , 729 hjellming , m.s .
1989 , ph.d .
thesis , university of illinois hjellming , m.s . , & webbink , r.f .
1987 , , 318 , 794 hobbs , g. , lorimer , d.r . ,
lyne , a.g .
, & kramer , m. 2005 , , 360 , 974 hurley , j. r. , pols , o. r. , & tout , c. a. 2000 , , 315 , 543 hurley , j. r. , tout , c. a. , & pols , o. r. 2002 , , 329 , 897 hut , p. 1981
, , 99 , 126 hut , p. , & paczynski , b. 1984 , , 284 , 675 ihm , m. , kalogera , v. , & belczynski , k. 2005 , , submitted ( astro - ph/0508626 ) ivanova , n. 2006 , , in press ( astro - ph/0509514 ) ivanova , n. , belczynski , k. , fregeau , j.m . , & rasio , f.a .
2005 , , 358 , 572 ivanova , n. , belczynski , k. , kalogera , v. , rasio , f. , & taam , r. e. 2003 , , 592 , 475 ivanova , n. , & taam , r. e. 2003 , , 599 , 516 ivanova , n. , & taam , r. e. 2004 , , 601 , 1058 janes , k.a .
, & phels , r.l . 1994 , , 108 , 1773 jonker , p. g. , wijnands , r. , & van der klis 2004 , , 349 , 94 kalogera , v. , & webbink , r. 1996 , , 458 , 301 kalogera , v. 1996 , , 471 , 352 kaper , l. 2001 , in proceedings of `` influence of binaries on stellar population studies , ed .
vanbeveren , kluwer acad . pub .
kato , m. , & hachisu , i. 1999 , , 513 , l41 kato , m. , & hachisu , i. 2004 , , 613 , l129 kim , d .- w .
, & fabbiano , g. 2004 , , 611 , 846 king , a. r. , & begelman , m. c. 1999 , , 519 , l169 kobulnicki , h.a . , fryer , c.l . , & kiminki , d.c . 2006 , , submitted ( astro - ph/0605069 ) de kool , m. 1990 , , 358 , 189 kroupa , p. , tout , c.a . , & gilmore , g. 1993 , , 262 , 545 kroupa , p. , & weidner , c. 2003 , , 598 , 1076 kucinskas , a. 1999 , ap&ss , 262 , 127 kuulkers , e. , norton , a. , schwope , a. , & warner , b. 2003 , to appear in compact stellar x - ray sources , cambridge university press , eds .
w.h.g . lewin and m. van der klis la barbera , a. , burderi , l. , di salvo , t. , & robba , n.a .
2001 , , 553 , 375 lai , d. , rasio , f.a . , &
shapiro , s.l .
1993 , , 88 , 205 lamers , h.j . , snow , t.p . , & lindholm , d.m .
1995 , , 455 , 269 lattimer , j.m . , & yahil , a. 1989 , , 340 , 426 levine , a. , rappaport , s.a .
, & zojcheski , g. 2000 , 541 , l194 liu , x .- w . , & li , x .- d . , 2006 , , in press ( astro - ph/0512019 ) liu , q.z . , van paradijs , j. , & van den heuvel , e.p.j .
2000 , , 147 , 25 liu , q.z . , van paradijs , j. , & van den heuvel , e.p.j .
2005 , , 442 , 1135 lyne , a.g . , & lorimer , d.r .
1994 , nature , 369 , 127 maccarone , t.j . , 2003 , 409 , 697 maccarone , t.j . ,
& coppi , p.s .
2003 , , 399 , 1151 mathieu , r. d. , duquennoy , a. , latham , d. w. , mayor , m. , mermilliod , t. , & mazeh , j. c. 1992 , proceedings of ' ' binaries as tracers of stellar formation `` , ed .
duquennoy , a. & mayor , m. , cambridge university press , p.278 mathieu , r. d. , latham , d. w. , & griffin , r. f. 1990 , , 100 , 1859 mcswain , m.v . , &
gies , d.r .
2005 , , accepted ( astro - ph/0505032 ) van der meer , a. , kaper , l. , van kerkwijk , m.h .
, & van den heuvel , e.p.j .
2005 , in the proceedings of the international workshop ' ' massive stars in interacting binaries , submitted ( astro - ph/0502313 ) meibom , s. , & mathieu , r.d .
2005 , , 620 , 970 menou , k. , perna , r. , & hernquist , l. 2002 , , 564 , l81 meurs , e.j.a .
, & van den heuvel , e.p.j .
1989 , , 226 , 88 miller , j.m . , fox , d.m . ,
di matteo , t. , wijnands , r. , belloni , t. , pooley , d. , kouvelioutou , c. , & lewin , w.h.g .
2001 , , 546 , 1055 morrison , i.a . ,
baumgarte , t.w . , & shapiro , s.l .
2004 , , 610 , 941 motch , c. , stella , l. , janot - pacheco , e. , & mouchet , m. 1991 , , 369 , 490 muno , m.p . , et al .
2003 , , 589 , 225 muno , m.p . ,
bauer , f.e . ,
bandyopadhyay , r.m . , & wang , q.d . , 2006 , , submitted ( astro - ph/0601627 ) nice d.j . , et al .
2005 , , 634 , 1242 nieuwenhuijzen , h. , & de jager , c. 1990 , , 231 , 134 nelemans , g. , & tout , c. a. 2005 , , 356 , 753 nutzman , p. , kalogera , v. , finn , l.s . , hendrickson , c. , & belczynski , k. 2004 , , 612 , 364 okazaki , a.t . , & negueruela , i. 2001 , , 377 , 161 orosz , j.a .
2003 , in a massive star odyssey : from main sequence to supernova , proceedings of iau symposium 212 , p.365 oshaughnessy , r. , kalogera , v. , & belczynski , k. 2005a , , 620 , 3850 oshaughnessy , r. , kaplan , j. , kalogera , v. , & belczynski , k. 2005b , , 632 , 1025 oshaughnessy , r. , kim , c. , frakgos , t. , kalogera , v. , & belczynski , k. 2005c , , 633 , 1076 paczynski , b. 1971 , ara&a , 9 , 183 paczynski , b. , & ziolkowski , j. 1967 , acta astronomica , 17 , 7 peters , p. c. 1964 , phys.rev . , 136 , b1224 pfahl , e.d . , rappaport , s. , & podsiadlowski , ph .
2002a , , 571 , l37 pfahl , e.d . ,
rappaport , s. , podsiadlowski , ph .
, & spruit , h. 2002b , , 574 , 364 pinsonneault , m.h . , & stanek , k.z . 2006 , , 639 , l67 podsiadlowski , p. , joss , p.c .
, & hsu , j.j.l .
1992 , , 391 , 246 podsiadlowski , p. , langer , n. , poelarends , a.j.t . ,
rappaport , s. , heger , a. , & pfahl , e.d .
2004 , , 612 , 1044 portegies zwart , s.f . , dewi , j. , & maccarone , t. 2004 , , 355 , 413 prialnik , d. , & kovetz , a. 1995 , , 445 , 789 rappaport , s. , verbunt , f. , & joss , p.c .
275 , 713 rasio , f.a . ,
tout , c.a . ,
lubov , s.h .
, & livio , m. 1996 , , 470 , 1187 ritter , h. 1999 , , 309 , 360 ruiter , a.j . ,
belczynski , k. , & harrison , t. 2006 , , 640 , l167 rutledge , r. e. , bildsten , l. , brown , e. f. , pavlov , g. g. , & zavlin , v. e. 2001 , , 559 , 1054 rutledge , r. e. , bildsten , l. , brown , e. f. , pavlov , g. g. , & zavlin , v. e. 2002 , , 577 , 346 saio , h. , & nomoto , k. 1998 , , 500 , 388 savonije , g.j . , de kool , m. , & van den heuvel , e.p.j .
1986 , , 155 , 51 schaller , g. , schaerer , d. , meynet , g. , & maeder , a. 1992 , a&as , 96 , 269 stauffer , j.b . , & hartmann , l.w .
1986 , pasp , 98 , 1233 saio , h. , & nomoto , k. 1998 , , 500 , 388 stella , l. , white , n.e .
, & rosner , r. 1986 , , 308 , 669 taam , r.e .
1980 , , 242 , 749 taam , r.e . ,
king , a.r .
, & ritter h. 2000 , , 541 , 329 tauris , t.m . , & savonije , g.j . 1999 , , 350 , 928 tauris , t. m. & takens , r. j. 1998 , , 330 , 1047 tavani , m. , & arons , j. 1997 , , 477 , 439 terquem , c. , papaloizou , j.c.b . , nelson , r.p . , & lin , d.n.c . 1998 , , 502 , 788 theuns , t. , boffin , h.m.j . , & jorissen , a. 1996 , , 280 , 1264 thorsett , s.e . , & chakrabarty , d. 1999 , , 512 , 288 timmes , f.x . , woosley , s.e . , & weaver , t.a . 1996 , , 457 , 834 tomsick , j.a
. , et al .
2003 , , 597 , l133 tomsick , j. a. , gelino , d. m. , halpern , j. p. , & kaaret , p. 2004
, , in press tout , c. , aarseth , s. , pols , o. , & eggleton , p. 1997
, , 291 , 732 truss , m. , & done , c. 2006 , , 368 , l25 wang , q.d . , gotthelf , e.v .
, & lang , c.c .
2002 , nature , 415 , 148 wellstein , s. , & langer , n. 1999 , , 350 , 148 wellstein , s. , langer , n. , & braun , h. 2001 , , 369 , 939 webbink , r. f. 1984 , , 277 , 355 willems , b. , & kolb , u. 2003 , , 343 , 949 woo , j.w . , et al .
1996 , , 467 , 811 woosley , s. e. , taam , r. e. , & weaver , t. a. 1986 , , 301 , 601 woosley , s. e. , & weaver , t. a. 1994 , , 423 , 371 van bever , j. , & vanbeveren , d. 2000 , , 358 , 462 vanbeveren , d. , van rensbergen , w. , & de loore , c. 1998 , in ` the brightest binaries' , astrophysics and space science library , vol .
232 zahn , j.p .
1975 , , 41 , 329 zahn , j.p .
1977 , , 57 , 383 ; erratum 67 , 162 zahn , j.p .
1989 , , 220 , 112 zangrilli , l. , tout , c.a . , & bianchini , a. 1997 , , 289 , 59 zorec , j. , & briot , d. 1997 , , 318 , 443 ) is much greater than the mass ratio at the moment when the orbit starts expanding ( @xmath273 ) then the system is dynamically unstable , otherwise rlof on a thermal timescale is assumed .
the arrow represents the partial derivative of donor radius ( equal to the roche lobe radius ) with respect to its mass , and points to the place where the donor is expected to regain thermal equilibrium .
the bottom panel shows a specific system : a @xmath277 hertzsprung gap donor with a @xmath278 ms companion in an 8-day orbit , for which the diagnostic diagram is plotted .
the mass transfer begins on a thermal timescale ( flat part ) and then evolves on a slower nuclear timescale ( decline ) . for more details
see 5.2 . ]
( for details see 6 ) .
the line oa is parallel to the vector @xmath618 , while the line ob , the asymptote of the hyperbola , is parallel to the vector @xmath619 .
the point o is the focus of the hyperbola . ]
ns + @xmath508 evolved he - star binary .
the critical eddington mass accretion rate onto the ns is @xmath625 .
panels same as in fig .
[ case1 ] .
note the very short duration of this rlof phase ; the ( finite ) timesteps taken by the code may be seen through lines showing orbital period and donor mass . ] , respectively ) as opposed to the standard prescription , presented on the top panel ( @xmath626 ) .
note the increase of cutoff period ( the longest period circular binary in a given sample ) with increasing @xmath532 .
the observed cutoff period for m67 is @xmath627 days . for more details
see 8.2.1 . ]
( marked with dotted line ) . predicted decay rates for different radii of the main sequence secondary in respect to its roche lobe ( @xmath628 ) are shown for @xmath629 .
each track is displayed after the massive star reaches the relative radius corresponding to a given track and for all tracks the starting point is then set to time=0 . for more details
see 8.2.2 . ]
|
we present a comprehensive description of the population synthesis code startrack .
the original code has been significantly modified and updated .
special emphasis is placed here on processes leading to the formation and further evolution of compact objects ( white dwarfs , neutron stars , and black holes ) . both single and binary
star populations are considered .
the code now incorporates detailed calculations of all mass - transfer phases , a full implementation of orbital evolution due to tides , as well as the most recent estimates of magnetic braking .
this updated version of startrack can be used for a wide variety of problems , with relevance to many current and planned observatories , e.g. , studies of x - ray binaries ( chandra , xmm - newton ) , gravitational radiation sources ( ligo , lisa ) , and gamma - ray burst progenitors ( hete - ii , swift ) .
the code has already been used in studies of galactic and extra - galactic x - ray binary populations , black holes in young star clusters , type ia supernova progenitors , and double compact object populations . here
we describe in detail the input physics , we present the code calibration and tests , and we outline our current studies in the context of x - ray binary populations .
|
a few years ago , william harris asked how the kinetic energy acquired from running is converted in the long and high jump events.@xcite the question was whether athletes running horizontally can change their velocity into one that forms an angle of 45@xmath0 with the horizontal without changing their speed .
the three negative answers@xcite were that athletes can not convert their initial horizontal velocity into a vertical one,@xcite because it is impossible to generate the necessary power required by the task,@xcite or equivalently , athletes can not sustain the necessary acceleration to acquire the vertical velocity at takeoff.@xcite if the answer were positive , the center of mass of a world class athlete running at 10 m/s and taking off at 45@xmath0 would go a horizontal distance of about 10.2 m . because the athlete s center of mass is forward of the front edge of the runway at takeoff and behind the point where her heels hit the ground at landing ( see sec .
iii ) , the actual total jump length would be about 11 m .
for the high jump , the athlete s center of mass would attain a maximum height of 3.5 m .
these results are much greater than actual records .
we conclude that athletes can not totally change their horizontal velocity into a vertical one , as stated by the negative answers .
the negative answers bring some new questions . if the magnitude of the velocity is assumed to be unchanged and thus the kinetic energy , why is extra power necessary ?
why does a high jumper run horizontally to jump vertically ?
how do runners change their horizontal velocity into a velocity with a vertical component ?
what is conserved and what is changed at takeoff ?
there are three systems that any animal uses to produce energy.@xcite in the case of humans , for activities whose duration is longer than a few minutes , aerobic energy is produced at a low rate by burning carbohydrates or fats . for activities that last a few tens of seconds
, the main source of energy is the breakdown of glycogen in the absence of oxygen and the production of lactic acid .
for very short duration activities the body uses the adenosine triphosphate ( atp ) stored in muscles , and energy is supplied immediately , following the conversion of atp into adenosine diphosphate ( adp ) .
the static force generated from muscle contraction can be very high .
however the magnitude of the force is limited by the rate that the muscles can produce energy , which is limited by the rate that atp is depleted ( atp depletion is completed in a few seconds ) .
for example , after a few seconds sprinters no longer accelerate , but decelerate,@xcite because they can not overcome the force due to air resistance .
atp is the source of energy in all explosive sports , such as jumps , sprints , and weightlifting .
thus , if we produce a force with a non - vanishing velocity , that is , we produce mechanical power , the force is limited by the rate that our muscles can convert atp into adp . to discuss the performance of a jumper
, we use the results of two observations .
one is that the maximum force that athletes can produce in a small time interval with their legs while producing work .
we will take this information from observations of the squat . in this exercise a weight is rested on the shoulders of an athlete .
he starts from the upright position , and bends his ankles , knees , and hips as if sitting .
at the lowest position , when his thighs are horizontal , he returns to the upright position . the maximum weight an athlete can lift doing a squat is about 5000n . for a first class male athlete with mass 100 kg , the maximum force that can be made with just one leg is about 3000n .
( in the high and long jump , athlete s use just one leg to push their center of mass . )
this force is different from a static force . in both the high and long jump , the height of the runner s center of mass remains almost constant until the last stride . in the last stride
the center of mass rises about 25 cm before the runner takes off and his foot loses contact with the ground .
from these two observations we conclude that an 80 kg runner ( assuming 80 kg athlete can produce the same 3000n force with each leg ) can add about @xmath1j to the kinetic energy .
we will use this result in the following .
an equivalent result can be obtained from an analysis of an elite 100 m runner . at 6 m/s ,
an elite runner has an acceleration of about 5m / s@xmath2 ( see , for instance , fig . 1 in ref . 6 ) .
for an 80 kg athlete this acceleration corresponds to a mechanical power of 2400w . at every stride , the runner s center of mass rises about 5 cm . at a rate of 5strides / s ( a typical rate in the 100 m dash )
, the up - down movement of the center of mass corresponds to 200w of additional power .
thus at each stride , the runner produces @xmath3j .
if we also include a small contribution due to the air resistance , this result is consistent with the previous estimate of 550j . for female runners , the time of the 100 m dash is about 10% greater , and the velocity about 10% smaller than the male records .
thus , we estimate that the acceleration and , hence the force , is about 20% smaller than that for male runners .
thus , for events that depend on velocity and force as do the long and high jumps we expect that female performance to be between 10% and 20% smaller .
the model for the long jump is simple . after running a few seconds the runner produces a big vertical force against the ground in the last stride .
some details of the movement of the athlete are shown in fig .
[ longjump ] . at takeoff ,
the runner s center of mass is about 0.4 m forward of the front edge of the runway , @xmath4 , in fig .
[ longjump ] . at landing ,
the runner s center of mass is behind the point where the heels hit the sand ( @xmath5 m ) .
the difference between the vertical height of the runner s center of mass at takeoff and at landing , which would lead to a small difference between @xmath4 and @xmath6,@xcite will be neglected .
the length of the jump is @xmath7 , where @xmath8 is the flight distance .
the horizontal velocity of the runner at the last stride , @xmath9 , is about 10 m/s .
( usually , long jumpers are good sprinters . )
the best a runner can do is to use the additional energy of 550j to obtain a vertical velocity .
for an 80 kg athlete , the vertical velocity gained is @xmath10 m/s .
thus , @xmath8 is given by @xmath11 and @xmath12 .
the latter result agrees with observations between 7.4 m and 8.8 m for elite male long jumpers and 6.5 m and 7.5 m for elite female jumpers.@xcite the takeoff angle can be calculated from the velocity diagram in fig .
[ diag1 ] , and we find @xmath13
. this result agrees with observations of @xmath14.@xcite
before jumping , a high jumper runs a few meters and attains a speed of about 7.5 m/s .
why does the jumper run horizontally before jumping vertically ?
what is the mechanism used by the jumper to change her horizontal velocity into a velocity with a vertical component ? ( a detailed description of the high jump can be found in ref . 8 . ) to answer these questions , consider a bar representing a jumper of length @xmath15 , which has a horizontal velocity @xmath9 and is inclined backward , as shown in fig .
[ highjump ] . at point
p , the bottom end of the bar stops suddenly , and the bar begins to rotate around p. due to the impact , the bar s velocity will change , but the angular momentum with respect to p is conserved .
conservation of angular momentum at the instant the bottom end stops gives @xmath16 where @xmath17 is the bar s center of mass velocity immediately after the impact and the velocity is perpendicular to the bar .
thus , the initial horizontal velocity has changed to a velocity with a vertical component .
the horizontal , @xmath18 , and vertical , @xmath19 , projections of the velocity just after the bottom end of the bar stops are @xmath20 if the bar stops the rotation immediately after hitting p and if the horizontal and vertical velocities at takeoff are given by eqs .
( [ vh ] ) and ( [ vv ] ) , the center of mass of the bar will attain a maximum height ( calculated by energy conservation ) given by @xmath21 if we use @xmath9 = 7.5 m/s and a 2 m
( @xmath22 m ) bar , we obtain the maximum @xmath23 for @xmath24 .
suppose that the runner is represented by a bar moving at 7.5 m/s inclined backward at 56@xmath0 ( in sec .
[ sec : discuss ] we will show how this motion is possible ) then the runner begins to rotate just for a moment around point p. as in the long jump , she can add 550j to the total kinetic energy .
the best the higher jumper can do is to gain an additional velocity in the vertical direction , which can be calculated from @xmath25 giving @xmath26 = 4.5 m/s .
the increase in her vertical velocity is @xmath27 m/s .
if the runner takes off with a vertical velocity equal to 4.5 m/s , the runner s center of mass will rise about 1.0 m . if we add this value to the initial position of the runner s center of mass at takeoff , which is about 1.1 m , we obtain a total height of 2.1 m .
because we know the horizontal and vertical velocities , we can calculate the takeoff angle as @xmath28 ( see fig . [ diag2 ] ) .
the world record is 2.45 m for men and 2.09 m for women and a typical takeoff angle is 50@xmath0 .
these values suggest that the model is reasonable . despite the simplification of representing a person by a bar to explain the high jump ,
the pole vault is even simpler to analyze .
the pole vaulter runs about 15 m and reaches a velocity of about 10 m/s .
her kinetic energy is changed to elastic potential energy of the bar , and then the energy is changed to the vaulter s gravitational potential energy . if we assume that the vaulter s center of mass at the beginning of the jump is about 1.1 m above the ground , we can use energy conservation to calculate that the height reached by a pole vaulter is 6.2 m .
the records are about 5 and 6 m for men and women , respectively .
this estimate is only approximate because it does not take into account that the vaulter s center of mass passes significantly under the bar , and it assumes a 100% efficiency of horizontal to vertical jump energy conversion .
can a high jumper run with his body inclined backward , as assumed in the model ?
the answer is that it is not possible . like everyone else
, the jumper runs in a vertical position .
however , in the last strides , the high jumper makes a circular curve with a radius of about 10 m . due to the centripetal force
, the jumper leans to the center of the circle and the trajectory of the center of mass differs from the trajectory of the footprints .
these trajectories are illustrated in fig .
[ route ] . at the beginning of the takeoff ,
corresponding to the instant that the bottom end of the bar stops , the jumper s center of mass is tilted toward the center of the curve , and his velocity is along the arrow shown in fig . [
route].@xcite during the next 0.2s , he changes his body orientation , tilting it outside the circle , and produces a very strong vertical force against the ground .
( this model is an oversimplification of the detailed model of ref . , but it retains the essence of the athlete s movement at takeoff as illustrated in fig . 1 of ref .
8) in our model we assumed that the long jumper does not change her initial velocity into a new velocity using the same trick used by the high jumper because fixing one foot against the ground would cause some loss of the horizontal velocity , which is very important to the long jumper .
we note that the high jump differs from the vertical standing jump , which has been studied in ref . .
in the standing jump the athlete uses both legs to jump . in the vertical standing jump the athlete does not run before jumping and the only source of energy comes from the leg muscles .
as can be estimated from the figures of ref .
10 , the mechanical energy gained at takeoff by a non - athlete is about 550j when using both legs ; our hypothesis that an elite athlete can add 550j when using only one leg seems reasonable .
|
a simple model based on the maximum energy that an athlete can produce in a small time interval is used to describe the high and long jump .
conservation of angular momentum is used to explain why an athlete should run horizontally to perform a vertical jump .
our results agree with world records .
|
host galaxy studies got the opportunity for a real boost in february 1997 when the near - infrared camera and multi - object spectrometer ( nicmos ) was installed on the hubble space telescope ( hst ) .
nicmos combines the superb spatial resolution of hst with the benefits that long wavelengths provide for imaging the redder hosts against the overwhelming glare of the bluer quasar nuclei .
we have used nicmos to image 16 radio quiet quasars as part of a project to investigate the `` luminosity / host - mass limit , '' the results of which were presented in mcleod , rieke , & storrie - lombardi ( 1999 ; hereafter mrs ) . in this paper , we present the images themselves , along with 1- and 2-dimensional analyses of the host galaxy properties .
the sample , listed in table [ tab-1d ] , is composed of all 10 quasars from our `` high - luminosity sample '' ( the 26 highest - luminosity pg quasars with @xmath1 ; @xcite ) that had not been previously observed with hst . to this we added 6 luminous quasars out to @xmath2 for which ground - based attempts to resolve a host galaxy had failed .
all 16 objects are in the redshift range @xmath3 with an average @xmath4 .
llllllllll pg0026 + 129 & 0.142 & 13.3 & -25.5 & 15.2 & -23.6 & -23.7 & 0.17 & 0.16 + pg0947 + 396 & 0.206 & 14.3 & -25.3 & 15.4 & -24.2 & -23.1 & 0.37 & 0.05 + pg1048 + 342 & 0.167 & 15.5 & -23.7 & 15.1 & -24.1 & -23.2 & 1.48 & 0.07 + pg1121 + 422 & 0.224 & 14.6 & -25.3 & 17.0 & -22.8 & -23.7 & 0.11 & 0.31 + pg1151 + 117 & 0.176 & 14.6 & -24.7 & 15.9 & -23.3 & -23.6 & 0.29 & 0.19 + pg1322 + 659 & 0.168 & 14.3 & -24.9 & 15.6 & -23.5 & -23.2 & 0.29 & 0.10 + pg1352 + 183 & 0.158 & 15.0 & -24.0 & 15.1 & -23.9 & -23.2 & 0.92 & 0.07 + pg1354 + 213 & 0.300 & 15.7 & -24.9 & 16.1 & -24.4 & -24.6 & 0.73 & 0.16 + pg1427 + 480 & 0.221 & 15.0 & -24.9 & 16.1 & -23.7 & -23.3 & 0.35 & 0.11 + pg2233 + 134 & 0.325 & 15.2 & -25.6 & 16.7 & -24.0 & -24.3 & 0.26 & 0.20 + mark876 & 0.129 & 13.6 & -25.0 & 13.3 & -25.3 & -23.0 & 1.36 & 0.02 + um357 & 0.334 & 15.1 & -25.8 & 16.7 & -24.0 & -24.2 & 0.22 & 0.17 + q0530 - 379 & 0.334 & 15.5 & -25.4 & 16.9 & -23.9 & -23.4 & 0.28 & 0.09 + nab1612 + 26 & 0.395 & 15.6 & -25.6 & 17.3 & -23.9 & -23.4 & 0.22 & 0.10 + 1628.6 + 3806 & 0.394 & 15.2 & -26.0 & 16.7 & -24.4 & -23.7 & 0.26 & 0.08 + kuv18217 + 6419 & 0.297 & 12.3 & -28.2 & 14.7 & -25.7 & -26.0 & 0.11 & 0.20
+ we have used a zero point of 21.80 magnitudes in the vega system for the f160w filter ( @xcite ) , and refer to the resulting magnitudes simply as `` h. '' we note that this is fainter by 0.31 mag than the value we used in mrs . for computing rest - frame properties of the hosts , we assume @xmath5 throughout , but our results are not strongly sensitive to cosmology .
for example , at @xmath2 , the difference in proper distance between @xmath6 and 1 is 10% .
we also apply galaxy h - band k - corrections appropriate for star - forming galaxies , but note these amount to less than 0.1 magnitude for the redshifts in our sample . for the nuclei ,
we have used colors and k - corrections from cristiani & vio ( 1990 ) . to compute @xmath7
, we have assumed that in the b - band all of the light belongs to the nucleus .
this is a reasonable assumption for these high - luminosity quasars because ( i ) we know _ ex post facto _ that the galaxies are generally less luminous than the nuclei in the h - band , and ( ii ) the galaxy contribution relative to the nucleus falls dramatically at shorter wavelengths ( see e.g. mcleod & rieke 1995 , fig .
we observed each quasar in a single orbit with the nic2 camera and the f160w filter ( approximately the h band ) .
a small amount of time at the end of each orbit was used to observe a star for characterizing the point - spread - function ( psf ) of the telescope .
these `` psf stars '' were chosen within 2 of the quasar , which is the maximum allowed slew that does not require overhead for acquisition of new guide stars . for 11 quasars
, we used our own ground - based h - band images to locate the infrared - brightest star within the 2 limit . for the remaining five quasars , we chose the visually brightest star from the guide star catalog .
these psf stars do not have the same colors as the quasars , but this has not caused problems ; we show below that the results are insensitive to psf used .
all of the quasars and psf stars were observed in multiaccum mode , which records data in a series of increasingly long nondestructive readouts .
the reduction software determines the average count rate for each pixel individually by fitting those parts of the integration ramps that are not saturated .
this allows us to build up an image that is linear over the entire field , including on the bright quasar core .
we can therefore avoid some of the difficulties previously encountered with psf - subtraction in wfpc2 exposures , where the quasar core is necessarily saturated out to @xmath8 .
the total on - source times for the quasars ranged from 1792 to 2304 sec , while the totals for the psf stars were 32 - 128 sec . for most of the quasars ,
the multiaccum sequences were obtained in a four - position spiral - dith pattern with a dither size of 10875 , corresponding to 14.5 nic2 pixels .
this offset was chosen to be large enough so that the quasar nucleus was on a different set of pixels on each of the four exposures , but small enough to provide a reliable half - pixel offset to improve the sampling of the otherwise slightly undersampled image .
the dithering also allowed recovery of data lost due to cosmic rays .
the psf stars were observed in the same four - position spiral - dith , and all of the quasar and psf dithers were centered on the same part of the chip ( within @xmath9 ) . for four objects with extended hosts and/or large companions ( pg0026 + 129 , pg0947 + 396 , pg1048 + 342 , mrk876 ) we extended the area covered by observing in a 7 or 8 position spiral - dith - chop pattern , four positions of which coincided with the positions used for the rest of the objects .
we reduced all of the quasar and psf images using _ nicred_1.8 _ , a set of c programs and iraf scripts developed for use with nicmos images ( @xcite ; @xcite ) .
this version improves upon earlier versions of the pipeline and _ nicred _ by lowering the threshhold at which saturation is flagged .
this is especially important for bright sources like our quasar cores , where significant charge can build up before the first readout
. the f160w flat and dark images are on - orbit exposures processed and provided by nicmos deputy pi m. rieke . for the rest of the calibration we used
the _ nicred _ files except that we modified the bad pixel mask to account for ( i ) the coronagraphic hole in the upper left portion of the images , and ( ii ) pixels that were obviously bad in many of our frames .
in addition , cosmic ray hits were flagged by hand for each image as part of the _ nicred _ reduction process .
the resulting images from each dither position were magnified by a factor of 2 and then aligned and combined to produce the final image for each object .
the central 96 of the 20 reduced quasar images are shown in the first column of fig .
[ fig - qimages ] , and those of the psfs are shown in fig .
[ fig - mosaicp ] .
the final pixel scale is @xmath10 .
the psf `` star '' selected for pg1121 + 422 turned out to be a compact galaxy and is not shown . for the quasar images ,
the @xmath11 scatter per pixel in blank sky corresponds to a limiting surface brightness of @xmath12 ( or @xmath13 for a typical galaxy moved to the average redshift of our sample ) .
[ fig - qimages ] .
contd fig .
[ fig - qimages ] .
to estimate the magnitudes of the quasar host galaxies , we first performed a model - independent 1d removal of the nuclear point source for each object .
this analysis differs somewhat from the technique used in mrs , and is better suited to the rereduced the data with the improved _ nicred_1.8 _ saturation limits .
the results here supersede those in the earlier paper .
we assume that the h - band light from the quasar is composed of only two components : the host galaxy , which in the h - band is largely due to light from red giant stars ; and the nucleus , taken to be a point source with image shape given by the psf of the telescope .
we do not assume a particular model for the galaxy , except to assume that its flux decreases monotonically away from its center .
we start by extracting from circular annuli 1d radial intensity profiles of the 16 quasars and the 15 psf stars .
the profile for the brightest psf star , corresponding to pg0026 + 129 and called hereafter psf0026 , is shown in fig .
[ fig - pg0026_p ] .
we then scale the psf profiles to have the same central flux as the quasar profile , and subtract from the quasar the highest psf fraction that leaves a monotonic profile inside the first airy minimum . without assuming a particular functional form for the host galaxy radial profile , this `` just monotonic '' subtraction provides a reasonable estimate of the host flux ( @xcite ) .
finally , we numerically integrate the difference profiles , excluding any light from companions .
we can typically integrate the profiles to a surface brightness of @xmath14 ( corresponding to @xmath15 ) .
the resulting host galaxy magnitudes are given in table [ tab-1d ] , and the profiles are shown in fig .
[ fig - profs ] .
[ fig - profs ] .
contd in four cases where the quasar s own psf was faint compared to the quasar , and in the case of pg1121 + 422 which has no psf star , we have used psf0026 for the analysis instead . to test the reliability of using another quasar s psf , we repeated our analysis using psf0026 on all of the objects .
the resulting host magnitudes were acceptably close , with @xmath16 .
this is an indication that the azimuthal average of the nic2 psf is relatively stable . in figure [ fig - profs_p ] , we plot the central parts all 15 psf star profiles together , to show graphically the extent to which this is so . the assumption inherent in our 1d method is that the galaxy contributes little flux to the profile inside the first airy minimum .
the procedure we use is effectively equivalent to the one used with our ground - based data , where we subtracted just enough psf to make the difference profile flatten out at the center .
we can now use nicmos data to test the effect of seeing on this procedure for the ground - based data , by comparing the magnitudes for the 8 quasars where we have ground - based detections ( @xcite ) .
we find the average value of @xmath17 , which is in good agreement with the 0.30 mag uncertainty quoted for our ground - based data ( @xcite ) .
one of the most discrepant points is due the presence of a companion that could not be resolved from the ground .
we conclude that , despite an order - of - magnitude difference in resolution , the ground - based host magnitudes derived by this method for low - redshift quasars are fairly robust when nearby companions can be resolved .
our 1d analysis is simplistic but has the advantage of not assuming _ a priori _ a model for the host galaxy .
the amount by which our technique underestimates the host flux is dependent on the exact profile of the galaxy . to understand the limitations of this method
, we generated a grid of model quasar images covering a range of type ( devaucouleurs and exponential ) , effective radius and scale length ( 0.5 - 1 , corresponding to 1.5 - 3 kpc for the average redshift of our sample ) , and total host intensity relative to the nucleus ( 1 - 30% , corresponding to galaxies in the range 13 - 16.7 mag ) . we modeled the quasar as point source + galaxy , and convolved the image with one our observed psf stars .
we then ran our 1d analysis technique on the images using psf0026 as the `` psf star . ''
we find that for all the model galaxies with @xmath18 of the nuclear luminosity , this technique underestimates the true host flux by @xmath19 mag .
for the galaxies at 1% , errors in estimating the background level for the profile dominate the procedure , and the resulting host magnitude errors can be much higher , up to 0.7 mag off in either direction .
for the real quasars in our sample , the 1d analysis yielded hosts with total luminosity from 10 - 150% of the nuclear luminosity . only one host , that of pg1121 + 422 , is at both a low fraction ( 11% ) and at the faint end where background errors are worrisome .
therefore , we treat its detection and magnitude as uncertain , but estimate that the rest of the host magnitudes have an accuracy of @xmath20 mag . in fig .
[ fig - profs ] , each quasar profile is shown plotted versus both linear radius @xmath21 and @xmath22 .
these plots can be used to judge how well the profiles approximate an exponential disk law ( straight line on the plots versus @xmath21 ) or devaucouleurs law ( straight line on the plots versus @xmath22 ) .
one can see from this figure that the shape of the galaxy s profile outside the central 04 is very little affected by even large errors in psf subtraction ( compare the `` just monotonic '' and `` 100@xmath23 '' curves ) .
eyeball inspection of these plots shows that some of the profiles clearly follow a devaucouleurs law , for example nab1612 + 26 and kuv18217 + 6419 . in some cases ,
the choice is dependent upon the psf fraction subtracted .
we now turn to a more complete analysis of host morphology using the full 2d information in the images .
to characterize the properties of the host galaxies using the full 2d information in the nicmos images , we have fit the quasars with combinations of point sources , devaucouleurs galaxies , and exponential disk galaxies , all convolved with a psf image .
the goodness - of - fit is determined by a merit that is the weighted sum of the squares of the residuals over all the pixels .
we modeled each quasar using different combinations of the following components : ( i ) the background level ( parametrized by a single number ) , ( ii ) the nucleus ( intensity @xmath24 , position @xmath25 ) , and ( iii ) the host galaxy ( @xmath26 , @xmath27 , position angle @xmath28 , axis ratio @xmath29 , and scale length @xmath30 for a devaucouleurs law or @xmath31 for an exponential disk ) .
when other galaxies were present in the field , we fit those simultaneously to make sure their contributions would not bias the host galaxy parameters .
we performed many tests of the fitting procedure s robustness , finding that we could not always vary all the parameters simultaneously ; in the case of tiny psf mismatches , the fitting program would sometimes turn the `` galaxy '' component into a long , off - center , thin spike to match a high spatial frequency , high signal - to - noise psf feature .
we therefore constrained the galaxy and quasar to have the same position , i.e. we assumed the nucleus was in the center of the galaxy . in tests where we relaxed this assumption , and where the galaxy was not turned into a spike , the fitted positions of nucleus and host galaxy agreed within 001 ( @xmath32 ) , corresponding to 30pc at the distance of the average quasar .
we list in tables [ tab-2d ] and [ tab-2e ] the galaxy parameters for the devaucouleurs and exponential models from unweighted fits using each quasar s own psf ( @xmath33 weighting is discussed in [ sec - chisqr ] ) .
the host galaxy magnitudes derived from the devaucouleurs fits are in good agreement with the 1d method , with an average difference of @xmath34 . as is commonly seen in host profile fitting , the exponential fits give hosts systematically fainter than the 1d fits and devaucouleurs fits by 0.7 mag , with a @xmath11 spread of 0.30 mag .
rrrrrrrrrrrrrrrrrrrrrrr pg0026 + 129&14.7&2.3&0.87&13.2&-25.6&14.7 & -24.1&2.3&4.9&0.87&0026 + pg0947 + 396& ... & ... & ... &14.3&-25.4&14.9 & -24.7&2.6&7.4&0.83&1821 + pg1048 + 342&15.5&0.4&0.67&15.2&-24.0&15.4 & -23.8&0.8&1.8&0.61&0026 + pg1121 + 422& ... & ... & ... &14.5&-25.4&17.2 & -22.6&3.9&11.8&0.92&0026 + pg1151 + 117&15.9&0.4&0.66&14.5&-24.8&15.7 & -23.6&0.6&1.6&0.82&0026 + pg1322 + 659& ... & ... & ... &14.1&-25.0&15.8 & -23.4&0.7&1.8&0.95&0026 + pg1352 + 183&15.3&1.2&0.83&14.8&-24.3&15.2 & -23.8&1.4&3.3&0.79&1821 + pg1354 + 213&16.1&1.0&0.71&15.6&-25.0&16.2 & -24.3&0.7&2.7&0.75&0026 + pg1427 + 480&16.0&0.5&0.89&14.9&-24.9&16.0 & -23.8&0.5&1.5&0.89&0026 + pg2233 + 134&16.5&0.6&0.95&15.2&-25.6&16.7 & -23.9&0.4&1.7&0.63&1821 + mark876&13.4&1.8&0.83&13.3&-25.3&13.0 & -25.5&4.1&8.0&0.80&0026 + um357&17.1&1.3&0.54&14.9&-25.9&17.2 & -23.5&1.3&5.0&0.43&0026
+ q0530 - 379&17.3&0.2&0.71&15.3&-25.5&17.1 & -23.6&0.7&2.7&0.71&0026 + nab1612 + 26&17.2&0.3&0.79&15.6&-25.6&17.2 & -23.9&0.3&1.4&0.79&1612 + 1628.6 + 3806&16.7&0.2&0.67&15.1&-26.1&16.8 & -24.4&0.6&2.6&0.73&0026 + kuv18217 + 6419& ... & ... & ...
&12.3&-28.2&14.7 & -25.7&0.3&1.1&0.87&1821 + rrrrrrrrrrrrrrrrrrrrrrr pg0026 + 129&15.8&0.4&0.81&13.2&-25.6&15.8 & -23.0&0.4&0.8&0.81&0026 + pg0947 + 396& ... & ... & ... &14.3&-25.4&15.9 & -23.7&0.5&1.5&0.77&1821 + pg1048 + 342&16.1&0.2&0.69&15.1&-24.0&16.0 & -23.1&0.2&0.6&0.64&0026 + pg1121 + 422& ... & ... & ... &14.5&-25.4&18.0 & -21.8 & 0.7 & 2.0 & 0.79 & 0026 + pg1151 + 117&16.3&0.2&0.81&14.5&-24.8&16.2 & -23.0&0.2&0.6&0.90&0026 + pg1322 + 659& ... & ... & ... &14.1&-25.1&16.4 & -22.8&0.2&0.6&0.92&0026 + pg1352 + 183&16.0&0.3&0.87&14.7&-24.3&15.9 & -23.1&0.4&0.9&0.89&0026 + pg1354 + 213&16.8&0.3&0.80&15.5&-25.0&16.8 & -23.7&0.2&0.9&0.81&0026 + pg1427 + 480&16.5&0.2&0.91&14.9&-24.9&16.5 & -23.3&0.2&0.5&0.91&0026 + pg2233 + 134&17.1&0.2&0.95&15.1&-25.6&17.0 &
-23.6&0.3&1.0&0.94&0026 + mark876&14.5&0.4&0.88&13.3&-25.3&14.3 & -24.2&0.5&1.0&0.87&0026 + um357&17.8&0.4&0.72&14.9&-25.9&17.8 & -22.9&0.4&1.4&0.56&0026 + q0530 - 379&17.7&0.1&0.74&15.3&-25.5&17.7 & -23.0&0.2&0.8&0.77&0026 + nab1612 + 26&17.7&0.2&0.74&15.5&-25.7&17.7 & -23.4&0.3&1.1&0.60&0026 + 1628.6 + 3806&17.2&0.1&0.73&15.1&-26.1&17.2 & -23.9&0.2&1.1&0.76&0026 + kuv18217 + 6419& ... & ... & ... &12.3&-28.3&15.3 & -25.1&0.3&1.3&0.46&0026 + in the second column of fig . [ fig - qimages ] , we show the quasars after removal of the nucleus based on 2d devaucouleurs fits using the high signal - to - noise psf0026 ( at this stretch , the ones based on 2d exponential fits are visually indistinguishable ) .
we also show in fig .
[ fig - qimages ] the residuals after subtracting both the nucleus and the model galaxy for both types of host .
the fitting technique has done a superb job of removing most of the complex features of the nicmos psf .
it is immediately obvious that few of these hosts are perfectly fit by ideal devaucouleurs and exponential models .
however , subtracting out the smooth models does reveal interesting structure , which we discuss in [ sec-2dresults ] .
in general , structure can be believed outside of the central 05 diameter noisy region , and , in a few galaxies , the regions around the diffraction spike residuals . to test the sensitivity of the fits to the properties of the psf star and to possible sampling effects in the psf cores
, we repeated the devaucouleurs and exponential fits for each quasar with the five psf stars having the highest signal - to - noise , running each fit both with and without a mask that excluded the quasar nucleus inside a 037 diameter .
thus , we have fit each quasar @xmath35 different ways .
we list in tables [ tab-2d ] and [ tab-2e ] the results of the `` best '' fits , i.e. those using the psf that gave the lowest merit for each quasar and galaxy type .
this was often the high signal - to - noise psf0026 .
we find that there are minor differences in the residuals around the first airy ring , which can probably be attributed to telescope breathing .
the effect on the photometry can be quantified by comparing the results of the `` own '' and `` best '' fits .
we find that the host magnitude is robust for both devaucouleurs and exponential fits , with @xmath36 and @xmath37 respectively .
the radii for exponential fits generally agree within 005 , corresponding to 0.15 kpc for a typical quasar in our sample .
however , with a scatter of 07 , the devaucouleurs radii are found to be much less certain .
this effect , seen also by mclure , dunlop , & kukula ( 1999 ; hereafter mdk ) , arises from a degeneracy whereby the fit can steal light from the nucleus to put into the peaky devaucouleurs galaxy . for both types of galaxies ,
the axial ratios agree within @xmath38 .
we find a nearly identical result when we compare the results from the normal fits to those where we mask out the central few pixels to avoid having the merit function dominated by noise in the bright centers . as a further test of the 2d method
, we explored the level to which our method might fabricate galaxies when none are detectable . to do this , we forced nucleus + galaxy fits to two true point sources , namely the bright psfs corresponding to kuv18217 + 6419 and pg1151 + 117 ( a psf that is not as well matched to the others ) .
we fitted them using the 5 psfs used with the quasars and assuming both devaucouleurs and exponential models for the `` galaxies . ''
of these 20 combinations , the fits diverged to give negative or linear `` galaxies '' 18 times .
the negative fluxes corresponded to a level only 0.4% that of the point source . in two cases with pg1151 + 117 s psf ,
the method fabricated an exponential disk with reasonable physical parameters .
these `` galaxies '' have fluxes of only 12% and 2% that of the point source . given the flux levels for our quasars , this test suggests that all the host detections are secure with the possible exception of pg1121 + 422 and kuv18217 + 6419 , for which any exponential solutions must be treated as suspect . finally , the nuclear magnitudes from all of the fits appear extremely robust .
we measure @xmath39 , @xmath40 , and @xmath41 the three quasars with large differences ( @xmath42 ) in nuclear magnitudes between the 1- and 2d fits are the ones in which the galaxy light constitutes @xmath43 of the total in the h - band .
we conclude based on these tests that , for pure exponentials and devaucouleurs hosts , the nuclear magnitudes and the host magnitudes , position angles , and exponential scale lengths are robust within the uncertainties quoted above , but that the devaucouleurs radii are less reliable . based on inspection of the residual images and the tests described above ,
we believe our unweighted fits give good estimates of the host parameters .
however , the merit function used does not provide a well - defined statistical discriminant between devaucouleurs and exponential models for the hosts .
following the approach of mdk , we have attempted to assess the preference for one or the other type using a @xmath33 analysis .
we first generated an azimuthally symmetric error map from each quasar image by masking out companions and assigning the error at each radius to be the standard deviation @xmath44 in circular annuli centered on the quasar .
we then re - ran the fits with the @xmath45 weight map and the high signal - to - noise psf0026 ( we found that the @xmath45 weighting scheme fails with fainter psfs where the fits become dominated by the noisy wings ) .
as in mdk , maps of @xmath46 generated by the psf0026 fits are very uniform , so that no one area dominates the fitting . however , the fits generally resulted in overall @xmath33 values less than the number of degrees of freedom , indicating that we have overestimated the size of the errors .
this is because azimuthal structure in the psf inflates the rms error around each annulus .
we have compared the resulting galaxy parameters to those generated by the best unweighted fits . excluding pg1121 + 422 , which is again suspect , and kuv18217 + 6419 , for which no unweighted devaucouleurs fit converged even though it is clearly an @xmath22 host based on the 1d profile , the host magnitudes and axial ratios have exactly the same spread as the uncertainties quoted above .
the galaxy sizes , however , are significantly different .
the weighted fits give systematically larger ellipticals and spirals by 06 ( or 1.5 times ) and 03 ( or 1.9 times ) respectively , with a large spread .
one possible reason is that down - weighting the nuclear region allows less of the point - source component , which has a small scale length , to be attributed to the host .
one of the most interesting results of quasar host galaxy research in the past few years has been the debunking of the textbook myth `` radio - loud quasars live in ellipticals , radio - quiet quasars live in spirals . ''
while early ground - based studies had shown this statement to be an oversimplification ( @xcite ) , the improvement of instruments and the arrival of hst has made more detailed analyses possible .
various studies over the past few years , each including one to several dozen nearby quasars , have claimed that radio - quiet quasars often inhabit non - spiral hosts , and that the elliptical fraction likely increases with nuclear luminosity .
there have been many techniques used to assess the host type .
mcleod & rieke ( 1995 ) combined their ground - based near - ir host magnitudes with visual inspection of wfpc2 images from several groups to determine that there is a strong preference for high - luminosity quasars to have smooth , early - type hosts .
taylor et al .
( 1996 ) applied a 2d @xmath33 modeling technique like that of mdk to their ground - based , near - ir images and found that almost half of the radio - quiet quasars lie in devaucouleurs galaxies .
rnnback et al .
( 1996 ) applied a @xmath33 test to 1d profiles to determine that the radio - quiet objects are found in both ellipticals and spirals ; however , they performed the fits to profiles that had already had the psf removed , the normalization of which can alter their results for reasons they outline .
bahcall et al . (
1997 ) took advantage of the superior resolution of wfpc2 to carry out visual morphological classification of their sample of 20 quasars , and found that more of their radio - quiet objects appeared to be in smooth hosts than in spirals .
although smoothness is not proof of an elliptical , the galaxies are also generally round , unlike the hosts of lower - luminosity quasars imaged with wfpc2 by hooper , impey , & foltz ( 1997 ) .
boyce et al .
( 1998 ) arrived at a similar conclusion for quasars imaged with the wfpc2 pc , based on both morphological and cross - correlation analyses ; for the latter , they selected the model that gave the smaller @xmath33 .
all of the techniques have limitations , and all are based on the possibly faulty premise that hosts are necessarily one type or the other .
nonetheless , we have attempted to discern the galaxy type for the quasars in our ( radio - quiet ) sample by several methods , the results of which are summarized in table [ tab - types ] .
first , we have examined the images and residual maps for evidence of spiral structure . although any star - forming regions will admittedly be harder to see in the near - ir than in the visible , we see clear evidence for arms / tidal features in only pg1322 + 659 and pg1048 + 342 ( which also has a spiral companion ) , and probably pg0947 + 396 .
second , we have used the profiles in fig .
[ fig - profs ] to determined by inspection which curve in each pair ( plotted vs. @xmath21 or @xmath22 ) better approximates a straight line .
we have shown various normalizations for the psf subtraction so the reader can judge the reliability of this technique .
third , we have used the merits from our various 2d fits , where we record the preferred type when the difference in merit is at least 10% .
unfortunately , _ even though the galaxy parameters are little affected , the preference for one type over the other depends on the weighting and masking used .
_ fourth , we apply the mdk @xmath33 test .
lllllll pg0026 + 129 & d & d & - & - & e & - + pg0947 + 396 & e ?
& e & - & e & e & - + pg1048 + 342 & e & d & d & d & d & d + pg1121 + 422 & d & d & - & - & d & - + pg1151 + 117 & d & e & d & - & e & - + pg1322 + 659 & e & - & - & - & d & - + pg1352 + 183 & d & d ? & - & e & e & e + pg1354 + 213 & d & d & d & e & d & d + pg1427 + 480 & d & d & d & e & d & - + pg2233 + 134 & d & d & - & - & d & - + mark876 & d & d ?
& d & e & d & - + um357 & d & d ?
& - & - & e & - + q0530 - 379 & d & d & - & - & e & - + nab1612 + 26 & d & d & - & - & e & - + 1628.6 + 3806 & d & d & - & - & e & - + kuv18217 + 6419 & d & d & - & - & - & - + one might be tempted on mathematical grounds to trust exclusively the @xmath33 statistic used by taylor et al .
( 1996 ) and mdk .
indeed , we find by their method a result similar to theirs , that roughly half of the radio - quiet objects are in devaucouleurs galaxies .
their criterion for preferring a galaxy type was that @xmath47 , where @xmath48 is the value of @xmath33 that gives 99.99% probability for the number of parameters in the fit .
however , for reasons outlined by jahnke & wisotzki ( 2000 ) , this too is problematic .
we therefore also consider the latter authors more conservative @xmath33 criterion , namely that a galaxy type is preferred if it can not be ruled out at 95% confidence , while the alternate model is rejected , say at the 99.9% level ( the results are insensitive to the exact level chosen ) . by this criterion
, we make a clean distinction in only _
one _ case : pg1352 + 183 is an exponential . as discussed above ,
we have likely overestimated the errors in many cases .
we therefore also rescale the errors to achieve @xmath49 and repeat the test .
this adds to the list pg1048 + 342 , with a devaucouleurs host , and pg1354 + 213 , with an exponential host .
interestingly , even by the conservative criteria , we find an exponential host with no spiral arms , and a devaucouleurs host that has arms or tidal features .
table [ tab - types ] shows that the methods used by various authors to discriminate the host types give discrepant results , and hints that all such result should be treated with caution . at least in our sample ,
exponentials and devaucouleurs models are nearly equally good ( or bad ) at fitting the 2d images . as the reader can see from figures [ fig - qimages ] and [ fig - profs ] , neither model is generally perfect , and the results can be sensitive to the fraction of nuclear light attributed to the galaxy .
we note that the methods that down - weight the nuclear region , namely the masked and @xmath33 methods , generate a higher fraction of disk systems .
one possible explanation is that these are disk galaxies with significant bulge components .
when the central regions are neglected , the fit is dominated by the outer , disky components . unfortunately , with the bulges expected to be comparable to the psf size
, we do not feel we can justify adding a bulge component to the disk fits .
however , the facts that arms / tidal features are visible in only two or three of the galaxies , and that all techniques indicate a high fraction of non - disk systems , together _ support the notion that many of the radio - quiet quasars live in non - spiral galaxies_. for completeness , we also examined the distributions of host - galaxy axial ratios @xmath29 . for both exponential and devaucouleurs models ,
there is a lack of hosts with low @xmath29 .
this is suggestive of an early - type population , but could also be the result of a selection effect . at least for low luminosities ,
there is a strong bias against finding active nuclei in edge - on spirals , where obscuration in the plane of the host obscures the nucleus ( @xcite ) .
interactions have long been suspected of triggering and fueling of quasars , and hst images of nearby quasars have revealed some spectacular examples of hosts with tidal distortions ( e.g. @xcite ; @xcite ; @xcite ; @xcite ) . in our sample ,
the fraction of hosts that appear to be involved in strong interactions ( approximately 4/16 ) is similar to what has been found in previous studies .
however , approximately half of the objects appear to have hosts that are smooth and symmetrical at hst resolution .
this is a lower rate than in other studies , though the small numbers make the difference insignificant . for a more detailed discussion of the interaction levels of the quasars from our high - luminosity sample compared to those of low - luminosity quasars ,
see mcleod & rieke 1994b .
mcleod & rieke ( 1995a ) , defined a luminosity / host - mass limit for quasar hosts based on a compilation of ground - based near - infrared data .
this is not a correlation , as recently misinterpreted by some groups , but rather a limit on the maximum nuclear luminosity possible in a host of a given near - ir luminosity ( and presumably mass ) . from the ground - based data
, we found a tight limit well - described by @xmath50 .
mrs interpret the limit in terms of physical parameters and provide a formula for calculating the fraction of the eddington luminosity @xmath51 at which the quasar radiates .
the calculation uses the nuclear b magnitude and the host - galaxy h magnitude along with assumptions about the galaxy mass - to - light ratio , the fraction of the galaxy s mass that resides in the central black hole , and the energy distribution of the nucleus . under these assumptions
, the original @xmath50 line translates to a constant eddington fraction of @xmath52 .
the data for the current project were taken to probe further the luminosity / host - mass limit , and to try intentionally to find galaxies that violate the limit . in figure [ fig - mags ]
we show the host and nuclear magnitudes based on the 1d analysis in this paper . owing to an updated f160w zero point , more accurate saturation limits , and an improved 1d psf - subtraction technique , the host magnitudes in the current paper
do differ , sometimes substantially , from those in mrs .
the zero point accounts for 0.31 mag of the difference .
the saturation limit has contributed to a major revision downward of the host magnitude for kuv1821 + 643 , which percival et al .
( 2000 ) noted had an unrealistically luminous host in mrs .
the improved 1d technique has corrected a tendency for the host flux to be overestimated in quasars with the lowest ratio of host - to - nuclear light ( especially the 6 objects we picked specifically because their hosts had not been detected from the ground , represented by the black pentagons in the figure ) .
the magnitudes derived for our sample from hst data make the limit less of a straight line in mag - mag space , but they reinforce the basic result that the most luminous quasars are not found in low - luminosity galaxies .
we list in table [ tab-1d ] the eddington fractions derived from the current 1d analysis , and note that the results from the 2d devaucouleurs analyses are similar .
the basic mrs conclusion holds , namely that the eddington fractions are generally @xmath53 .
in fact , the only quasar exceeding this limit in the present study is pg1121 + 422 , with @xmath54 , but our detection of this galaxy is suspect . its host could be 3 times fainter without crossing the eddington line .
recently , percival et al . (
2000 ) have probed the limit by observing at k the hosts of 14 high - luminosity , radio - quiet quasars .
they find that their galaxies lie `` to the right of the mcleod & rieke points '' on a plot of nuclear absolute magnitude versus host absolute magnitude , i.e. at higher nuclear luminosity for a given host luminosity than the hosts in our ground - based studies .
as shown in figure [ fig - mags ] , some of our quasars do also .
percival et al .
( 2000 ) also state that our limit is difficult to transfer quantitatively to their figure 5 ( the pg survey b magnitudes we used were shown to be systematically too bright by goldschmidt et al .
( 1992 ) , but only by @xmath55 ; the color transformations from r to b and h to k are sensitive to the assumed spectra ; etc . ) .
however , we can make an approximate comparison by noting that on the mrs plot , the eddington line crosses the @xmath56 host galaxy line two nuclear magnitudes brighter than the envelope described by our ground - based limit .
this is precisely the region occupied by the percival et al .
( 2000 ) quasars .
we conclude that the percival et al .
( 2000 ) quasars also likely are radiating below the eddington limit , though perhaps closer to it .
this is not surprising given that they targeted the most luminous nuclei .
further evidence for consistency with the limit can be seen from their figure 5 , where their quasars overlap with those that mclure et al .
( 1999 ) have shown to be sub - eddington radiators . a re - statement of the luminosity / host - mass limit based on the physical model is to say that quasars today typically radiate below @xmath57 of the eddington rate .
despite choosing quasars that have high nuclear luminosity , and ones for which ground - based attempts to image the hosts had failed , we have not yet found any object that violates the limit corresponding to @xmath58 .
however , we expect such object will be found at higher redshifts according to the predictions of kauffmann & haehnelt ( 2000 ) .
their hierarchical galaxy formation models imply that the massive hosts of today s highest - luminosity quasars formed relatively recently , but that at earlier epochs , luminous quasars will be found in progressively lower - luminosity hosts .
the following notes are based on inspection on an image display of fig .
[ fig - qimages ] as well as our full ( 20 ) reduced images .
the magnitudes of other objects in the fields are simple aperture magnitudes . * smooth host with no distortions visible in psf - subtracted image .
a galaxy with @xmath59 is seen at 47 separation , and a faint compact object is at 31 . *
possible broad spiral structure visible in psf - subtracted image .
this may be tidal in origin , due to an interaction with the large edge - on spiral galaxy with @xmath60 seen at 99 . if at the redshift of the quasar , this spiral is at a projected separation of 28 kpc .
a second galaxy with @xmath61 is at 50 .
* spiral structure visible in psf - subracted image , but the 1d profile indicates a devaucouleurs profile . as with pg0947 , the structure could be tidal in origin , due to interaction with the large spiral ( @xmath62 , and similar in size to the quasar host ) at a projected separation of 34 ( 8.2 kpc if at the quasar s redshift ) .
this galaxy appears to be part of a group .
besides the large spiral , there are at least 5 fainter spirals and distorted galaxies within 18 . *
no host is visible on the images , but the 1d profile suggests that a compact one is present . *
smooth host is visible on the psf - subtracted image .
devaucouleurs fit residual shows an elongated structure right - left ( north - south ) .
an @xmath63 galaxy is at 36 , and a faint point source is roughly twice that far away .
* beautiful 2 armed spiral host with bar - like structure right - left ( north - south ) .
several galaxies , including one that is likely similar in size to the host , are visible on the edge of our full frame at @xmath64
. if at the redshift of the quasar , the large galaxy is at a projected separation of 24 kpc and could be responsible for the host s broad arms . *
smooth host .
a faint ( @xmath65 ) , elongated galaxy is seen at 52 .
* smooth host .
* smooth host .
an @xmath66 galaxy is at 31 . *
smooth host .
there are two small galaxies in our full frame : @xmath67 at 36 , @xmath66 at 97 . *
[ also called pg1613 + 658 ] severely distorted host . the host galaxy
filled so much of the frame that removal of the nicmos quadrant effect was problematic . as a result ,
the frame is not as flat as for the other objects .
however , the nearly triangular outer contours with a straight left edge is not an artifact , but is also visible on ground - based images .
what the nicmos image reveals is that a companion , with @xmath68 and at a projected separation of 22 ( 4.3 kpc ) , is appears to be embedded within the host .
the psf - subtracted image shows large - scale , asymmetric , tidal features that probably resulted from this interaction .
* host is very compact , and too small for structure to be visible . *
smooth host .
there are several other galaxies visible in the full frame : @xmath69 at 42 , @xmath70 at 65 , @xmath71 at 87 , and @xmath67 at 97 .
* host is compact but clearly elongated left - right ( northwest - southeast ) .
two diffuse , faint galaxies are visible in our full frame : @xmath63 at 60 , and @xmath72 at 92 . * psf - subtracted image shows asymmetric extension to lower left ( northeast ) .
there are two small galaxies in the frame : @xmath73 at 35 , and @xmath74 at 56 .
there is also a star at 56 .
* 1d profile shows that this is clearly a devaucouleurs host , yet devaucouleurs fits converged only when masked .
this is perhaps due to interference from a linear feature extending up ( southwest ) from the host .
however , it is also possible that the extremely high nuclear brightness has compromised the fitting ; it has by far the brightest nucleus in our sample , and the host - to - nucleus flux ratio is low .
still , the galaxy is intrinsically the most luminous in our sample .
there are two galaxies at 95 ( @xmath75 ) seen on opposite sides of the quasar in our full frame .
we have imaged 16 low - redshift , high - luminosity quasars with nicmos . using stellar images as psfs has worked extremely well , and nicmos has done a superb job of showing the host galaxies with high contrast against the bright nucleus .
we find the following results .
* for redshifts @xmath76 , the host - galaxy magnitudes derived from ground - based data using our 1d analysis technique are very reliable ( within @xmath77 ) unless there are companions that can not be resolved . therefore , general conclusions about host luminosities from our previous papers should be robust . * distinguishing between devaucouleurs and exponential hosts based on various fitting techniques is questionable , especially given that neither law is likely perfect .
for 2d fits , we recommend a very conservative @xmath33 criterion for discrimination .
while some previously used methods appear to distinguish between types in almost every case , they are often at odds with the more conservative criterion , which can distinguish between types in only @xmath78 cases out of 16 .
the 1d radial profiles are also very useful .
fortunately , hst allows us to look directly for spiral arms and other morphological features in the hosts .
we concur with previous studies that have found radio - quiet quasars often live in devaucouleurs hosts . *
approximately 4 of the 16 hosts are undergoing strong interactions with companions .
however , nearly half of the hosts are smooth and symmetric , a reminder that current - epoch interactions are not a necessary condition of quasar activity . * assuming that galaxies contain central black holes with 0.6% of the galaxy s mass ( see mrs for details ) , 15 of the 16 quasars in our sample radiate at @xmath0 of the eddington rate .
our host detection for pg1121 + 422 is uncertain , so its rate of 31% eddington may be a lower limit . despite intentionally choosing high - luminosity quasars whose hosts were hard to detect from the ground , we have failed to find any object that violates a luminosity / host - mass limit corresponding to @xmath79 thanks to george rieke and lisa storrie - lombardi for their work on the hst proposal and mrs , and to lisa also for especially valuable comments on this paper .
we thank the nicmos team for putting such a superb instrument in orbit , erin condy for help with the data reduction , ray weymann for helpful ideas during the course of this project , and marcia rieke for the calibration files used with @xmath80 .
we thank the anonymous referee for comments that helped us to improve the presentation significantly .
support for this work was provided by nasa through grant number go-07421.01 - 96a from the space telescope science institute , which is operated by the association of universities for research in astronomy , inc .
, under nasa contract nas5 - 26555 .
we also gratefully acknowledge support from the keck northeast astronomy consortium .
|
we have obtained near - infrared camera and multi - object spectrometer images of 16 radio quiet quasars observed as part of a project to investigate the `` luminosity / host - mass limit . ''
the limit results were presented in mcleod , rieke , & storrie - lombardi ( 1999 ) . in this paper , we present the images themselves , along with 1- and 2-dimensional analyses of the host galaxy properties .
we find that our model - independent 1d technique is reliable for use on ground - based data at low redshifts ; that many radio - quiet quasars live in devaucouleurs - law hosts , although some of the techniques used to determine host type are questionable ; that complex structure is found in many of the hosts , but that there are some hosts that are very smooth and symmetric ; and that the nuclei radiate at @xmath0 of the eddington rate based on the assumption that all galaxies have central black holes with a constant mass fraction of 0.6% . despite targeting hard - to - resolve hosts ,
we have failed to find any that imply super - eddington accretion rates .
|
one of the most fascinating phenomena in physics is that of a phase transition .
initially observed in macroscopic systems and in electromagnetic interactions , phase transitions have been seen manifesting also in strongly interacting microscopic systems and nowadays two specific areas are receiving a great deal of attention .
one involves the loss of stability of excited nuclear systems which , under certain conditions of temperature and density , may lead to the total disassembly of the nucleus into particles and fragments .
the second , at much higher energies , concerns the transition from hadrons to quarks and gluons , and the possibility of observing new phenomena in quark matter .
+ nucleus - nucleus collisions at intermediate and relativistic energies have been shown @xcite to be an ideal tool to produce pieces of finite nuclear matter at extremely different thermodynamical conditions . in order to analyze thermodynamical properties of microscopic systems
we need to produce and select samples which can be associated with microcanonical ensembles at high excitation energies . + in a series of experiments @xcite , multifragment decays of projectile spectators have been studied with the aladin forward - spectrometer at the sis accelerator ( gsi - darmstadt - germany ) . in these collisions , energy depositions are reached , which cover the range from particle evaporation to multifragment emission and further to the total disassembly of the system .
the most prominent feature of the multifragment decay is the universality of the fragment multiplicities and the fragment charge correlations .
the loss of memory of the entrance channel is an indication that equilibrium is attained prior to the fragmentation stage of the reaction .
clearly , new experiments are mandatory for having a better knowledge of the thermodynamics of a finite nucleus and its decay .
+ recently a systematic investigation on projectile - spectator fragmentation has been undertaken at the aladin spectrometer at the gsi @xcite : four different projectiles , @xmath0au , @xmath1la , @xmath1sn and @xmath2sn , all with an incident energy of 600 amev on @xmath3sn and @xmath0au targets , have been studied .
the fragments emerging from the decay of the projectile spectators are well localized in rapidity .
the distributions are concentrated around a rapidity value very close to the beam rapidity and become increasingly narrower with increasing mass of the fragment .
the sorting out of the collision according to its centrality is done using the observable @xmath4 defined as the sum of the charges of the projectile fragments ( @xmath5 ) .
to first approximation @xmath4 is a measure of the charge of the remaining projectile - spectator except for the number of evaporated z=1 .
it follows that large values of @xmath4 correspond to big projectile - spectators , i.e. peripheral collisions , whereas small values of @xmath4 correspond to small projectile - spectators , i.e. more central collisions .
the excitation of the projectile - spectator depends on the impact parameter , since it occurs via knocking - out nucleons : @xmath4 is therefore not only a measure of the impact parameter , but also of the excitation energy .
+ in the left panel of fig .
[ fig : charge ] the charge distributions measured in the fragmentation of @xmath1sn for 9 different bins of @xmath4 are shown .
the shapes of the charge distributions as a function of @xmath4 are similar for all the studied systems .
we have fitted these charge distributions with a power - law parameterization @xmath6 .
the fitting range was @xmath7 .
the right panel of fig .
[ fig : charge ] presents the evolution of the @xmath8 parameter as a function of @xmath9 .
the @xmath8 parameter for all the three reactions lies on an almost universal curve and shows a minimum near @xmath9=0.5 .
specific isotopic effects , even though small , can be nevertheless observed : in particular , the inversion in the hierarchy of the three studied systems by going from small to high values of @xmath4 ( the values for @xmath1sn are larger than those for the other two systems in central collisions and become slowly smaller than them by going towards larger impact parameters ) has been shown to be related to an isospin dependence of the surface energy as a function of the excitation energy of the system @xcite .
+ sn fragmentation for different bin of @xmath4 .
right panel : the extracted @xmath8 parameters as a function of @xmath4 for @xmath1la , @xmath1sn and @xmath2sn at 600 amev , compared with the values of @xmath0au at the same energy @xcite . ]
sn fragmentation for different bin of @xmath4 .
right panel : the extracted @xmath8 parameters as a function of @xmath4 for @xmath1la , @xmath1sn and @xmath2sn at 600 amev , compared with the values of @xmath0au at the same energy @xcite . ]
since long it has been observed that some characteristics of the exited nuclear - system , such as the largest emitted charge ( @xmath10 ) and the asymmetry between the three largest - charges @xmath11 vary by going from the evaporation to the vaporization regions through the multifragmentation regime .
if the lowly - excited nuclear system evaporates light fragments , large @xmath10 and @xmath12 are expected , whereas in case of vaporization , both @xmath10 and @xmath12 are expected to be small .
[ bimod ] shows the event distributions in the @xmath10 and @xmath12 plane for different cuts of @xmath4 in the case of the @xmath1sn projectile .
-@xmath12 plane for the @xmath13 projectile .
each panel refers to a different cut in @xmath14 . ] for large values of @xmath4 , i.e. small excitation energies , the system is in the evaporation regime and therefore almost all the events present high values of @xmath10 and an @xmath12 value near 1 ( lower panels in fig .
[ bimod ] ) .
the distribution of @xmath10 as well as for @xmath12 , is characterized by the presence of a single peak . by going towards smaller values of @xmath4 ,
the event distributions start to populate the region at intermediate @xmath10 and @xmath12 until , in the multifragmentation region , events with high @xmath10 and @xmath12 and small @xmath10 and @xmath12 are simultaneously present ( middle panels of fig . [ bimod ] ) .
this is the region where the two variables exhibit the largest fluctuations and correspondingly a bimodal distribution ( central panel in fig . [ bimod ] ) .
when , finally , @xmath4 is small enough , all the events are located in the vaporization zone ( upper panels in fig .
[ bimod ] ) characterized by simultaneously small values of @xmath10 and @xmath12 .
similar behaviors have been found by analyzing other systems @xcite . in particular
this analysis has also been extended to projectile - spectator systems studied by the aladin group in previous experiments @xcite in order to explore a wider mass - range : for each system the occurrence of a bimodal distribution for a specific impact parameter has been observed . +
whether such results could be interpreted as a genuine bimodal signal characteristic of a first - order phase transition is still a matter of investigation : indeed the observation of a bimodal distribution in the data could be the result of a mixing of residue and multifragment emission for some impact parameter bins .
signals of the predicted phase coexistence at low density and temperature ( @xmath16 and @xmath17 ) , often associated with multifragmentation , may be revealed by abnormal fluctuations of experimental observables @xcite .
the theory of the universal character of order parameter fluctuations in finite systems @xcite addresses this question in a model - independent framework , for systems with a second - order phase transition .
experimental observables that may be related to an order - parameter can be identified , in particular through their @xmath15-scaling behavior , and recently a model - independent tracking of criticality signals in nuclear multifragmentation data at intermediate energies has been proposed @xcite .
+ the @xmath15-scaling is observed when two or more probability distributions @xmath18 $ ] of the observable _ m _ for a system size _
n _ collapse onto a single scaling curve @xmath19 independent of system size when plotted in terms of the scaling variables : @xmath20= \phi(z_{(\delta ) } ) = \phi \left ( \frac{m-\left < m \right > ^{\delta } } { \left < m \right >
^{\delta}}\right ) \label{eqn : dscaling}\ ] ] where @xmath21 is the mean value of the distribution @xmath22 $ ] and 1/2@xmath231 .
@xmath21 plays the role of a scale parameter and can replace _
n _ as a measure of the size of the system .
the scaling law with @xmath15=1/2 is associated with low temperature ( _ ordered _ systems ) , or with observables which are not related to an order parameter . scaling with @xmath15=1
is seen at high temperature ( _ disordered _ system ) and also for systems at critical conditions . a necessary condition for _ m _ to be an order
parameter is that it must exhibit a corresponding change of @xmath15-scaling regime when some suitable control parameter ( e.g. available energy , temperature , etc . )
is varied .
+ using the data from this experiment , it has been investigated whether the @xmath10 distributions follow the @xmath15-scaling law .
the multiplicity of charged particles measured with the catania hodoscope @xcite was used as impact parameter selector in oder to avoid a possible autocorrelation of @xmath4 with @xmath10 .
an additional selection has , moreover , been applied to the data by defining two cuts on the asymmetry variable : @xmath24 and @xmath25 corresponding to the _ liquid _ and _ gas _ phases respectively .
the correlation between the mean value and the fluctuation of @xmath10 is shown in fig .
[ dsca1 ] .
the dotted lines represent results of the fits providing @xmath26 1 for high excitation energy ( left - hand side of fig .
[ dsca1 ] ) and @xmath26 0.5 for low excitation energy ( right - hand side of fig .
[ dsca1 ] ) .
the form of the @xmath10 distributions also evolves with excitation energy : it is nearly gaussian in the @xmath15=1/2 regime ( right panel of fig .
[ dsca ] ) beside for @xmath15=1 the shape is asymmetric with a near - exponential tail for large positive values of the scaling variable , @xmath27 ( left panel of fig .
[ dsca ] ) : this distribution approaches that of the modified gumbel - distribution @xcite .
+ interestingly enough the transition from the _ ordered _ to the _ disordered _ regimes is observed for the same impact parameter bin , for which the appearance of a bimodal distribution has been observed ( see fig . [ bimod ] ) .
similar results have been obtained in the study of the _ quasi_-projectile decay at lower energies @xcite .
the interpretation of the @xmath15-scaling result is still an open question .
the two @xmath15-scaling regimes seem to suggest that @xmath10 is behaving like the order - parameter of a second - order phase transition .
moreover , other analyses of similar data have found evidence for multifragmentation being a first - order phase transition , through signals of liquid - gas coexistence in the form , for example , of a flattening of the caloric curve @xcite .
this apparent contradiction could be understood as a finite - size effect @xcite .
the fact that many reactions exhibit a universal behavior , dependent only on the available energy , can again be an indication of a microcanonical equilibrium .
+ + _ c.sf . acknowledges the receipt of an alexander - von - humboldt fellowship .
this work was supported by the european community under contract no .
rii3-ct-2004 - 506078 and hpri - ct-1999 - 00001 and by the polish scientific research committee under contract no .
2p03b11023 . _ 9 j. pochodzalla et al .
, prog .
39 ( 1997 ) 443 .
a. schttauf et al .
, nucl .
phys . a 607 ( 1996 ) 457 .
chomaz and f. gulminelli in `` dynamics and thermodynamics of systems with long range interactions '' , eds . t. dauxois et al , springer ( 2002 ) .
c.sfienti et al .
, nucl .
phys . a 734 ( 2004 ) 528 .
a. botvina et al .
, phys .
c in press , nucl - th/0606060 .
p. kreutz et al .
, nucl .
phys . a 556 ( 1993 ) 672 .
o. lopez and m. f. rivet in dynamics and thermodynamics with nuclear degrees of freedom , edited by ph .
chomaz et al .
j. a , in press ( 2006 ) . r. botet and m. ploszajczak phys .
e 62 ( 2000 ) 1825 .
j. frankland et al .
, phys .
c 71 ( 2005 ) 034607 .
m. de napoli , phd thesis , university of catania , december 2005 .
j. pochodzalla et al , phys .
75 ( 1995 ) 1040 .
+ j. b. natowitz et al .
, phys .
c 65 ( 2002 ) 034618 .
f. gulminelli et al .
, phys .
c 71 ( 2005 ) 054607 .
|
a systematic study of isotopic effects in the break - up of projectile spectators at relativistic energies has been performed with the aladin spectrometer at the gsi laboratory .
+ searching for signals of criticality in the fragment production we have applied the model - independent universal fluctuations theory already proposed to track criticality signals in multifragmentation to our data .
the fluctuation of the largest fragment charge and of the asymmetry of the two and three largest fragments and their bimodal distribution have also been analysed .
|
measurements of hadronic states with charm quarks such as the @xmath3 have led to extensive and still ongoing discussions about our deeper understanding of mesons and baryons @xcite , traditionally thought to be composed of a pair of quark and antiquark or three quarks in the naive quark model . with its mass ( @xmath4 mev )
about 100 mev lower than the lowest @xmath5 scalar state in the naive quark model , the @xmath1 can not be a conventional @xmath6 state @xcite .
one possible interpretation is that of a compound dynamically generated by the strong @xmath2 interaction in coupled - channels dynamics @xcite .
such approaches have provided many useful insights into the nature of some most intriguing new resonances ( see , e.g. , refs .
@xcite for some recent applications ) . in order to clarify the nature of the @xmath3 , or of any other meson of similar kind ,
it is useful to study such objects from various perspectives and compare the results with experimental and lattice qcd ( lqcd ) data . in this respect
, it has been argued that the isospin - breaking decay width @xmath7 @xcite , the light - quark mass dependence @xcite , and the volume dependence @xcite of the @xmath3 properties can provide valuable information on its nature . at the same time
it should also be noted that , in addition to the @xmath3 , coupled - channels unitary dynamics predicts several other states in sectors or channels related to the @xmath3 by heavy - quark spin and flavor symmetry and ( approximate ) chiral symmetry [ or broken su(4 ) symmetry ] @xcite .
once the mass and width of the @xmath3 are fixed , so are those of the other related states . future experiments in search for those resonances in the predicted energy regions are therefore strongly encouraged .
all these predictions are subject to potentially sizable symmetry - breaking corrections . in particular , a comprehensive study of recoil corrections is necessary because the velocity of the charm quark in @xmath8 mesons is only about @xmath9 , not small enough to allow for a complete neglect of recoil corrections . for the scattering lengths of the nambu - goldstone bosons off the @xmath0 mesons , such a study has been performed in ref .
@xcite , and it was shown that indeed recoil corrections are sizable . in refs .
@xcite , covariant chiral perturbation theory ( chpt ) , supplemented with the extended - on - mass - shell ( eoms ) scheme , was applied to study the decay constants of the @xmath8/@xmath10 mesons .
it was shown that the covariant chpt converges faster than its nonrelativistic ( heavy - meson ) counterpart .
these findings can , to some extent , be deemed as repercussions of the one - baryon sector .
for instance , it has been shown that the eoms formulation of the baryon chpt is capable of better describing three - flavor observables and their light - quark mass evolutions than its nonrelativistic ( heavy - baryon ) counterpart , see , e.g. , refs .
@xcite and references cited therein . in the present work we study the interactions of the heavy - light mesons ( @xmath0 , @xmath11 , @xmath12 , @xmath13 and their strange counterparts ) with nambu - goldstone bosons ( the octet of the lightest pseudoscalar mesons ) in covariant chpt and its unitary version .
we calculate the interaction potentials up to next - to - leading order ( nlo ) and perform an iteration of these potentials to all orders using the bethe - salpeter equation .
it was pointed out that in the covariant calculation of the loop function appearing in the bethe - salpeter equation , one loses the heavy - quark spin and flavor symmetry @xcite .
we study this problem in detail and propose a new renormalization scheme , similar in spirit to the eoms scheme widely used in the one - baryon sector @xcite and also used in refs .
@xcite , to recover heavy - quark spin and flavor symmetry up to @xmath14 corrections , where @xmath15 is a generic heavy - light meson mass .
we apply our approach to describe the most recent fully dynamical lqcd simulations for the scattering lengths of nambu - goldstone bosons off the @xmath0 mesons @xcite and fix the relevant low - energy and subtraction constants . , @xmath16 , and @xmath2 scattering lengths
have also been calculated on the lattice using both quark - antiquark and meson - meson interpolating fields @xcite and the @xmath1 is found to be a bound state in the @xmath2 channel @xcite .
] we then solve the corresponding bethe - salpeter equations and search for poles in the complex energy plane , identified as dynamically generated states .
we show that a number of @xmath17 and @xmath18 states emerge naturally , including the @xmath1 , the @xmath19 and their bottom - quark counterparts .
charm sector , and in addition ref .
@xcite studied the preliminary lqcd results of ref .
@xcite . ] this article is organized as follows . in sec .
ii , the relevant terms of the effective chiral lagrangian are summarized and the driving potentials up to nlo are constructed . in sec .
iii we propose a new renormalization scheme to be used in the bethe - salpeter equation , which manifestly satisfies the chiral power counting rules and heavy - quark spin and flavor symmetries .
we discuss the advantage of this scheme in comparison with others widely used in unitary chpt . in sec .
iv , we apply both the unitary heavy - meson and covariant formulations of chpt to fit the lqcd data and make predictions for the existence of a number of dynamically generated resonances in both the charm and the bottom sectors . a short summary is given in sec .
introducing the chiral effective lagrangians in the present context , one first has to specify a power counting rule . in the present work ,
the nambu - goldstone boson ( ngb ) masses @xmath20 and the field gradients @xmath21 are counted as @xmath22 as usual , where @xmath23 denotes a ngb boson of the pseudoscalar octet . for @xmath0 mesons , the triplets are @xmath24 and @xmath25 , and for @xmath26 mesons , they are @xmath27 and @xmath28 . their field gradients @xmath29 and @xmath30 and masses @xmath31 and @xmath32 are counted as @xmath33 . the ngb propagator @xmath34 is counted as @xmath35 , while the heavy - light pseudoscalar and vector meson propagators @xmath36 and @xmath37 are counted as @xmath38 .
the chiral order of the propagators of the heavy - light mesons can be understood as follows . as in standard heavy - meson ( hm ) chpt ,
one can write the momentum @xmath39 as a sum of a large component and a residual small component , i.e. , @xmath40 , where @xmath41 is the velocity of the heavy - light meson and @xmath42 is the small residual component counted as @xmath43
. therefore , the heavy - light pseudoscalar meson propagator becomes @xmath44 , which is counted as @xmath38 .
the same is true for the heavy - light vector meson propagator .
the leading order covariant chiral lagrangian describing the interactions of the ngbs with the heavy - light pseudoscalar and vector mesons has the following form : @xmath45 where @xmath46 and @xmath32 are the @xmath47 and @xmath48 masses in the chiral limit , respectively , and @xmath49 denotes trace in the @xmath50 , @xmath51 , and @xmath52 flavor space .
the coupling constant @xmath53 has mass dimension @xmath54 , whereas @xmath55 is dimensionless .
the axial current is defined as @xmath56 and the chiral covariant derivative is @xmath57 with the vector current @xmath58 . in these equations , @xmath59 with @xmath60 being the ngb decay constant in the chiral limit and @xmath61 collecting the octet of ngb fields : @xmath62 the coupling @xmath63 is known empirically .
it can be determined from the decay width @xmath64 together with the branching ratio @xmath65 @xcite . at tree level , @xmath66 , which gives @xmath67 .
the coupling @xmath68 can be related to @xmath69 through the heavy - quark spin symmetry , i.e. , @xmath70 , keeping in mind that there could be sizable deviations of higher order in @xmath71 . the couplings @xmath72 and @xmath73 can be related to their @xmath0 counterparts through heavy - quark flavor symmetry . in a similar way , one can construct the covariant nlo terms of the effective lagrangian : @xmath74\mathcal d_\nu p^\dagger\rangle]\nonumber\\ & & + 2[\tilde c_0\langle p^*_\mu
p^{*\mu\dagger}\rangle\langle \chi_+\rangle-\tilde c_1\langle p^*_\mu \chi_+ p^{*\mu\dagger}\rangle- \tilde c_2\langle p^*_\mu p^{*\mu\dagger}\rangle\langle u^\mu u_\mu\rangle-\tilde c_3\langle p^*_\nu u^\mu u_\mu p^{*\nu\dagger}\rangle\nonumber\\&&+\frac{\tilde c_4}{m_{p^*}^2}\langle \mathcal d_\mu p^*_\alpha\mathcal d_\nu p^{*\alpha\dagger}\rangle\langle\{u^\mu , u^\nu\}\rangle+\frac{\tilde c_5}{m_{p^*}^2}\langle \mathcal d_\mu p^*_\alpha\{u^\mu , u^\nu\ } \mathcal d_\nu p^{*\alpha\dagger}\rangle\nonumber\\&&+\frac{\tilde c_6}{m_{p^*}^2}\langle \mathcal d_\mu p_\alpha^{*}[u^\mu , u^\nu]\mathcal d_\nu p^{*\alpha\dagger}\rangle]\,,\label{nlolag}\end{aligned}\ ] ] where @xmath75 with @xmath76 . in the infinite heavy - quark mass limit
, one has @xmath77 for @xmath78 and @xmath79 . for the numerical results presented in this work , we have fixed the @xmath31 and @xmath32 appearing in eq .
( [ nlolag ] ) , which are needed to make the @xmath80 , @xmath81 , and @xmath82 low - energy constants ( lecs ) dimensionless , to the following values : @xmath83 and @xmath84 ( see table [ table : par ] ) , where @xmath85 ( @xmath86 ) is the su(3 ) average of strange and non - strange d(b ) and @xmath87 masses .
such a choice is taken in order to avoid introducing su(3 ) breaking corrections to the lecs by hand in the covariant framework . as a first estimate of the size of spin or flavor symmetry - breaking effects
, one can determine the constants @xmath88 and @xmath89 from the masses of strange and nonstrange @xmath0 and @xmath11 mesons .
at the nlo chiral order , the masses of the @xmath0 , @xmath90 , @xmath11 and @xmath91 mesons are given by @xmath92 where the @xmath93 meson mass in the chiral limit is denoted as @xmath94 @xmath95 .
inserting the physical masses listed in table [ table : par ] leads to @xmath96 and @xmath97 . repeating the same argument for the @xmath98 mesons , we obtain @xmath99 and @xmath100 .
the heavy - quark flavor symmetry dictates that @xmath101=const . using an su(3)-averaged mass for @xmath15 for each sector , we find @xmath102 gev@xmath103 , @xmath104 gev@xmath103 , @xmath105 gev@xmath103 , and @xmath106 gev@xmath103 .
these numbers provide a hint about the expected order of magnitude for the breaking of heavy - quark spin and flavor symmetry : about @xmath107 between @xmath0 vs @xmath11 and @xmath12 vs @xmath13 , whereas it amounts to about @xmath108 between @xmath0 vs @xmath12 and @xmath11 vs @xmath13 . in this section
we derive the chiral potentials contributing to @xmath109 scattering up to nlo in both the covariant and the heavy - meson formulations .
the corresponding feynman diagrams are shown in figs .
[ lofeynman ] and [ vectorlofeynman ] . at lo ( a)-(c ) and nlo ( d )
chiral order .
the pseudoscalar ( @xmath47 ) mesons are represented by solid lines , the vector ( @xmath48 ) mesons by double lines , and the nambu - goldstone bosons by dashed lines.,scaledwidth=100.0% ] at lo ( a)-(e ) and nlo ( f ) chiral order .
the pseudoscalar ( @xmath47 ) mesons are represented by solid lines , the vector ( @xmath48 ) mesons by double lines , and the nambu - goldstone bosons by dashed lines . , scaledwidth=80.0% ] for the processes @xmath111 , the leading order ( lo ) potential can be written as @xmath112 where @xmath113 , @xmath114 and @xmath115 are the weinberg - tomozawa term and the @xmath52- and @xmath116-channel exchange contributions , respectively .
the weinberg - tomozawa term @xmath113 has the following form : @xmath117 with the mandelstam variables @xmath118 and @xmath119 .
the coefficients @xmath120 for different strangeness and isospin combinations @xmath121 are listed in table [ coefficients ] .
the @xmath122-channel exchange terms @xmath114 and @xmath115 are suppressed by @xmath14 compared to the weinberg - tomozawa term . at threshold
they are in fact of second chiral order and can be absorbed into the available @xmath123 lecs .
in addition , we have checked numerically that they play a negligible role in the present study , and hence we neglect their contributions in the following . the same statements hold for the @xmath50-channel exchange diagrams in the @xmath124 process .
the @xmath52-channel diagrams in the @xmath124 process do not contribute to @xmath125-wave interactions .
[ cols="^,^,^,^,^,^,^,^,^ " , ] in table [ tab : boom ] we compare the predicted @xmath17 and @xmath18 states from several different formulations of uchpt in the bottom sector .
it is seen that the absolute positions can differ by as much as 80 mev , which is not surprising because the heavy - quark flavor symmetry was implemented differently . and the @xmath19 . [
fig : pole ] ] it has been argued that the light - quark mass evolution of the masses of mesons and baryons can provide important hints about their nature ( see , e.g. , refs .
@xcite ) . in the left panel of fig .
[ fig : pole ] , we show how the pole positions of the @xmath1 and the @xmath19 evolve as a function of @xmath126 .
the strange - quark mass is fixed to its physical value using the leading - order chpt .
the light - quark mass dependences of the @xmath127 and @xmath128 are given by the nlo chpt formulas of eqs .
( [ massform1])-([massform4 ] ) .
the right panel of fig .
[ fig : pole ] shows the evolution of the @xmath1 and @xmath19 pole position as a function of the kaon mass ( or equivalently the strange - quark mass ) as we fix the pion mass to its physical value . as has been argued in ref .
@xcite , the feature of being dynamically generated dictates that the dependences of the masses of these states on @xmath129 are linear with a slope close to unity , which can be clearly seen from fig .
[ fig : pole ] .
we have studied the latest fully dynamical lqcd simulations for the scattering lengths of nambu - goldstone bosons off @xmath0 mesons in covariant chiral perturbation theory and its unitary version up to next - to - leading order .
it is shown that the covariant ( u)chpt describes the lqcd data better than its nonrelativistic ( heavy - meson ) counterpart .
in addition , we show that the @xmath1 can be dynamically generated without _ a priori _ assumption of its existence .
we have proposed a new subtraction scheme to ensure that the loop function appearing in the bethe - salpeter equation satisfies the chiral power counting rule and has a well - defined behavior in the limit of infinite heavy - quark mass .
it is shown that this scheme has a similar @xmath14 scaling as the hmchpt loop function but provides a better description of the light - quark mass dependence of the lqcd scattering lengths , in agreement with the findings in the one - baryon sector .
with such a scheme , we have predicted the counterparts of the @xmath1 in the @xmath130 sector and in the bottom sector .
the experimental confirmation of the dynamically generated states in the bottom sector can serve as a stringent test of our theoretical model and the interpretation of the @xmath1 as a dynamically generated state from the strong @xmath2 interaction .
this work is supported in part by bmbf , by the a.v .
humboldt foundation , the fundamental research funds for the central universities , the national natural science foundation of china ( grant no .
11005007 ) , the new century excellent talents in university program of ministry of education of china under grant no .
ncet-10 - 0029 , the dfg cluster of excellence origin and structure of the universe , " and by dfg and nsfc through the sino - german crc 110 symmetries and emergence of structure in qcd . "
b. aubert _ et al . _ [ babar collaboration ] , phys .
* 90 * , 242001 ( 2003 ) [ hep - ex/0304021 ] .
p. krokovny _ et al . _ [ belle collaboration ] , phys .
lett . * 91 * , 262002 ( 2003 ) [ hep - ex/0308019 ] .
d. besson _ et al . _
[ cleo collaboration ] , phys .
d * 68 * , 032002 ( 2003 ) [ erratum - ibid .
d * 75 * , 119908 ( 2007 ) ] [ hep - ex/0305100 ] .
w. a. bardeen , e. j. eichten and c. t. hill , phys .
d * 68 * , 054024 ( 2003 ) .
m. a. nowak , m. rho and i. zahed , acta phys .
b * 35 * , 2377 ( 2004 ) .
e. van beveren and g. rupp , phys .
lett . *
91 * , 012003 ( 2003 ) .
y. b. dai , c. s. huang , c. liu and s. l. zhu , phys .
d * 68 * , 114011 ( 2003 ) .
s. narison , phys .
b * 605 * , 319 ( 2005 ) .
y. q. chen and x. q. li , phys .
* 93 * , 232001 ( 2004 ) .
a. p. szczepaniak , phys .
b * 567 * , 23 ( 2003 ) .
t. e. browder , s. pakvasa and a. a. petrov , phys .
b * 578 * , 365 ( 2004 ) .
h. y. cheng and w. s. hou , phys .
b * 566 * , 193 ( 2003 ) .
t. barnes , f. e. close and h. j. lipkin , phys .
d * 68 * , 054006 ( 2003 ) .
e. e. kolomeitsev and m. f. m. lutz , phys .
b * 582 * , 39 ( 2004 ) .
f. k. guo , p. n. shen , h. c. chiang and r. g. ping , phys .
b * 641 * , 278 ( 2006 ) .
d. gamermann , e. oset , d. strottman and m. j. vicente vacas , phys .
d * 76 * , 074016 ( 2007 ) .
s. -l .
zhu , int .
j. mod .
e * 17 * , 283 ( 2008 ) [ hep - ph/0703225 ] . c. garca - recio , l. s. geng , j. nieves , l. l. salcedo , e. wang and j. -j .
xie , phys .
d * 87 * , 096006 ( 2013 ) [ arxiv:1304.1021 [ hep - ph ] ] .
m. f. m. lutz and m. soyeur , nucl .
phys . a * 813 * , 14 ( 2008 ) [ arxiv:0710.1545 [ hep - ph ] ] . f. -k .
guo , c. hanhart , s. krewald and u. -g .
meissner , phys .
b * 666 * , 251 ( 2008 ) [ arxiv:0806.3374 [ hep - ph ] ] .
m. altenbuchinger , l. s. geng and w. weise , phys .
b * 713 * , 453 ( 2012 ) [ arxiv:1109.0460 [ hep - ph ] ] .
l. s. geng , j. martin camalich , l. alvarez - ruso and m. j. vicente vacas , phys .
* 101 * , 222002 ( 2008 ) [ arxiv:0805.1419 [ hep - ph ] ]
. j. martin camalich , l. s. geng and m. j. vicente vacas , phys .
d * 82 * , 074504 ( 2010 ) [ arxiv:1003.1929 [ hep - lat ] ] .
l. -s .
geng , front .
phys . china .
* 8 * , 328 ( 2013 ) [ arxiv:1301.6815 [ nucl - th ] ] .
j. gegelia and g. japaridze , phys .
d * 60 * , 114038 ( 1999 ) [ hep - ph/9908377 ] .
t. fuchs , j. gegelia , g. japaridze and s. scherer , phys .
d * 68 * , 056005 ( 2003 ) [ hep - ph/0302117 ] .
l. liu , k. orginos , f. -k .
guo , c. hanhart and u. -g .
meissner , phys . rev .
d * 87 * , 014508 ( 2013 ) [ arxiv:1208.4535 [ hep - lat ] ] .
d. mohler , s. prelovsek and r. m. woloshyn , phys .
d * 87 * , no . 3 , 034501 ( 2013 ) [ arxiv:1208.4059 [ hep - lat ] ] .
d. mohler , c. b. lang , l. leskovec , s. prelovsek and r. m. woloshyn , arxiv:1308.3175 [ hep - lat ] .
wang and x. g. wang , phys .
d * 86 * , 014030 ( 2012 ) [ arxiv:1204.5553 [ hep - ph ] ] .
l. liu , h. -w .
lin and k. orginos , pos lattice * 2008 * , 112 ( 2008 ) [ arxiv:0810.5412 [ hep - lat ] ] .
j. beringer _
[ particle data group collaboration ] , phys .
d * 86 * , 010001 ( 2012 ) .
l. m. abreu , d. cabrera , f. j. llanes - estrada and j. m. torres - rincon , annals phys . * 326 * , 2737 ( 2011 ) [ arxiv:1104.3815 [ hep - ph ] ] .
m. f. m. lutz and e. e. kolomeitsev , nucl .
a * 730 * , 392 ( 2004 ) [ nucl - th/0307039 ] .
l. roca , e. oset and j. singh , phys .
d * 72 * , 014002 ( 2005 ) [ hep - ph/0503273 ] .
n. kaiser , p. b. siegel and w. weise , nucl .
a * 594 * , 325 ( 1995 ) [ nucl - th/9505043 ] .
a. dobado and j. r. pelaez , phys .
d * 56 * , 3057 ( 1997 ) [ hep - ph/9604416 ] .
j. a. oller and e. oset , nucl .
a * 620 * , 438 ( 1997 ) [ erratum - ibid .
a * 652 * , 407 ( 1999 ) ] [ hep - ph/9702314 ] .
e. oset and a. ramos , nucl .
a * 635 * , 99 ( 1998 ) [ nucl - th/9711022 ] .
j. a. oller , e. oset and j. r. pelaez , phys .
d * 59 * , 074001 ( 1999 ) [ erratum - ibid .
d * 60 * , 099906 ( 1999 ) ] [ erratum - ibid .
d * 75 * , 099903 ( 2007 ) ] [ hep - ph/9804209 ] .
n. kaiser , eur .
phys . j. a * 3 * , 307 ( 1998 ) .
|
recent lattice qcd simulations of the scattering lengths of nambu - goldstone bosons off the @xmath0 mesons are studied using unitary chiral perturbation theory .
we show that the lattice qcd data are better described in the covariant formulation than in the heavy - meson formulation .
the @xmath1 can be dynamically generated from the coupled - channels @xmath2 interaction without _ a priori _ assumption of its existence .
a new renormalization scheme is proposed which manifestly satisfies chiral power counting rules and has well - defined behavior in the infinite heavy - quark mass limit . using this scheme we predict the heavy - quark spin and flavor symmetry counterparts of the @xmath1 .
|
experimental studies of neutrino - nucleus interactions carried out over the past decade @xcite have provided ample evidence of the inadequacy of the relativistic fermi gas model ( rfgm ) , routinely employed in event generators , to account for both the complexity of nuclear dynamics and the variety of reaction mechanisms other than single nucleon knock out contributing to the observed cross section .
a striking manifestation of the above problem is the large discrepancy between the predictions of monte carlo simulations and the double differential charged current ( cc ) quasi elastic ( qe ) cross section measured by the miniboone collaboration using a carbon target @xcite . as pointed out by the authors of ref .
@xcite , improving the treatment of nuclear effects , which turns out to be one of the main sources of systematic uncertainty in the oscillation analysis @xcite , will require the development of a _ comprehensive _ and _ consistent _ description of neutrino - nucleus interactions , _ validated _ through extensive comparison to the large body of electron - nucleus scattering data @xcite .
the main difficulty involved in the generalisation of the approaches successfully employed to study electron scattering to the case of neutrino interactions stems from the fact that , while the energy of the electron beam is fixed , in neutrino scattering the measured cross section results from the average over different beam energies , broadly distributed according to a flux @xmath0 .
therefore , a measurement of the energy of the outgoing charged lepton in a cc qe interaction _ does not _ specify the energy transfer to the nuclear target , which largely determines the reaction mechanism . as shown in refs .
@xcite , the miniboone double differential cross section corresponding to a specific muon energy bin turns out to receive comparable contributions from different mechanisms , which must be all taken into account in a consistent fashion .
many authors have suggested that the excess cc qe cross section observed by the miniboone collaboration is to be ascribed to the occurrence of events with two particle - two hole final states , not taken into account by the rfgm employed for data analysis @xcite .
the description of these processes within a realistic model of nuclear dynamics requires that all mechanisms leading to their occurrence
initial state correlations ( isc ) among nucleons in the target nucleus , final state correlations ( fsc ) between the struck nucleon and the spectator particles , and interactions involving two - nucleon meson - exchange currents ( mec)be included . within the independent particle model ( ipm ) of the nucleus
, however , correlations are not taken into account , and two particle - two hole final states can only be excited through the action of two - body operators , such as those involved in the definition of mec . in this paper , we analyse the mechanisms leading to the appearance of two particle - two hole final states in the response of interacting many - body systems , and argue that the interference between amplitudes involving one- and two - nucleon currents plays an important role .
this feature clearly emerges from the results of a calculation of the transverse electromagnetic response of @xmath1he and of the corresponding sum rule of @xmath2c , computed using state - of - the - art models of the nuclear hamiltonian and currents , within the green s function monte carlo ( gfmc ) computational scheme @xcite . in view of the extension of our study to the kinematical regime
in which the non relativistic approximation is no longer applicable , we also outline a novel approach , based on a generalisation of the factorisation _ ansatz _ , underlying the spectral function formalism .
this scheme , allowing for a consistent treatment of one- and two - nucleon current contributions , appears to be quite promising for applications to neutrino scattering .
the structure of the nuclear cross section , as well as its expression in terms of longitudinal and transverse structure functions , are reviewed in section [ response ] , while section [ nmbt ] describes the theoretical approaches , based on nuclear many - body theory , developed to study electron - nucleus scattering . in section [ nonrel ]
we discuss the non relativistic regime and the results of gfmc calculations , while in sections [ rel ] and [ rel2 ] we derive the explicit expression of the two particle - two hole contribution to the cross section obtained from our approach , and discuss the preliminary results of its application .
finally , in section [ conclusions ] we summarise our findings and state the conclusions .
in the one - photon - exchange approximation , the double differential electron - nucleus cross section can be written in the form [ xsec ] = l_w_a^ , where @xmath3 and @xmath4 are the four - momenta of the incoming and outgoing electrons , respectively , @xmath5 is the fine structure constant , @xmath6 is the differential solid angle in the direction specified by @xmath7 , and @xmath8 is the four momentum transfer .
the lepton tensor @xmath9 is completely determined by lepton kinematics , while the nuclear response is described by the tensor @xmath10 , defined as @xmath11 where @xmath12 and @xmath13 denote the initial and final hadronic states , the four - momenta of which are @xmath14 and @xmath15 .
the nuclear current can be written as a sum of one- and two - nucleon contributions , according to ( see , e.g. , ref .
@xcite ) [ nuclear : current ] j_a^= _
i j^_i+_j > i j^_ij .
the current @xmath16 describes interactions involving a single nucleon . in the qe sector , it can be expressed in terms of the measured vector form factors @xcite .
the two - nucleon contribution @xmath17 , on the other hand , accounts for processes in which the beam particle couples to the currents arising from meson exchange between two interacting nucleons .
equation can be rewritten in terms of two response functions , denoted @xmath18 and @xmath19 , describing interactions with longitudinally ( l ) and transversely ( t ) polarised photons , respectively .
the resulting expression reads @xmath20 \ , \end{aligned}\ ] ] where @xmath21 and @xmath22 ^ 2 $ ] is the mott cross section .
the @xmath23 and @xmath24 structure functions can be readily expressed in terms of the components of the response tensor of eq . .
choosing the @xmath25-axis along the direction of the momentum transfer one finds @xmath26 note that the above expressions are completely general , and describe processes involving both one- and two - nucleon current operators .
+ it follows from eqs . and that the nuclear cross section and response functions can be written as a sum of contributions corresponding to different hadronic final states @xmath27 .
consider , for example , the case of qe scattering , in which the final state particles are nucleons only .
for a carbon target we find latexmath:[\ ] ] extensive numerical calculations of the electron - carbon cross section based on the formalism described in this paper are under way .
however , they involve a number of non trivial novel developments , e.g. the derivation of the two - hole contributions to the nuclear spectral function , the discussion of which will require a separate publication .
figure [ rt_12c ] shows the transverse electromagnetic response of carbon at @xmath125 mev computed using the carbon spectral function of ref .
@xcite and approximating the two - hole spectral function of carbon with that of uniform nuclear matter , at density corresponding to fermi momentum @xmath126 mev .
note that this is _ not _ the same as working within the fermi gas model .
we use overlaps the functions @xmath120 defined by eqs . and obtained within the _ ab initio _ approach of refs .
@xcite , based on a realistic nuclear hamiltonian including two- and three- nucleon interactions . owing to short range correlations , which move strength from the 1p1h to the 2p2h sector , the resulting occupation of the momentum eigenstates is reduced by @xmath127 20% .
interactions effects also affect the initial state energies of the knocked - out nucleons @xcite , thus shifting the threshold of the two - nucleon current contributions with respect to the predictions of the frmi gas model @xcite .
it has to be pointed out that the correlation contribution to the carbon spectral function of ref .
@xcite is obtained from nuclear matter results .
therefore , the use of nuclear matter overlaps in the matrix elements of the two - nucleon current entering the interference terms appears to be consistent .
we have used the fully relativistic expression of the two - nucleon current described in refs .
@xcite , with the same form factors and @xmath128-width . the solid line of fig
. [ rt_12c ] represents the results of the full calculation , whereas the dashed line has been obtained including only the amplitudes involving the one - body current .
the contributions arising from the two - nucleon current are illustrated by the dash - dot and dotted lines , corresponding to the pure two - body current transition probability and the interference term , respectively .
the latter turns out to be sizable , its contribution being comparable to the total two - body current response for @xmath129 mev . although our results still need to be improved , and do not include the corrections taking into account the effects of fsi , in fig .
[ rt_12c ] we have also included , for comparison , the data resulting from the analysis of ref .
@xcite .
the solid line represents the results of the full calculation , whereas the dashed line has been obtained including only amplitudes involving the one - body current .
the contributions arising from the two - nucleon current are illustrated by the dot - dash and dotted lines , corresponding to the pure two - body current transition probability and the interference term , respectively .
the experimental data are taken from ref .
we have analysed the mechanisms correlations in the initial and final states and coupling to meson exchange currents leading to the excitation of 2p2h final states in the nuclear response to electromagnetic interactions . in the non relativistic regime ,
in which highly accurate calculations consistently taking into account all these mechanisms are feasible , our results confirm the findings of ref .
@xcite . in the transverse channel ,
the contribution of processes involving the two - nucleon current is sizeable , and extends well into the kinematical region corresponding to energy transfer @xmath130 , in which single nucleon knock out is dominant .
the important role played by interference between the amplitudes involving one- and two- body currents clearly implies that correlation effects must be included in any model aimed at describing the nuclear cross section in the 2p2h sector .
this point was clearly stated , over three decades ago , in the pioneering work of ref .
@xcite , the authors of which also remarked the inadequacy of the treatment of correlations based on lowest - order perturbative pion exchange .
however , combining a realistic and consistent description of correlations and meson - exchange currents in the kinematical region in which the non relativistic approximation is no longer applicable involves serious difficulties . to overcome this problem ,
we have developed a novel approach based on the factorisation _
ansatz _ underlying the spectral function formalism , widely and successfully employed to describe the nuclear response in the 1p1h sector . the preliminary results obtained within this approach , shown in fig .
[ rt_12c ] , provide a fairly good description of the measured transverse response of carbon at @xmath125 mev . a comparison between the results of fig . [ rt_12c ] and the gfmc results of fig .
[ 4he ] shows distinctive discrepancies in both magnitude and energy dependence of the two - body current contributions . while part of the disagreement is likely to originate from differences in the two - nucleon currents employed in ref .
@xcite , as well as from the non relativistic nature of the gfmc calculations , the large interference contribution in the region of the quasi elastic peak observed in fig .
[ 4he ] may arise from interference between amplitudes involving the one- and two - body currents and 1p1h final states . a careful analysis of these terms , that were found to be sizable in the pioneering work of ref .
@xcite , is being carried out , and will be discussed elsewhere .
the main assumption implied in the factorisation of the 2p2h final states is the treatment of the knocked out nucleons as free particles , which amounts to neglecting their interactions , both among themselves and with the spectator nucleons .
antisymmetrization under exchange between any of the outgoing particles and the spectators is also disregarded .
the factorized nuclear transition amplitudes involving the one - nucleon current can be corrected to include the effects of final state interactions in the quasi elastic sector using an extension of the spectral function formalism , as discussed in ref .
the resulting modifications lead to ( i ) a shift in energy transfer of the differential cross section , arising from interactions between the knocked out nucleon and the mean field of the recoiling nucleus , and ( ii ) a redistribution of the strength from the quasi free peak to the tails , resulting from rescattering processes .
theoretical studies of electron - nucleus scattering suggest that in the kinematical region relevant to the miniboone analysis the former mechanism which does not involve the appearance of 2p2h final states provides the dominant contribution , and can be taken into account through an optical potential @xcite .
the corrections to the factorized amplitudes involving the two - nucleon current also include interactions between the two knocked out particles . a careful analysis of these processes is certainly needed .
however , the results of shen et al , who carried out an accurate calculation of the neutrino - deuteron cross section over a broad kinematical range , suggests that their effect becomes negligibly small at beam energies larger than @xmath131 mev @xcite . in conclusion , we believe that the approach described in this paper provide a viable and promising scheme for the development of a unified treatment of processes involving one- and two - nucleon currents , applicable in the kinematical region relevant to neutrino oscillation searches .
therefore , it may in fact be regarded as a step towards the _ new paradigm _ advocated by the authors of refs .
@xcite .
we are deeply indebted to a.m. ankowski , j. carlson , s. gandolfi , c. mariani , s. pieper and r. schiavilla for many illuminating discussions .
the work of ob and nr was supported by infn under grant manybody .
the work of al was supported by the u.s .
department of energy , office of science , office of nuclear physics , under contracts de - ac02- 06ch11357 . under an award of computer time provided by the incite program ,
this research used resources of the argonne leadership computing facility at argonne national laboratory , which is supported by the office of science of the u.s .
department of energy under contract de - ac02 - 06ch11357 .
|
the excitation of two particle - two hole final states in neutrino - nucleus scattering has been advocated by many authors as the source of the excess cross section observed by the miniboone collaboration in the quasi elastic sector . we analyse the mechanisms leading to the appearance of these final states , and illustrate their significance through the results of accurate calculations of the nuclear electromagnetic response in the transverse channel .
a novel approach , allowing for a consistent treatment of the amplitudes involving one- and two - nucleon currents in the kinematical region in which the non relativistic approximation breaks down is outlined , and its preliminary results are reported .
|
the internet and the networks it facilitates including the web and email networks are the largest artificial complex networks in existence , and understanding their structural and dynamic properties is important if we wish to understand social and technological networks in general .
moreover , efforts to design novel dynamic protocols for communication and fault tolerance are well served by knowing these properties .
one structural property of particular interest is the degree distribution at the router level of the internet .
this distribution has been inferred @xcite both by sampling _ traceroutes _ , i.e. , the paths chosen by internet routers , which approximate shortest paths in the network , and by taking `` snapshots '' of bgp ( border gateway protocol ) routing tables @xcite .
these methods have been criticized as being noisy and imperfect @xcite . however , lakhina et al .
@xcite recently argued that such methods have a more fundamental flaw . due to the fact that these methods use a small number of sources for the inference
, they were able to show that only a small fraction of the edges of a graph are `` visible '' in such a sample .
moreover , the set of visible edges is biased in such a way that erds - rnyi random graphs @xcite , whose underlying distribution is poisson , will appear to have a power law degree distribution . while no one would argue that the internet is a purely random graph , this certainly calls into question the standard measurements of power law or `` scale - free '' degree distributions on the internet , and reopens the problem of characterizing internet topology . in this paper
we explain this bias phenomenon analytically . specifically , by modeling the growth of a breadth - first spanning tree with differential equations , we show that sampling shortest paths from a single source in an erds - rnyi random graph gives rise to a power law degree distribution of the form @xmath2 , up to a cutoff @xmath4 where @xmath1 is the average degree of the underlying graph .
while sampling traceroutes from a single source is rather limited , barford @xcite provides empirical evidence that , on the internet , merging shortest paths from several sources leads to only marginally improved surveys of internet topology . in the conclusions we discuss generalizing our approach to sampling from multiple sources .
finally , even if the internet has a power - law degree distribution , the exponent may be rather different from the one observed in traceroute samples . in future work
we plan to extend our approach to graphs with arbitrary degree distributions , to study the relationship between the observed exponent and the underlying one .
most mapping projects have inferred the internet s degree distribution by implicitly building a map of the network from the union of a large number of traceroutes from a single source , or from a small number of sources . assuming that traceroutes are shortest paths , sampling from a single source is equivalent to building a spanning tree .
we therefore model this sampling method by modeling the growth of a spanning tree on a graph .
there are several ways one might build a spanning tree .
we will consider a family of methods , in which at each step , every in the graph is labeled _ reached _ , _ pending _ , or .
the pending vertices are the leaves of the current tree , the reached vertices are those in its interior , and the unknown vertices are those not yet connected . to model traceroutes from a single source , we initialize the process by labeling the source vertex pending , and all other vertices unknown .
then the growth of the spanning tree is given by the following pseudocode .
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ while there are pending vertices : + choose a pending vertex @xmath5 + label @xmath5 reached + for very unknown neighbor @xmath6 of @xmath5 , + label @xmath6 pending . _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ the type of spanning tree is determined by how we choose which pending vertex @xmath5 we will use to extend the tree . to model shortest paths , we store the pending vertices in a queue , take them in fifo ( first - in , first - out ) order , and build a breadth - first tree ; if we like we can break ties randomly between vertices of the same age in the queue , which is equivalent to adding a small noise term to the length of each edge as in @xcite .
if we store the pending vertices on a stack and take them in lifo ( last - in , first - out ) order , we build a depth - first tree .
finally , we can choose from among the pending vertices uniformly at random , giving a `` random - first '' tree .
surprisingly , while these three processes build different trees , and traverse them in different orders , we will see in the next section that they all yield the same degree distribution when @xmath7 is large . to illustrate this , fig .
[ fig : graph ] shows empirical degree distributions of breadth - first , depth - first , and random - first spanning trees for a random graph @xmath0 where @xmath8 and @xmath9 .
the three degree distributions are indistinguishable ; further , they are well - matched by the analytic results given in the next section , and obey a power law with exponent @xmath10 for degrees less than @xmath1 . for comparison , we also show the poisson degree distribution of the underlying graph .
[ htbp ] and average degree @xmath9 , and our analytic results ( the black dots ) .
the agreement between the three types of trees and our analytic results is extremely good . for comparison ,
the black line shows the poisson degree distribution of the underlying graph .
note the power law behavior of the apparent degree distribution for @xmath11 , which extends up to a cutoff at @xmath4 . ]
in this section we show analytically that building spanning trees in erds - rnyi random graphs @xmath0 , using any of the processes described above , gives rise to an apparent power law degree distribution @xmath2 for @xmath3 .
we focus here on the case where the average degree @xmath1 is large , but constant with respect to @xmath7 ; we believe our results also hold if @xmath1 is a moderately growing function of @xmath7 , such as @xmath12 or @xmath13 for small @xmath14 , but it seems more difficult to make our analysis rigorous in that case
. we will model the progress of the while loop described above .
let @xmath15 and @xmath16 denote the number of pending and unknown vertices at step @xmath17 respectively .
the expected changes in these variables at each step are ( where @xmath18}$ ] denotes the expectation ) @xmath19 } & = & - p u(t ) \nonumber \\ { { \rm e}[s(t+1)-s(t ) ] } & = & p u(t ) - 1 \label{eq : diff}\end{aligned}\ ] ] here the @xmath20 terms come from the fact that a given unknown vertex @xmath6 is connected to the chosen pending vertex @xmath5 with probability @xmath21 , in which case we change its label from unknown to pending ; the @xmath10 term comes from the fact that we also change @xmath5 s label from pending to reached .
moreover , these equations apply no matter how we choose @xmath5 ; whether @xmath5 is the `` oldest '' vertex ( breadth - first ) , the `` youngest '' one ( depth - first ) , or a random one ( random - first ) . by the principle of deferred decisions , the events that @xmath5 is connected to each unknown vertex @xmath6 are independent and
occur with probability @xmath21 .
our experiments do indeed show that these three processes result in the same degree distribution .
writing @xmath22 , @xmath23 and @xmath24 , the difference equations become the following system of differential equations , @xmath25 with the initial conditions @xmath26 and @xmath27 , the solution to is @xmath28 the algorithm ends at the smallest positive root @xmath29 of @xmath30 ; using lambert s function @xmath31 , defined as @xmath32 where @xmath33 , we can write @xmath34 note that @xmath29 is the fraction of vertices which are reached at the end of the process , and this is simply the size of the giant component of @xmath35 . now , we wish to calculate the degree distribution @xmath36 of this tree .
the degree of each vertex @xmath5 is the number of its previously unknown neighbors , plus one for the edge by which it became attached ( except for the root ) . now ,
if @xmath5 is chosen at time @xmath37 , in the limit @xmath38 the probability it has @xmath11 unknown neighbors is given by the poisson distribution with mean @xmath39 , @xmath40 averaging over all the vertices in the tree gives @xmath41 it is helpful to change the variable of integration to @xmath42 . since @xmath43 we have @xmath44 , and @xmath45 here in the second line we use the fact that @xmath46 when @xmath1 is large ( i.e. , the giant component encompasses almost all of the graph ) .
the integral in is given by the difference between two incomplete gamma functions .
however , since the integrand is peaked at @xmath47 and falls off exponentially for larger @xmath42 , for @xmath3 it coincides almost exactly with the full gamma function @xmath48 .
specifically , for any @xmath49 we have @xmath50 and , if @xmath51 for @xmath52 , then @xmath53 this is @xmath54 if @xmath55 , i.e. , if @xmath56 for some @xmath57 .
in that case we have @xmath58 giving a power law of exponent @xmath10 up to @xmath4 .
although we omit some technical details , this derivation can be made mathematically rigorous using results of wormald @xcite , who showed that under fairly generic conditions , the state of discrete stochastic processes like this one is well - modeled by the corresponding rescaled differential equations .
specifically it can be shown that if we condition on the initial source vertex being in the giant component , then with high probability , for all @xmath37 such that @xmath59 , @xmath60 and @xmath61 .
it follows that with high probability our calculations give the correct degree distribution of the spanning tree within @xmath62 .
lakhina et al .
@xcite , argued that sampling traceroutes from a small number of sources profoundly underestimates the degrees of vertices far from the source , and that this effect can cause graphs to appear to have a power law degree distribution even when their underlying distribution is poisson . in this work
, we modeled this sampling process as the construction of a spanning tree , and showed analytically that for sparse random graphs of large average degree , the apparent degree distribution does indeed obey a power law of the form @xmath63 for @xmath11 below the average degree .
this illustrates the danger of concluding the existence of a power - law from data over too small a range of degrees , and , more specifically , the danger of sampling traceroutes from just a few sources .
while our analytic results hold for traceroutes from a single source , we conjecture that if we use any constant number of sources and take the union of the resultant spanning trees , the observed distribution will remain subject to the same sampling bias , and one will again observe a power - law degree distribution .
certainly this exponent @xmath10 is not the one observed for the internet .
for instance , faloutsos , faloutsos and faloutsos @xcite observed degree distributions @xmath64 at the router level , and @xmath65 for out - degrees at the inter - domain ( bgp ) level . while the real degree distribution of the internet may be a power law
, one possibility is that the true exponent is rather different from the observed one ; this possibility was recently explored by peterman and de los rios @xcite ( they also study another mechanism for observing apparent power laws in the degree distribution , which samples the original graph via a probabilistic pruning strategy and gives an observed exponent of @xmath66 for random graphs ) . in future work
we will extend our differential equation model to random graphs with power law degree distributions to explore this possibility .
although our analytic model of spanning trees illustrates one possible mechanism for apparent power - law degree distributions , another possibility is that decisions ( such as those made by the border - gateway protocol ) made by actual internet routers interact in complex ways with the the underlying link - level topology .
it may well be that routers only use a small fraction of the edges in the network , and that the edges they actually use give rise to an _ effective _ degree distribution with roughly a power - law form .
this possibility implies that even if the real degree distribution of the internet is something very different from a power law , it may not matter as these extra links are not utilized in normal routing decisions .
another difference of unknown , but perhaps small , significance is that on the internet , routing algorithms are intended to optimize for end - to - end traffic flow , and routers in the `` interior '' of the internet such as those on the backbone are not themselves destinations or sources . unlike the empirical studies cited earlier , we selected sources uniformly at random for our spanning trees .
the story of the topology of the internet is far from over and will undoubtedly remain a topic of great interest for many years to come .
however , knowing decisively that current methods are fundamentally biased will serve to push the state - of - the - art forward toward more robust methods of characterization .
we see the following three lines of inquiry as enticing : first , having analytically shown that a poisson degree distributions can lead to apparent power laws , we naturally wish to generalize our approach to random graphs with arbitrary degree distributions , characterize the family of distributions which generate apparent power laws , and understand the relationship between observed and underlying exponents for power law distributions .
having this generalization may allow us to make firm and useful claims about the topology of the internet .
secondly , in what manner is the traceroute - sampled degree distribution of a random graph dependent on the length of the paths taken between a source and destination ? if we simulate a `` crawl '' by using random walks rather than short paths , how would the observed degree distribution change ?
finally , if we use build spanning trees from @xmath42 sources and take their unions , how does the observed degree distribution vary with @xmath42 ?
if we are correct that any constant @xmath42 leads to the same kind of bias , how does @xmath42 need to grow with the network to obtain an accurate sample ?
the authors are grateful to david kempe , mark newman , michel morvan , paolo de los rios , mark crovella and tracy conrad for helpful conversations , and to american airlines for providing a conducive working environment .
this work was funded by nsf grant phy-0200909 .
cooperative association for internet data analysis ( caida ) .
d. meyer , university of oregon route views project .
t. kernen , traceroute organization + http://www.traceroute.org b. lyon , opte project http://www.opte.org the border gateway protocol ( bgp ) is a distributed method by which routers which bridge two distinct networks construct forwarding tables based on the ip addresses which are `` reachable '' from the current router via `` advertised paths '' of other bgp - routers .
such routers decide which packets cross to and from the network it guards , and may also serve to forward packets between networks .
|
the topology of the internet has typically been measured by sampling _ traceroutes _ , which are roughly shortest paths from sources to destinations .
the resulting measurements have been used to infer that the internet s degree distribution is scale - free ; however , many of these measurements have relied on sampling traceroutes from a small number of sources .
it was recently argued that sampling in this way can introduce a fundamental bias in the degree distribution , for instance , causing random ( erds - rnyi ) graphs to appear to have power law degree distributions .
we explain this phenomenon analytically using differential equations to model the growth of a breadth - first tree in a random graph @xmath0 of average degree @xmath1 , and show that sampling from a single source gives an apparent power law degree distribution @xmath2 for @xmath3 .
|
this is an update to the _ chandra _ high energy transmission grating ( hetg ) calibration based on in - orbit observations .
the hetg was described by canizares et al .
( 2005 @xcite ) and previous flight calibration results were reported by marshall et al .
( 2004 @xcite ) .
there are two grating types on the hetg : the high energy gratings ( hegs ) have @xmath22 higher dispersion than the medium energy gratings ( megs ) .
when combined with the detectors , the system is called the hetg spectrometer ( hetgs ) .
a new set of hetg efficiencies was released as part of _ chandra _ caldb v4.4.7 , which was intended to reduce systematic differences between the heg and meg fluxes as well as improve overall model fits .
here is provided details of the steps that went into the release and what is in store for the next update .
updates will also be available at the hetgs calibration web site http://space.mit.edu/asc/calib/hetgcal.html and are presented regularly at meetings of the international astronomical consortium for high energy calibration ( iachec ) ( see http://web.mit.edu/iachec/meetings/index.html ) .
first , data were selected from the tgcat on - line catalog of hetgs spectra .
see http://tgcat.mit.edu to obtain the data used here , which are given in table [ tab : observations ] .
i started with sources of all sorts but relied most on extragalactic sources with relatively smooth continua .
a simple type is given : `` bll '' for the bl lac objects ( without strong optical emission lines ) , and `` agn '' for the remaining quasars and seyfert galaxies .
support for this work was provided by the national aeronautics and space administration ( nasa ) through the smithsonian astrophysical observatory ( sao ) contract sv3 - 73016 to mit for support of the chandra x - ray center ( cxc ) , which is operated by sao for and on behalf of nasa under contract nas8 - 03060 .
canizares , c.r . ,
et al . , 2005 , pasp , 117 , 1144 .
marshall , h. l. , dewey , d. , & ishibashi , k. 2004 , , 5165 , 457 marshall , h. l. , dewey , d. , schulz , n. s. , & flanagan , k. a. 1998 , , 3444 , 64 perlman , e. s. , et al .
2005 , , 625 , 727
|
the efficiencies of the gratings in the high energy transmission grating spectrometer ( hetgs ) were updated using in - flight observations of bright continuum sources .
the procedure first involved verifying that fluxes obtained from the @xmath0 and @xmath1 orders match , which checks that the contaminant model and the ccd quantum efficiencies agree .
then the fluxes derived using the high energy gratings ( hegs ) were compared to those derived from the medium energy gratings ( megs ) .
the flux ratio was fit to a low order polynomial , which was allocated to the megs above 1 kev or the hegs below 1 kev .
the resultant efficiencies were tested by examining fits to blazar spectra .
|
when an incident x - ray photon ejects an electron from a core - level in a metal , the conduction band electrons feel a local attractive potential due to the created hole .
it was discovered by anderson @xcite that the electronic ground states before and after the creation of the hole are orthogonal to each other .
this many - body effect has dramatic consequences in x - ray photoemission spectroscopy ( xps ) experiments where an asymmetric power - law divergence is observed.@xcite for a non - interacting metal , the exponent of the power - law and the relative intensity of the peaks in the xps spectra are well understood .
however , the behavior of the power - law divergence in a strongly interacting metal has received little theoretical attention besides the one - dimensional case .
@xcite recently , there have been several xps studies of strongly correlated transition - metal oxides,@xcite which addressed the changes in the core - level spectrum across the metal to mott insulator transition ( mit ) .
it was observed in particular that the ( strongly renormalized ) low - energy quasiparticles present in the metallic phase close to the mit strongly modify the satellite peaks measured in xps , an effect also discussed theoretically by kim _
@xcite in this paper , we provide a detailed theory of the core - level photoemission lineshape across the metal - insulator transition .
a power - law behavior is found throughout the metallic phase up to the mit where it is destroyed .
both the exponent and the intensity of the power - law are strongly renormalized by interactions and two different regimes can be identified depending on the ratio between the valence band interaction and the core - hole potential intensities .
the rest of the paper is organized as follows . in sec .
[ sec : model ] we describe the model that we use in our calculations .
the numerical results for the xps spectra across the mit transition are presented and discussed in sec .
[ sec : numesp ] .
section [ sec : molorb ] contains the results of a molecular orbital approximation .
finally , we state the conclusions of our study in sec .
[ sec : conclus ] .
the simplest model to study the correlation induced mit is the hubbard model @xmath7where @xmath8 creates an electron with spin @xmath9 at site @xmath10 and @xmath11 .
a key quantity for this problem is the spectral function @xmath12 $ ] that contains information about the valence - band single particle photoemission spectrum .
it can be obtained in the limit of large lattice coordination using dynamical mean - field theory ( dmft ) .
@xcite in the dmft framework the hubbard model reduces to an anderson impurity model coupled to a non - interacting electron bath that is calculated in a self - consistent way . in the xps experiment
a photon excites a core electron out of the sample and the resulting core - hole interacts with the band electrons attractively . in the sudden approximation considered below ,
the self - consistent bath remains unchanged while the energy of the effective anderson impurity is shifted .
the core - hole potential is taken as local and momentum independent @xcite @xmath13 where @xmath14(@xmath15 ) creates ( destroys ) a core - hole at site @xmath16 , and @xmath17 is the core - level energy . in the sudden approximation
the xps spectrum is given by the core - hole spectral density @xmath18 where the final states @xmath19 with corresponding energies @xmath20 satisfy the sum rule @xmath21 and @xmath22 is the initial ground state with energy @xmath23 .
the spectrum is zero for energies larger than the threshold energy @xmath24 where @xmath25 is the ground state energy in the final configuration . in what follows
we will consider a bethe lattice geometry where the free density of states is given by @xcite @xmath26 we will set the half - bandwidth @xmath27 as the energy unit , and focus on half - filling ( @xmath28 ) .
in the non interacting case ( @xmath29 ) the spectral function @xmath30 is simply is given by eq .
( [ eq : freedens ] ) and the xps spectrum then presents a doniach - unji power - law behavior at the threshold @xcite @xmath31 here @xmath32 , @xmath33 , where @xmath34 is the phase - shift at the fermi level due to scattering with the core - hole , and the factor @xmath35 is due to spin degeneracy . for @xmath36
a bound state appears in the final configuration that is reflected in the xps spectra as an additional peak at @xmath37 .
these analytical results have been nicely reproduced numerically using wilson s numerical renormalization group@xcite ( nrg ) by lbero _
et al._@xcite here we will use a different approach within the nrg scheme . to improve the accuracy at high energies we use a density matrix
formulation.@xcite we have @xmath38 where @xmath39 is the density matrix .
to evaluate the matrix elements @xmath40 and @xmath41 two simultaneous runs of nrg are performed ( with and without the core - hole energy shift ) .
the @xmath40 are calculated on the first iteration and then transformed to the new basis at each nrg iteration .
the spectra are constructed following ref . .
the high numerical resolution of the nrg at low energies allows us to recover the power - law behavior with parameters that reproduce the analytical results ( see fig . [
fig : nonint ] ) .
a bound state emerges in the final state for @xmath36 and the associated peak in the xps spectra can be observed at @xmath37 . .
inset : numerically obtained @xmath4 ( dots ) and analytical expression @xmath42 ( solid line ) .
, width=321 ] for a non - zero hubbard interaction , the spectral function @xmath30 needs to be calculated in a self - consistent way . in what follows we will use the density matrix numerical renormalization group ( nrg ) scheme @xcite to solve the dmft equations @xcite and to calculate the xps spectra @xcite for different values of @xmath1 .
the ratio @xmath43 can be varied in some materials across the mit using chemical or mechanical pressure .
the ratio @xmath44 , however , is expected to remain approximately constant for a given material . in ref .
the spectra in ru 3d systems were fitted using @xmath45 .
the spectrum is gapped while for @xmath46 there is a narrow peak at the fermi level .
b ) core - hole spectra for a core - hole potential @xmath47 across the mit showing a power - law behavior close to the threshold in the metallic phase .
c ) same as b ) with @xmath48 .
d ) and e ) detail of the high energy peaks of b ) and c ) , respectively.,width=321 ] figure [ fig : fig1]a shows numerical results for the valence band spectral function .
a small interaction in the band ( @xmath49 ) produces only a slight modification of @xmath30 that close to the fermi level is approximately given by eq .
( [ eq : freedens ] ) . for intermediate values of the interaction ( @xmath50 ) the system is in the strongly correlated regime and the valence band has a three three peak structure .
the central _ quasiparticle _ peak has a width @xmath51 , where @xmath52/\partial\omega|_{\omega=0}$ ] and @xmath53 is the self - energy .
the incoherent upper and lower hubbard bands have a width @xmath54 and their positions are determined by the atomic energies .
the fermi liquid behavior is restricted to a narrow region close to the fermi level inside the quasiparticle peak which narrows with increasing interaction as @xmath55 . for values of @xmath1 larger than the critical interaction @xmath56
the system is in the mott insulator phase , there is no quasiparticle peak and the charge excitations are gapped . within the dmft framework
the interactions in the band produce a twofold modification of the non - interacting x - ray problem .
the atomic level coupled to the core - hole potential has now a hubbard interaction @xmath1 and is hybridized with a renormalized electronic bath .
to lowest order in @xmath1 we may neglect the changes in the valence band and perform a hartree - fock approximation to treat the local interaction .
we are left with a non - interacting x - ray problem with a renormalized core - hole potential @xmath57 , where @xmath58 is the charge on the impurity in the presence of the core - hole potential . in general
we have @xmath59 and we therefore expect the bound state peak ( if present ) to be shifted toward the threshold ( @xmath60 ) and the power - law to become more divergent as @xmath4 is reduced to @xmath61 . in fig .
[ fig : fig1]b - e we show xps spectra for different values of the interaction @xmath1 and the core - hole potential @xmath2 .
we obtain a power - law behavior ( straight lines ) throughout the metallic phase that disappears at the mit . for small @xmath62 the xps spectrum
can be understood within the hartree - fock scheme .
this approximation , however , breaks down rapidly with increasing @xmath1 as the system approaches the strongly correlated regime .
we may argue in this regime that the width of the quasiparticle peak @xmath63 acts as an effective bandwidth for the x - ray problem . while this allows us to understand the decrease of the onset energy for the power - law behavior ( see figs .
[ fig : fig1]b and [ fig : fig1]c ) it does not give the complete picture as we will see below .
close to the mit there are some important differences in the spectra for the two values of @xmath2 shown in fig .
[ fig : fig1 ] . while the power - laws in fig .
[ fig : fig1]b for @xmath64 have a significant increase in the slope when approaching the mit ( we have @xmath65 for @xmath5 ) , those for @xmath48 in fig . [
fig : fig1]c have nearly parallel displacements with only a very small increase in the slope throughout the metallic phase . for @xmath47
the high energy peaks shift toward the edge with increasing @xmath1 and disappear at the mit leaving a delta - function peak at the threshold in the insulator .
for @xmath66 the main difference in the xps spectra just below ( @xmath46 ) and above ( @xmath67 ) the mit is the presence of the power - law in the metallic phase .
the high energy peaks are essentially unchanged and the integrated intensity of the power - law vanishes at the mit with @xmath68 .
some insight on the origin of these two distinct behaviors can be obtained by analyzing the atomic limit in the insulator . for values of @xmath2 such that @xmath69
, we have @xmath70 throughout the insulating phase , which means that the core - hole potential is not strong enough ( in the electron - hole symmetric case ) to overcome the local repulsion and change the charge locally in the atomic level .
the initial and final ground states will be identical and we therefore expect a peak at the threshold . in the opposite situation , @xmath71 , there will be a region close to the mit in which the initial and final states have a different occupation of the atomic level .
for these values of @xmath2 , the initial and final ground states are orthogonal and no peak is expected at the threshold . as a function of @xmath1 , for different values of @xmath72 and @xmath73 ( from top to bottom ) .
results from nrg ( symbols ) , hartree - fock results ( solid lines ) , and parabolic fits @xmath74 with @xmath75 ( dashed lines ) .
right panel : integrated intensity of the power - law peak for @xmath76 and @xmath73 ( from bottom to top ) .
the figure illustrates that the peak at threshold disappears at the mit in two different manners , for @xmath6 and @xmath3 ( see text ) .
, width=321 ] the results for the exponent and the integrated intensity of the power - law are summarized in fig .
[ fig : p - shiftu ] .
the exponent @xmath4 increases with @xmath2 and decreases with increasing interaction . for small @xmath77 the main effect of the interaction
is a compensation ( screening ) of the core - hole potential that can be explained within a hartree - fock approach ( solid lines in fig .
[ fig : p - shiftu ] ) . for large @xmath78
, @xmath4 is close to its maximum value @xmath79 and the interaction produces only a small reduction throughout the metallic phase .
the intensity of the peak at the threshold , however , vanishes continuously at the mit . for small @xmath69 the opposite behavior is observed with @xmath4 vanishing at the transition while the integrated intensity of the peak remains finite .
the behavior of @xmath4 can be understood through a fermi liquid analysis .
the friedel sum rule gives an exact zero - temperature relation between the scattering phase - shift and the _ total _ charge @xmath80 ( @xmath81-orbital plus conduction band ) displaced when the core - hole potential is turned on , @xcite namely : @xmath82 .
in the insulating phase , the valence band has a gap and the core - hole potential @xmath2 will not produce a displacement of charge ( @xmath83 ) unless @xmath2 is large enough to overcame the local repulsion ( i.e. @xmath84 ) .
this reflects the incompressibility of the insulator .
in contrast , in the metallic phase , charge will gather around the hole to screen it even for @xmath85 .
the amount of displaced charge reflects the vanishing compressibility of the metal as the mit is approached , so that we expect at small @xmath2 : @xmath86 . since fermi - liquid analysis suggests that the nozires and de dominicis value for the exponent [ @xmath87 is valid throughout the metallic phase even in the presence of interactions , this leads to a parabolic behavior @xmath88 , [ where we used @xmath89 .
this expression fits our data remarkably well , close to @xmath56 and for @xmath69 ( see fig . [
fig : p - shiftu ] ) . ) for fixed @xmath90 and three values of @xmath91 . left inset : in a logarithmic scale three peaks can be identified .
right inset : hubbard bands in the valence spectra.,width=321 ]
the intensity and position of the peaks in the insulating phase can be understood in the molecular orbital approximation .
the effective bath in dmft is modeled with a single site having a hubbard interaction @xmath1 and energy @xmath91 .
the resulting two site hubbard model can be solved analytically . in the absence of the core - hole potential ,
the ground state is in a subspace with 2 electrons .
when the core - hole potential is small ( @xmath92 ) the ground state is in the same charge sector and is essentially unchanged .
therefore , the xps spectrum has a delta - function peak at the threshold which carries almost all the spectral weight ( see fig . [
fig : intensities ] ) . for large @xmath93
the final ground state is in a subspace with three electrons and is therefore orthogonal to the initial ground state ( @xmath94=0 ) . in this regime
three peaks can be identified in the xps spectra and there is no peak at the threshold .
the main peak has a weight @xmath95 and there are two satellites at a distance @xmath96 ( shake - up ) and @xmath97 ( shake - down ) .
the shake - down peak is at a distance @xmath91 from the threshold and has a larger intensity ( @xmath98 ) than the shake - up peak ( @xmath99 ) . , @xmath100 , and @xmath101 .
, width=321 ] the molecular results are in excellent agreement with the numerical results in the insulating phase ( see fig . [
fig : insulating ] ) . the shake - up ( -down ) peak
is associated to the creation of an electron ( hole ) in the upper ( lower ) hubbard band and has an intrinsic width of order @xmath102 . for a finite hole lifetime ( finite temperature )
the spectrum will have an additional lorentzian ( gaussian ) broadening .
we note that , although it is likely to be difficult in practice , a measurement of the _ three _ peaks provides in principle a direct estimate of @xmath1 .
we stress that a measurement of the shake - down peak alone is not enough for such an estimation . as shown in fig .
[ fig : insulating ] , a change in @xmath91 only produces a small redistribution of the spectral intensity and a global shift of the peaks relative to the threshold
. the molecular orbital results also give a good estimation for the position of the peaks away from the threshold in the metallic phase ( see fig . [
fig : fig1]e ) .
in summary , we have studied theoretically the behavior of the core - level photoemission spectra across the correlation - driven mit .
away from the photoemission threshold and far from the mit , both the position and relative intensity of the peaks are well described by a molecular orbital approach .
close to the threshold or to the mit , more sophisticated techniques ( such as dmft and nrg ) are necessary to describe the spectra .
the changes in the xps spectra across the mit may be used to detect the transition . a symmetric peak ( or no peak ) at the edge implies an insulating phase , while an asymmetric peak or a power - law corresponds to a metal . for large @xmath84
there are three peaks in the xps spectra . while the shake - up peak is usually weak and
will probably be difficult to detect , its observation in conjunction with the shake - down peak allows in principle for a direct estimation of @xmath1 .
the most interesting results concern the behavior of the xps spectra at the threshold as summarized in fig .
[ fig : pdiag ] . in the metallic phase
there is a power - law behavior with an exponent @xmath103 , where @xmath4 depends on both the local interactions on the band @xmath1 and the intensity of the core - hole potential @xmath2 .
the power - law regime is restricted to an energy range @xmath51 close to the threshold and disappears at the mit .
for @xmath3 the exponent @xmath4 depends strongly on @xmath1 and vanishes as @xmath104 at the transition .
the integrated intensity , however , remains finite giving rise to a delta - function peak in the insulator .
in contrast , for @xmath6 , @xmath4 is finite at the transition but the integrated intensity of the peak vanishes and there is no delta - function peak in the insulator .
future high - resolution experiments might be able to test our theoretical predictions for the edge - singularity in a correlated metal close to the mit .
|
the zero temperature core - level photoemission spectrum of a hubbard system is studied across the metal to mott insulator transition using dynamical mean - field theory and wilson s numerical renormalization group .
an asymmetric power - law divergence is obtained in the metallic phase with an exponent @xmath0 which depends on the strength of both the hubbard interaction @xmath1 and the core - hole potential @xmath2 . for @xmath3 ,
@xmath4 decreases with increasing @xmath1 and vanishes at the transition ( @xmath5 ) leading to a symmetric peak in the insulating phase . for @xmath6 ,
@xmath4 remains finite close to the transition , but the integrated intensity of the power - law vanishes and there is no associated peak in the insulator .
the weight and position of the remaining peaks in the spectra can be understood within a molecular orbital approach .
|
planets can gravitationally perturb debris discs by various dynamical processes , such as secular interactions , where an eccentric or inclined planet can force the disc eccentricity or inclination @xcite , or resonance interactions , where the planet traps dust at a specific location , resulting in the creation of dust clumps in the disc @xcite .
these processes inducing eccentricity , a disc position offset with respect to the star , or clumps into the disc can result in brightness asymmetries . due to the limitations in the detection techniques ,
most of the confirmed exoplanets are located within 10 au of their host star , and potential planets located beyond this limit ( beyond saturn in our solar system ) remain undetected .
however , even if those distant planets are too small to be detected with our current telescopes , they can still leave an observational signature by gravitationally perturbing the dust of their debris disc . therefore investigating the dynamical relationship between debris discs and exoplanets
can not only provide some insights on the origin of debris disc asymmetries , but also provides clues to the presence of hidden planets in the outer part of stellar systems , a region currently difficult to observe .
hd 115600 is a young f2 type star at a distance of 110 pc in the scorpius centaurus association of age @xmath315 myr @xcite .
although an ir excess was detected by @xcite using _ spitzer / mips _ , its debris disc was imaged for the first time in the h - band at @xmath4 @xmath5 m by @xcite using the _ gemini planet imager _
( _ gpi _ ) .
the observation revealed a very broad nearly edge - on disc with an observed width to mean radius ratio @xmath6/@xmath7 0.37 , which makes the hd 115600 debris disc one of the broadest ever observed see table [ table1 ] for the main parameters of hd 115600 .
@xcite used the radiative transfer code grater to model the disc emission .
they found that the disc , most likely primarily constituted of ice - icy / silicate grains , appears eccentric with @xmath8 and its projected centre offset by @xmath9 and @xmath10 compared to the stellar location leading to a total offset of @xmath11 or @xmath12 au .
they proposed that such disc eccentricity and offset could have been sculpted by a potential planetary companion .
the absence of a direct detection of such a companion by _ gpi _
places an upper limit of the mass of the potential companion of @xmath13 @xmath14 if it is ideally located on an orbit outside the coronograph mask .
the authors point out that their model , however , respectively overestimates ( underestimates ) the disc flux at the southeast ( southwest ) corner . using predictions from the gap opening model developed by @xcite , as well as the planet - stirring scenario established by @xcite , only very loose constraints on the potential planetary companion
could be derived and the authors concluded that either a superjovian planet orbiting at @xmath15au or a super - earth located at the very inner edge of the disc could sculpt the disc .
n - body simulations can be used to model the interaction between a debris disc and a planet . in these simulations , a disc , modeled by an ensemble of massless particles , orbits around a central body representing the star and feels the gravitational perturbation of another massive body . to test different planet configurations
, a suite of these simulations must be run , and the planet parameter space ( comprised of the planet mass and orbital elements ) is traditionally explored by using a grid of set values . in this approach , the entire parameter space must be explored to isolate the best fit , and the precision of the best fit parameters directly depends on the size of the grid , and therefore a large number of simulations are required to reach high precision .
given the low number of currently resolved debris discs , n - body simulations have been so far used to study individual objects , however with the increased number of expected discoveries from the latest and next generation of instruments such as _ alma _ and _ jwst _ , a more efficient and statistical approach must be found .
for example , the _ monte carlo markov chain _ ( mcmc ) approach explores closely located points of the parameter space along a chain , and by accepting parameter space point resulting in better fits and rejecting point resulting in worse fits , the chain quickly converges toward the best fit region .
this method has been used in astrophysics to estimate cosmological parameters with high precision @xcite , adjusting semi - analytic models of galaxy formation @xcite or fitting planetary orbital element from transit and radial velocity data @xcite . @xcite
present the first study integrating n - body simulations within an _ mcmc _ algorithm to estimate the pre - infall mass of the carina galaxy before it joins the milky way satellites group .
one focus of our study is to present a framework for testing if the _
algorithm can be used to more efficiently probe the parameter space of n - body simulations of a planet shaping a debris disc than a traditional grid method . in this paper
, we investigate the role of planets in determining the morphology of the hd 115600 debris disc .
we dynamically model the interaction between an exoplanet and the hd 115600 disc using our modified n - body integrator , and explore the parameter space of the planets orbit and mass using both the sampling method _ mcmc _ and a traditional grid method . to compare the results of our numerical simulations with observations of hd 115600 , we create synthetic images from the simulations at a similar wavelength and resolution as the _ gpi _ image .
thus this study has two primary aims to both test the _ mcmc _ algorithm for exploring the parameter space in n - body simulations and also provide information on a potentially hidden companion that planet hunters can use in the future searches .
we first present our general method in section 2 . in order to derive the grain composition and properties
required to turn our simulations into synthetic images , section 3 presents our spectral modeling of the hd 115600 disc using the radiative transfer code _ mcfost_. in section 4 , we introduce our numerical method for performing the dynamical simulations , before presenting our results in section 5 .
a summary of our findings and conclusions are given in section 6 .
.properties for hd 115600 from currie et al .
( 2015 ) . [ cols="^,^",options="header " , ] while increasing the number of iterations would help derive stronger confidence intervals , we decided to keep the number of iterations low for several reasons .
firstly , running dynamical simulations are computationally expensive ( a run of 720 simulations requires @xmath16 days with a eight - cores intel xeon e5 - 2660 cpu of 2.2 ghz ) and as a consequence , very few dynamical studies of debris discs in the literature use a set of more than a couple of hundred simulations . to further ensure that enough planetary models resulting in disc parameters comparable to @xcite were explored within our limited number of iterations , we compare the disc parameters achieved by planetary models along the _ mcmc _ chain 1 with the observational parameters derived by @xcite and their associated uncertainties in figure [ fig : c1 ] au to derive the uncertainty on the disc width , which led to an uncertainty of 0.05 on the observational @xmath17 .
after a hundred iterations , most planetary models resulted in disc parameters similar to the observational value within the observational uncertainty , and we therefore conclude that given the uncertainty associated with the observational parameters , our number of iterations is enough to probe the best - fit region with appropriate accuracy . finally , direct imaging is currently the only technique capable of detecting planets with semi - major axes @xmath18au .
checking the exoplanet database for the 8 confirmed exoplanets which have been detected via imaging and with semi - major axes @xmath19au , we found that , when available , the average of uncertainties on the planetary parameters are @xmath20 au and @xmath21 . our short _
runs are therefore sufficient to also provide better constraints than those derived from current observations . + using _ mcfost
_ , we create the spectral energy distribution and synthetic image at 1.6 @xmath5 m corresponding to the best fit model see figure [ fig:6 ] .
the sed of the best fit model has @xmath22 , which is larger than the @xmath23 of the best fit model using a parametric disc in section 3 to estimate the grain properties .
the high @xmath24 value comes from the fact that we are trying to reproduce the disc features and emission observed in the near - ir and therefore our model , which is using small grains as they are the main contributors to the emission at short wavelengths , fails to reproduce the disc emission at the mid - ir wavelength ( 10 - 25 @xmath5 m ) accurately . on the other hand
, the model reproduces the observed sed for @xmath25 m very well . the projected synthetic image at @xmath26 m produced a disc with a projected offset of @xmath27 au , and the deprojected disc has a peak brightness location of @xmath28au , a disc width ratio of @xmath29 , and an eccentricity of @xmath30 , which is a good match to the observed characteristics of table [ table1 ] within the uncertainties .
to further check our implementation of the _ mcmc _ sampling method , we conduct an exploration of the parameter space using a traditional grid sampling method .
we sample the parameter space as follows : we used 6 values of @xmath31 between 1040 au in steps of 5 au ; 10 values of @xmath32 between 0.05 and 0.35 in steps of 0.06 ; and 6 values of @xmath33 between 2 @xmath34 and 8 @xmath34 in steps of 1.2 .
those ranges corresponds to the values typically explored in the _ mcmc _ chains ( see table [ table3 ] ) .
a total of 360 models were conducted , and we aim here to compare the best fit model and the statistics of the result found when using a grid sampling method versus a _ mcmc _
method for which only the first 360 iterations minus the 100 for the burn - in period are used . + following a grid method , the best fit ( @xmath35 )
is found for the model with @xmath36au , @xmath37 and @xmath38 , while by using only the first 360 simulations of the _ mcmc _ chain 1 , a better fit ( @xmath39 ) is found for @xmath40au , @xmath41 and @xmath42 .
while the best fit values are in excellent agreement between the two methods , figure [ fig:7 ] shows a @xmath23 map of the grid and illustrates how the grid method can be wasteful in terms of computational effort .
more than 250 out of 360 simulations returned @xmath43 , highlighting that @xmath44 of the time was spent outside of the best fit region , while the _ mcmc _ chain converged quickly toward the best fit region with all simulations having @xmath45 after the first 100 iterations .
as a result , only poor statistics can be derived by following a grid method : figure [ fig:8 ] presents the marginalized probability distribution of each parameter .
while the planet mass , eccentricity and semi - major axis distribution peak corresponds to the best fit model , the @xmath46 interval confidence for @xmath33 is @xmath47 and no interval can be derived for @xmath31 or @xmath32 .
if we look now at the probability distribution of each parameter derived using the first 260 iterations of the _ mcmc _ chain 1 ( the first 360 simulations excluding 100 iterations as a burn - in period ) presented as the red distribution in figure [ fig:8 ] , the distribution for all parameters peaks around the best fit value with @xmath48 , while no confidence interval can be derived for @xmath31 and @xmath32 .
although these _ mcmc _ probability distributions clearly provide less constraints on the interval confidence than the distribution derived using the full mcmc iterations sample ( figure [ fig:5 ] ) , it still presents slightly more constraints on the planetary mass confidence interval than the interval derived using the grid method .
au and @xmath41 , while the lines delimit the extended orbit crossing regions between the massive planet and a potential companion the mass of vesta when the asteroid is either apse aligned with the planet ( red ) or not ( blue dashed).,width=340,height=264 ] we now examine the dynamics at work that is shaping the disc .
the shaping of the debris disc by the planet is done via a mix of secular and resonant interactions , where secular interactions shape the disc eccentricity and orientation , while resonant interactions create dust traps and dictate their location in the disc . to understand and analyze the stability state of this system
, we first use the criteria from @xcite to estimate the width of the chaotic zone around a planet with the best fit parameters for an asteroid of the mass of the moon used as a proxy for the total mass of the disc see figure [ fig:9 ] .
this criteria predicts than in the absence of planet - disc apse alignment , the outer stable zone around the planet is restricted to a small portion of the plot with @xmath49au and @xmath50 , while if the planet and disc are aligned , the stable zone extend to @xmath51au with @xmath52 . at the beginning of the simulations , since the disc and planet are not initially aligned , particles with @xmath53 au and @xmath54 are rapidly removed as they lie within the chaotic zone highlighted by the red region of figure [ fig:9 ] . however , after @xmath55 years ( which corresponds to @xmath56 for this @xmath57 planet at 31.07 au to sculpt the disc at 48 au ) , secular interactions start to align the disc and planet . as a consequence ,
the chaotic zone shrinks to the blue region of figure [ fig:9 ] , allowing the particles to evolve over a broader range of semi - major axes and eccentricities .
30% of the initial test particles population has survived by this stage .
the top panel of figure [ fig:10 ] illustrates the impact of secular interaction on the disc eccentricity and orientation .
initially the particles complex eccentricities ( @xmath58 , @xmath59 ) are restricted to the green disc of radius @xmath2 centred on the origin , which corresponds to particles having an initial eccentricity range between @xmath60 ( as set in the initial conditions ) . by @xmath61
, the complex eccentricities are now precessing about the forced eccentricity , @xmath62 , as indicated by the blue arrow and form the red disc of radius @xmath63 .
in addition , because the planet is initially set with its pericentre @xmath64 , the direction of the disc forced pericentre illustrated by the blue arrow is the @xmath65-axis , i.e. , @xmath66 .
the disc and planet are therefore aligned .
, @xmath59 ) of all test particles constituting the disc at @xmath67 ( green ) and at @xmath68yrs ( red ) .
the blue arrow shows the location and direction of the forced eccentricity , @xmath69 , of the particles population at @xmath70 yrs .
( bottom ) evolution plot of the semi - major axis and eccentricity of three test particles trapped in a 4:3 , 3:1 and 5:2 mmr with the planet .
, scaledwidth=35.0% ] in addition to secular interactions , the disc location is also shaped by a set of resonances due to the interaction with the planet , which result in the presence of dusty clumps in the deprojected synthetic image of the best fit disc model in figure [ fig:6 ] . at the beginning of the simulation , while 70% of the test particles are scattered by the planet , the surviving particles are clustered around 38.5 , 48 and 54 au , which correspond to the 4:3 , 3:1 and 5:2 mean motion resonances .
the evolution of the semi - major axis and eccentricity of two particles trapped in the 4:3 , 3:1 and 5:2 mean motion resonances is shown in figure [ fig:10 ] .
the 3:1 mmr is trapping test particles at @xmath71au and , as a result , the particles eccentricity is undergoing adiabatic growth , while the 5:2 mmr is trapping particles around 54 au .
these resonances induce stability in the disc and the final disc is made of long term surviving particles with semi major axis mainly clustered around these mmrs with an eccentricity @xmath72 .
the new generation of high contrast imaging instruments , such as the _ gpi _ , can reach contrasts of @xmath73 within the inner @xmath74 . using the isochrones computed by for the evolutionary models of brown dwarfs and giant exoplanets applied to a 7 jupiter mass of a same age than hd 115600
, we estimate the flux ratio contrast to be @xmath75 in the h band . while technically within the current detection threshold , there are several reasons why our best fit planet was not detected on the currie et al . image .
first , given the high inclination of the system ( @xmath76 ) , the apparent orbit of the planet may be hidden by the disc emission , and secondly , most of the planet orbit is likely concealed by the coronograph mask which has an apparent radius of @xmath77au , as can be seen on the right panel of figure [ fig:6 ] .
we however note that a small portion of the planet s orbit around the apocentre is situated beyond the mask , and the planet at this position could be marginally detectable .
in this work , we conducted a numerical search for a planet responsible for shaping the debris disc of hd 115600 recently imaged in the h band by @xcite using _
gpi_. we first use the radiative transfer code _ mcfost _ with a parametric disc structure to derive the dust grain properties needed to fit the observed sed in order to create synthetic images .
we then used a modified n - body integrator which incorporates radiation forces to dynamically model the interaction between the potential companion and the debris disc , and used _
mcfost _ to produce synthetic images to compare with observations .
we explored the 3d parameter space ( @xmath31,@xmath32 and @xmath33 ) of the potential planet using two different methods : a classic grid exploration over 360 grid points , and a _ mcmc _ approach over 720 iterations .
our main results are as follows : * from our sed modeling , and by selecting the greenburg model where the grains are made of a core of silicate with a coating of water ice , the disc emission is best reproduced by a disc with a total dust mass of 0.2 @xmath78 with a minimal and maximal grain size of @xmath79 m and @xmath80 m .
we however note that these parameters are not well constrained . * from image modeling and using the _ mcmc _ scheme to explore the parameter space
, we reproduce the hd115600 disc eccentricity , projected offset , disc width and brightness peak with a @xmath57 planet located at @xmath81au with @xmath82 .
such a planet shapes the disc via a mix of secular interactions forcing the disc eccentricity , orientation and offset , as well as a set of planet resonances , at 4:3 , 3:1 and 5:2 , trapping dust at @xmath83 , @xmath84 and @xmath85 au respectively . *
while technically detectable by _ gpi _ , the apparent and projected planet orbit is likely hidden by the disc emission as well as partially concealed by the coronograph mask due to the high inclination of the system . * using a same total number of simulations , we compared two methods of exploring the parameter space : a _ mcmc _ sampling of the parameter space versus a grid method .
we find that the _ mcmc _ scheme not only finds a better fit to the observations , but also converges faster toward the best fit region .
however we find that using a _ mcmc _
scheme on such a short number of iterations presents some limitations : while our _ mcmc _ implementation barely passed the convergence test such as the @xcite , it also only provides weak constraints on the confidence intervals .
increasing the numbers of iterations would increase the quality of the best fit estimation .
we chose to conduct our study with a small number ( 720 ) of simulations for multiple reasons : first very few studies in the literature uses more than a few hundred dynamical simulations as they are computationally expensive and , secondly , after about a hundred iterations , the majority of subsequent models resulted in disc parameters comparable to the values obtained by @xcite within the observational uncertainties .
finally our planetary parameters uncertainties estimated with 720 _ mcmc _ iterations are well below the current uncertainties of parameters for planets detected via direct imaging . for low dimensional problems ( low @xmath86 ) , like our study
, a traditional grid of @xmath87 points only requires @xmath88 evaluations and presents the advantage of being trivially parallelized . for higher dimensional problems
, it should be noted the
_ efficiently converges like @xmath89 , which is dimensionally independant and therefore a large advantage over the grid technique . in this study
, we compare the efficiency of exploring our @xmath90 parameter space with a grid and a _ mcmc _ algorithm .
we conclude that this work demonstrates that the _ mcmc _ scheme is a promising method to efficiently explore the parameter space of dynamical simulations and allows us to localize a better - fit region faster than the grid method , although we also stress that the limited number of simulations makes reaching the convergence state quite challenging .
there are numerous schemes existing in the literature to explore multidimensional parameter spaces which could represent very good alternatives , such as the _ sparse grid mcmc _ @xcite a scheme combining the use of both a grid and _ mcmc _ , or the use of grids of different resolutions successively @xcite .
the debris disc of hd 115600 shows interesting features which can be reproduced by simulating the gravitational influence of an inner massive planet on the disc structure .
additional observations of hd 115600 at different wavelengths could help constraining the planetary parameters furthermore .
this work was performed on the swinstar supercomputer at swinburne university of technology . e.t .
would like to thank chris blake and caitlin adams for useful discussions about the _ mcmc _ algorithm , as well as thayne currie and tadashi mukai for providing information about the grain composition and modeling .
we also thank the anonymous referee for constructive feedback which have helped improve this paper . e.t .
was supported by a swinburne university postgraduate research award ( supra ) .
99 v. , guaita l. , gawiser e. , padilla n. , 2011 , in aasma 217 vol .
43 of bulletin of the aas , sed fitting with markov chain monte carlo : the case of z=2.1 lyman alpha emitters .
p.407.04 j. c. , lagrange a. m. , mouillet d. , papaloizou j. c. b. , grorod p. a. , 1999 , a&a , 348 , 557 h. h. , beichman c. a. , gillett f. c. , de jong t. , houck j. r. , low f. j. , neugebauer g. , walker r. g. , wesselius p. r. , 1984 , apjl , 278 , l23 i. , chabrier g. , barman t. s. , allard f. , hauschildt p. h. , 2003 , a&a , 402 , 701 , h. , augereau , j .- c . , bonsor , a. , graham , j. r. , kalas , p. , lebreton , j. , lagrange , a .- m . , ertel , s. , faramaz , v. , & thbault , p. 2014 , a&a , 561 , a43 c. h. , mamajek e. e. , bitner m. a. , pecaut m. , su k. y. l. , weinberger a. j. , 2011 , apj , 738 , 122 c. h. , mittal t. , kuchner m. , forrest w. j. , lisse c. m. , manoj p. , sargent b. a. , watson d. m. , 2014 , apjs , 211 , 25 m. k. , carlin b. p. , 1996
assoc , 91 , 883 t. , lisse c. m. , kuchner m. , madhusudhan n. , kenyon s. j. , thalmann c. , carson j. , debes j. , 2015 , apjl , 807 , l7 r. m. , et al .
2012 , vizier online data catalog , 2311 , a. t. , & maddison , s. t. 2005 , apj , 625 , 398 , j. s. , 1969 , jgr , 74 , 2531 j. , bucher m. , ferreira p. g. , moodley k. , skordis c. , 2005 , mnras , 356 , 925 j. , gaudi b. s. , agol e. , 2013 , pasp , 125 , 83 125 , 83 , v. , beust , h. , thbault , p. , augereau , j .- c . , bonsor , a. , del burgo , c. , ertel , s. , marshall , j. p. , milli , j. , montesinos , b. , mora , a. , bryden , g. , danchi , w. , eiroa , c. , white , g. j. , & wolf , s. 2014 , a&a , 563 , a72 a. , rubin d. b. , 1992 , statist .
, 7 , 457 t. m. , et al . , 2011 , aj 142 , 15 , c. a. , morais , m. h. m. , & correia , a. c. m. 2013 , mnras , 436 , 3547 j. m. , zhao n. s. , hage j. i. , 1989 , advances in space research , 9 , 3 a. a. , templeton m. , terrell d. , smith t. c. , levine s. , welch d. , 2016 , vizier online data catalog , 2336 j. , wittenmyer r. a. , hinse t. c. , marshall j. p. , mustill a. j. , tinney c. g. , 2013,mnras , 435 , 2033 d. , et al .
, 2010 , a&a , 514 , a1 p. , fitzgerald m. p. , graham j. r. , 2007 , apjl , 661 , l85 j. e. , stapelfeldt k. r. , bryden g. , plavchan p. , 2012
, aj , 144 , 45 a .- m .
, et al . , 2016 ,
a&a , 586 , l8 h. f. , duncan m. j. , 1994 , icarus , 108 , 18 a. , bridle s. , 2002 , prd , 66 , 103511 menze b. h. , van leemput k. , honkela a. , konukoglu e. , weber m .- a . , ayache n. , golland p. , 2011
, a generative approach for image - based modeling of tumor growth , springer berlin heidelberg , berlin , heidelberg , pp 735747 t. , yamamoto t. , 1982 , a&a , 107 , 97 a. j. , wyatt m. c. , 2009 , mnras , 399 , 1403 s. j. , poole g. b. , croton d. j. , 2013 , mnras , 428 , 2001 e. r. , kuchner m. j. , 2015 , apj , 798 , 83 e. r. , kuchner , m. j. , 2015 , apj , 815 , 61 , t. d. & wyatt , m. c. 2014 , mnras , 443 , 2541 m. j. , mamajek e. e. , bubar e. j. , 2012 , apj , 746 , 154 c. , mnard f. , duchne g. , bastien p. , 2006 ,
a&a , 459 , 797 , a. c. & thorndike , s. 2002 , apjl , 578 , l149 raftery a. e. , lewis s. m. , 1995 , in in practical markov chain monte carlo ( w.r .
gilks , d.j .
spiegelhalter and . chapman and hall ) , pp 115130 g. o. , rosenthal j. s. , 2001 , statist .
, 16 , 351 g. o. , rosenthal j. s. , schwartz p. o. , 1998 , j. appl .
probab . , 35 , 1 s. , demleitner m. , schilbach e. , 2010 , aj , 139 ,
2440 sinharay s. , 2003 , ets research report series , 2003 , i t. p. , berendsen h. j. c. , stam a. j. , 1986 , molecular physics , 57 , 89 d. , 2014 , mnras , 438 , 3577 e. , maddison s. t. , 2015 , pasa , 32 , e039 e. , maddison s. t. , 2016 , mnras , 457 , 1690 j. j. , et al . , 2015 , apjl 811 , l19
u. , wilkinson m. i. , read j. i. , walker m. g. , 2015 , nature communications , 6 , 7599 f. , 2007 , a&a , 474 , 653 , m. c. , dermott , s. f. , telesco , c. m. , fisher , r. s. , grogan , k. , holmes , e. k. , & pia , r. k. 1999 , apj , 527 , 918
the small dust grains experience orbital perturbation by stellar radiation forces and are short - lived .
collisions are the assumed mechanism by which new dust is constinuously created , and following our discussion in section 4.1.2 , we assume the debris disc of hd 115600 to be either ( i ) in a steady - state where larger planetesimals collide and constinuously replenish the disc with small grains or , given the young age of the system , ( ii ) is replenished by collisions of primordial grains before all are removed by stellar radiation forces . either way , collisions are expected to occur within hd 115600 debris disc on a timescale shorter than the simulation duration , and by stacking the dust distribution over the simulation duration , at any time @xmath91 we therefore mimic the dust distribution for grains whose ages range from @xmath91 years old to those newly created .
it is valid to then ask if mimicing the effect of collisions smoothens out the planet - induced features in the disc and therefore impacts the resulting synthetic disc image ( as well as the subsequent determination of the disc parameters ) . in figure
[ fig : a1 ] , snapshots of the dust distribution at the early stage of a simulation ( @xmath92 ) shows a broad slighly perturbed disc , while later snapshots taken over @xmath93 ( or 45000 years ) all display a clear eccentric ring that is apse - aligned with the planet . by @xmath94 , the presence of two dust clumps corresponding to mmr trapping can be clearly distinguished and little disc evolution occurs after that .
interestingly , the structure observed at @xmath94 is consistent with the disc structure illustrated in the deprojected _ mcfost _ synthetic image in figure [ fig:6 ] , for which the dust density distribution was obtained by stacking all the snapshots over the lifetime of the simulations .
therefore , because both the secular interactions and resonnance trapping occur rapidly within the disc ( see also figure [ fig:10 ] ) , no significant difference is expected between the dust density distribution at the end of the simulations and the dust distribution obtained by stacking .
a quick way to verify that the parameter space has been efficiently sampled is to have an overall acceptance rate of 25 to 40% in the _ mcmc _ chain .
this requires adjusting the standard deviation of the three normal prior parameter distributions , from which a new set of parameters is drawn at every _
mcmc _ step .
a commonly used method is to have a some correlation stepping when drawing a new set of parameters at each _ mcmc _ iteration , meaning that the value of the standard deviation for each prior distribution is based on how the three parameters are correlated . in this section ,
we provide a short description on how to achieve this but we refer to @xcite for more details . 1 . step 1 : the method consists in initially setting the normal prior distributions with a standard deviation of @xmath95 for all parameters , and to run the _ mcmc _ chain for a few hundreds iterations .
2 . step 2 : by recording how the chain moves through the parameter space , we can estimate the covariance matrix , * @xmath96 * , of the three parameters and therefore know how the sets of parameters over all _ mcmc _ iterations are correlated .
3 . step 3 : the next stage is to adjust the standard deviation of the prior distributions so that the sets of parameters would follow the covariance we just estimated , and this requires to generate correlated random numbers .
this process can be easily done by first performing a cholesky decomposition on the covariance matrix , which consists in finding an upper diagonal matrix * @xmath97 * such that * * @xmath96**=*@xmath98*. 4 .
step 4 : the cholesky matrix , * @xmath97 * , then needs to be updated to * @xmath99 * to obtain a better acceptance rate , where * * @xmath99**@xmath100*@xmath97 * , with @xmath90 , the number of parameters constituting the parameter space . 5 .
step 5 : the new standard deviation for each parameters , @xmath101 and @xmath102 is then obtained by applying * @xmath99 * to the vector ( @xmath103 ) .
step 6 : a new _ mcmc _ chain can be run using the correlated stepping @xmath101 and @xmath102 , to generate the set of parameters at each iteration of the chain .
if the new acceptance rate is not inside the ideal range , then steps 2 to 6 must be repeated until the acceptance rate enters the correct range .
[ lastpage ]
|
radial and azimuthal features ( such as disc offsets and eccentric rings ) seen in high resolution images of debris discs , provide us with the unique opportunity of finding potential planetary companions which betray their presence by gravitationally sculpting such asymmetric features .
the young debris disc around hd 115600 , imaged recently by the _
gemini planet imager _ , is such a disc with an eccentricity @xmath0 and a projected offset from the star of @xmath1 au . using our modified n - body code which incorporates radiation forces
, we firstly aim to determine the orbit of a hidden planetary companion potentially responsible for shaping the disc .
we run a suite of simulations covering a broad range of planetary parameters using a _ monte carlo markov chain _ sampling method and create synthetic images from which we extract the geometric disc parameters to be compared with the observed and model - derived quantities .
we then repeat the study using a traditional grid to explore the planetary parameter space and aim secondly to compare the efficiency of both sampling methods .
we find a planet of 7.8 jupiter mass orbiting at 30 au with an eccentricity of @xmath2 to be the best fit to the observations of hd 115600 .
technically , such planet has a contrast detectable by direct imaging , however the system s orientation does not favour such detection . in this study , at equal number of explored planetary configurations , the monte carlo markov chain not only converges faster but provides a better fit than a traditional grid . [ firstpage ] planetary systems - circumstellar matter - methods : numerical - methods : statistical - stars : individual ( hd 115600 )
|
principal component analysis ( pca ) is arguably the most successful of dimensionality reduction techniques .
given samples @xmath3 from a @xmath4-dimensional distribution , @xmath5 , pca seeks the direction of maximum variability .
assuming for simplicity the @xmath6 s to be centered ( i.e. @xmath7 ) , and denoting by @xmath8 a random vector distributed as @xmath6 , the objective is to estimate the solution of @xmath9 the solution of this problem is the principal eigenvector of the covariance matrix @xmath10 .
this is normally estimated by replacing expectation above by the sample mean , i.e. solving @xmath11 denoting by @xmath12 the matrix with rows @xmath3 , the solution is of course given by the principal eigenvector of the sample covariance @xmath13 , that we will denote by @xmath14 .
this approach is known to be consistent in low dimension .
let @xmath15 be the solution of problem ( [ eq : populationpca ] ) . if @xmath16 , then @xmath17 in probability @xcite .
on the other hand , it is well understood that consistency can break dramatically in the high - dimensional regime @xmath18 .
this phenomenon is crisply captured by the spiked covariance model @xcite , that postulates @xmath19 where @xmath15 has unit norm , @xmath20 are i.i.d .
@xmath4-dimensional standard normal vectors @xmath21 , and @xmath22 is a unit - norm vector but for our purposes it is more convenient to consider @xmath23 as a given deterministic vector .
equivalently , we can condition on @xmath24 . ] .
the above model can also be written as @xmath25 where @xmath26 has i.i.d .
entries @xmath27 .
the spectral properties of the random matrix @xmath28 defined by the [ eq : spikedmatrix ] have been studied in detail across statistics , signal processing and probability theory @xcite . in the limit @xmath29 with @xmath30 ,
the leading eigenvector @xmath31 undergoes a phase transition : @xmath32 in other words , [ eq : samplepca ] contains information about the signal @xmath15 if and only if the signal - to - noise ratio is above the threshold @xmath33 . below that threshold ,
the principal component is asymptotically orthogonal to the signal .
the failure of pca has motivated significant effort aimed at developing better estimation methods .
a recurring idea is to use additional structural information about the principal eigenvector @xmath15 , such as its sparsity @xcite or its distribution ( within a bayesian framework ) @xcite .
here we focus on the simplest type of structural information , namely we assume @xmath15 is known to be non - negative with @xmath34 an arbitrary , known , orthant , can be reduced to the present one . ] .
it is then natural to replace the [ eq : samplepca ] problem with the following one ( whereby we use the matrix @xmath28 to represent the data ) : @xmath35 notice that this problem in non - convex and can not be solved by standard singular value decomposition .
indeed it is in general np - hard by reduction from maximum independent set @xcite .
two questions are therefore natural : given the additional complexity induced by the non - negativity constraint , does this constraint reduce the statistical error significantly ? are there efficient algorithms to solve the [ eq : positivepca ] problem ? in this paper we answer _ positively _ to both questions within the spiked covariance model .
namely denoting by @xmath36 the solution of the [ eq : positivepca ] problem , we provide the following contributions : 1 .
we unveil a new phase transition phenomenon concerning @xmath36 that is analogous to the classical one , see eq .
( [ eq : pcaphasetransition ] ) .
namely , for @xmath37 , @xmath38 stays bounded away from @xmath39 , while , for @xmath40 , there exists vectors @xmath15 such that @xmath41 as @xmath29 .
+ non - negative pca is superior to classical pca in this respect since @xmath42 strictly .
2 . we prove an explicit formula for the asymptotic scalar product @xmath43 . non - negative pca is superior to [ eq : samplepca ] also in this respect .
namely @xmath38 is strictly larger than @xmath44 with high probability as @xmath29 .
+ note that the non - negativity constraint breaks the rotational invariance of classical pca ( under the spiked model ) . as a consequence , not all spikes @xmath15 are equally hard or easy to estimate .
we use our theory to characterize the least favorable vectors @xmath15 .
3 . we prove that ( for any fixed @xmath45 ) a @xmath46 approximation to the non - convex optimization [ eq : positivepca ] problem can be found efficiently with high probability with respect to the noise realization .
our algorithm has complexity of order @xmath47 , where @xmath48 is the maximum of the complexity of multiplying a vector by @xmath28 or by @xmath49 .
technically , our approach has two components .
we use sudakov - fernique inequality to upper bound the expected value of the [ eq : positivepca ] optimization problem .
we then define an iterative algorithm to solve the optimization problem , and evaluate the value achieved by the algorithm after any number @xmath50 of iterations .
this provides a sequence of lower bounds which we prove converge to the upper bound as the number of iterations increase .
more precisely , we use an approximate message passing ( amp ) algorithm of the type introduced in @xcite .
each iteration requires a multiplication by @xmath28 and a multiplication by @xmath49 plus some lower complexity operations .
while amp is not guaranteed to solve the [ eq : positivepca ] problem for arbitrary matrices @xmath28 , we establish the following properties : 1 . after any number of iterations @xmath50 , the algorithm produces a running estimate @xmath51 that satisfies the constraints @xmath52 and @xmath53 .
+ further the limit @xmath54 exists almost surely , and @xmath55 can be computed explicitly as a function of the empirical law of entries of @xmath15 .
analogously , the asymptotic correlation @xmath56 can be computed explicitly .
2 . denoting by @xmath57 the upper bound on the value of the optimization [ eq : positivepca ]
problem implied by sudakov - fernique inequality , we prove that @xmath58 for all @xmath59 for some dimension - independent @xmath60 .
this implies that sudakov - fernique inequality is asymptotically tight in the high - dimensional limit .
the asymptotic correlation converges to a limit as the number of iteration tends to infinity @xmath61 ( the convergence is , again , exponentially fast ) .
further , if we add the constraint @xmath62 to the [ eq : positivepca ] optimization problem , sudakov - fernique s upper bound on the resulting value is asymptotically smaller than @xmath63 for any @xmath45 . +
this implies that @xmath64 .
finally , we generalize our analysis to the case of symmetric matrices , namely assuming that data consist of a @xmath65 symmetric matrix @xmath28 : @xmath66 with @xmath67 , @xmath68 . here
@xmath69 is a noise matrix such that @xmath70 are independent with @xmath71 for @xmath72 and @xmath73 . in this case
we study the analogue of the [ eq : positivepca ] problem , namely @xmath74 the non - negativity constraint on principal components arises naturally in many situations : we briefly discuss a few related areas .
let us emphasize that the theoretical understanding of the methods discussed below is much more limited than for [ eq : samplepca ] . * microarray data .
* microarray measurements of gene expression result in a matrix @xmath12 whereby @xmath75 denotes the expression level of gene @xmath76 in sample @xmath77 .
several authors @xcite seek for a subset of genes that are simultaneously over - expressed ( or under - expressed ) in a subset of samples .
lazzeroni and owen @xcite propose a model of the form @xmath78 where @xmath79 indexes such gene groups ( or ` layers ' ) , and @xmath80 , @xmath81 indicate the level of participation of different samples or different genes in group @xmath79 .
these authors assume @xmath82 , but it is natural to relax this condition allowing for partial participation in group @xmath79 , i.e. @xmath83 $ ] , by a change of normalization , this constraint can be simplified to @xmath84 .
note a few differences with respect to our work : * we study a model with only one non - negative component .
while eq .
( [ eq : plaid ] ) corresponds to a model with multiple @xmath85 components , in practice several authors fit one ` layer ' at a time , hence effectively reducing the problem to a single - component case . +
extending our analysis to the multiple component case will be the object of future work .
* the non - negativity constraint is imposed in the model ( [ eq : plaid ] ) on both components .
this is a relatively straightforward modification of our setting .
* several studies ( e.g. @xcite ) fit models of the form eq .
( [ eq : plaid ] ) using greedy optimization methods .
their conclusions are based on the unproven belief that these methods approximately solve the optimization problem .
our results ( establishing convergence , with high probability , of an iterative method ) provide some mathematical justification for this approach .
* neural signal processing . *
neurons activity can be recorded through thin implanted electrodes .
the resulting signal is a superposition of localized effects of single neurons ( spikes ) . in order to reconstruct the single neuron activity , it is necessary to assign each spike to a specific neuron that created it , a process known as ` spike sorting ' @xcite .
once spikes are aligned , the resulting data can be viewed as a matrix @xmath86 , j\in [ p]}$ ] , where @xmath77 indexes the spikes and @xmath76 time ( or a transform domain , e.g. wavelet domain ) . in this context , principal component analysis is often used to project each row of @xmath28 ( i.e. each recorded spike ) in a low dimensional space , or decomposing it as a sum of single neurons activity , see e.g. @xcite .
clustering may be carried out after dimensionality reduction .
note that each spike is a sum of single neuron activity _ with non - negative _ coefficients .
in other words , the @xmath77-th row of @xmath28 reads @xmath87 where @xmath88 , @xmath89 are the signatures of @xmath90 neurons and @xmath91 are non - negative coefficients . again , this corresponds to a multiple component version of the problem we study here . to the best of our knowledge , the non - negativity constraint has not been exploited in this context .
* non - negative matrix factorization . *
initially introduced in the context of chemometrics @xcite , non - negative matrix factorization attracted considerable interest because of its applications in computer vision and topic modeling .
in particular , lee and seung @xcite demonstrated empirically that non - negative matrix factorization successfully identifies parts of images , or topics in documents corpora .
a mathematical model to understand these findings was put forward in @xcite and most recently studied , for instance , in @xcite .
note that these results only apply under a no - noise or very - weak noise conditions , but for multiple components .
further , the aim is to approximate the original data matrix , rather than estimating the principal components . in this sense ,
non - negative matrix factorization is the farther among all related areas to the scope of our work .
* approximate message passing .
* approximate message passing algorithms proved successful as a fast first - order method for compressed sensing reconstruction @xcite .
their definition is inspired by ideas from statistical mechanics and coding theory @xcite , see also @xcite for further background
. one attractive feature of amp algorithms is that their high - dimensional asymptotics can be characterized exactly and in close form , through ` state - evolution ' @xcite .
several applications and generalizations were developed by rangan @xcite , schniter @xcite and collaborators .
in particular schniter and cevher @xcite apply amp the problem of reconstructing a vector from bilinear noisy observations , a problem that is mathematically equivalent to the one explored here .
these authors consider however more complex bayesian models , and evaluate performances through empirical simulations , while we characterize a fundamental threshold phenomenon in a worst case setting .
similar ideas were applied in @xcite to the problem of dictionary learning , and in @xcite to hyperspectral imaging . finally , kabashima and collaborators @xcite study low - rank matrix reconstruction using a similar approach , but focus on the case in which the rank scales linearly with the matrix dimensions . in section
[ sec : main ] we present formally our results , both for symmetric matrices and rectangular matrices . as mentioned above , the proof is obtained by establishing an upper bound on the value of the [ eq : positivepca ] optimization problem using sudakov - fernique inequality , and a lower bound by analyzing an amp algorithm . the upper bound is outlined in section [ sec : upperboundslepiannonnegative ] . section [ sec : amp ] introduces formally amp and its analysis , hence establishing the desired lower bound as well as the convergence properties of this algorithm .
section [ sec : numerical ] presents a numerical illustration of the phase transition phenomenon , and of the behavior of our algorithm .
finally , section [ sec : proofupperbounds ] contains proofs , with some technical details deferred to the appendices .
in this section we present formally our results . for the sake of clarity , we consider first the case of symmetric ( wigner ) matrices , and then the case of rectangular ( or sample covariance , wishart ) matrices .
indeed formul for symmetric matrices are somewhat simpler . before doing that , it is convenient to introduce some definitions .
( for basic notations , we invite the reader to consult section [ sec : notations ] . ) our results concern sequences of matrices @xmath28 with diverging dimensions @xmath92 , and are expressed in terms of the asymptotic empirical distribution of the entries of @xmath15 .
this is formalized through the following definition .
[ def : convergesempirical ] let @xmath93 be a sequence of vectors with , for each @xmath94 , @xmath95 , and @xmath96 be a ( borel ) probability measure on the real line @xmath97 .
then we say that @xmath98 _ converges in empirical distribution _ to @xmath96 if the probability measure @xmath99 converges weakly to @xmath96 and the second moment of @xmath100 converges as well or , equivalently , @xmath101 . with an abuse of terminology , we will say that @xmath93 converges in empirical distribution to @xmath102 if @xmath102 is a random variable with law @xmath96 .
given a random variable @xmath102 , we let @xmath103 denote its law .
we next define a few functions of such a law .
[ def : functionsfgrts ] let @xmath104 be a real non - negative random variable independent of @xmath105 and @xmath106 be a real number .
we define the two functions @xmath107 using @xmath108 and @xmath109 define the following ` rayleigh functions ' @xmath110 for @xmath111 , we also define @xmath112 as the unique non - negative solution of @xmath113 and @xmath114 as the unique non - negative solution of @xmath115 .
note that the above functions depend on the random variable @xmath104 only through its law @xmath116 , but we prefer the notation say @xmath117 to the more indirect @xmath118 .
existence and well - definedness of @xmath119 and @xmath120 are proved in lemma [ lem : twelldefined ] below .
further in lemma [ lem : rvariations ] we prove that the functions @xmath121 respectively @xmath122 have a unique maximum reached respectively at @xmath112 and at @xmath114 .
our results become particularly explicit in case @xmath15 is sparse which ( in the asymptotic setting ) is equivalent to @xmath123 small .
we introduce some terminology to address this case .
[ def : uniformconvergence ] given a real random variable @xmath104 , we let @xmath124 denote its sparsity level .
we let @xmath125 be the set of probability measures @xmath96 supported on @xmath126 , with second moment equal to one , and , for @xmath127 , @xmath128 . given a function @xmath129 , @xmath130 , and a number @xmath131
, we write that @xmath132 _ uniformly over @xmath125 _
if @xmath133 in the following , we will often state that an event holds almost surely as the dimensions of the random matrix @xmath28 tend to infinity .
it is understood that such statements hold with respect to the law of a sequence @xmath134 of _ independent _ random matrices distributed according to the [ eq : spikedmatrix ] or the [ eq : symmetricmodel ] . for the sake of comparison , we begin by recalling some asymptotic properties of [ eq : samplepca ] . given @xmath135 symmetric distributed according to the [ eq : symmetricmodel ] , we denote by @xmath14 its principal eigenvector , and by @xmath136 the corresponding eigenvalue .
this model has been studied in probability theory under the name of ` low rank deformation of a wigner matrix ' .
the following is a simplified version of the main theorem in @xcite .
let @xmath137 be a rank - one deformation of the gaussian symmetric matrix @xmath138 , with @xmath71 independent for @xmath72 , and @xmath139 .
then we have , almost surely @xmath140 further @xmath141 numerous refinements exist on this basic result , see for instance @xcite .
our analysis provides a version of this theorem that holds for non - negative pca , and is intriguingly similar to the original one .
its proof can be found in appendix [ sec : proofofmaintheorems ] .
[ th : mainsym ] let @xmath137 be a rank - one deformation of the symmetric gaussian matrix @xmath138 with @xmath71 independent for @xmath72 , and @xmath139 . further let @xmath142 be the value of the [ eq : positivepcasymm ] problem , and @xmath143 be any of the optimizers .
finally assume that @xmath144 is such that @xmath145 converges in empirical distribution to @xmath146 .
then ( with the notation introduced in definition [ def : functionsfgrts ] ) , we have almost surely @xmath147 further , uniformly over @xmath125 , @xmath148 and @xmath149 the statement in theorem [ th : mainsym ] is dependent on the empirical distribution of the entries of @xmath15 .
it is of special interest to characterize the least favorable situation , i.e. the distribution corresponding to the smallest scalar product @xmath38 .
this has two motivations : @xmath0 to guarantee the minimum value of @xmath150 achieved by a solution @xmath36 of the optimization problem [ eq : positivepcasymm ] and @xmath1 to describe the least favorable signal @xmath15 .
the worst - case scenario is realized for a particularly simple distribution , namely 2-atoms distribution , with an atom at @xmath39 .
however , unlike in classical denoising @xcite , the worst case mixture is not obtained by setting all the allowed coordinates to non - zero . in the following theorem
we are interested in the worst case among @xmath151-sparse signals , or equivalently in vector sequences @xmath152 such that @xmath153 , or @xmath154 since sparse signals are naturally interesting for applications .
[ th : fworstcase ] consider the [ eq : symmetricmodel ] with the [ eq : positivepcasymm ] estimator . if @xmath155 , then there exists a sequence of vectors @xmath156 such that @xmath157 and , almost surely , @xmath158 for any @xmath159 , there exists @xmath160 $ ] such that the following is true
. let @xmath161 be the random variable with law @xmath162 then for any sequence of vectors @xmath163 such that @xmath164 we have , almost surely , @xmath165 equality holds if @xmath166 is the vector with @xmath167 non - zero entries , all equal to @xmath168 .
we defer this proof to section [ sec : worstcase ] .
the worst case mixture @xmath169 as well as the function @xmath170 can be expressed explicitly in terms of the gaussian distribution function , see section [ sec : worstcase ] .
we develop a very similar theory for the case of rectangular matrices .
our first result characterizes the value of the [ eq : positivepca ] problem , and the estimation error , in analogy with theorem [ th : mainsym ] .
the proof can be found in appendix [ sec : proofofmaintheorems ] .
[ th : mainrec ] let @xmath171 be a rank - one deformation of the gaussian matrix @xmath138 with @xmath71 independent , and @xmath172 . further let @xmath173 be the expected value of the [ eq : positivepca ] problem , and @xmath143 be any of the optimizers .
assume that @xmath29 with convergent aspect ratio @xmath174 , and that @xmath175 converges in empirical distribution to @xmath146 .
then ( with the notation introduced in definition [ def : functionsfgrts ] ) , we have almost surely @xmath176 further , uniformly over @xmath125 , @xmath177 and @xmath178 finally , in the same fashion as theorem [ th : fworstcase ] , we can characterize the worst case signals @xmath15 .
[ th : fworstcaserec ] consider the [ eq : spikedmatrix ] , with the [ eq : positivepca ] estimator .
if @xmath179 , then there exists a sequence of vectors @xmath180 such that @xmath181 and , almost surely , @xmath182 for any @xmath183 , there exists @xmath184 $ ] such that the following is true .
let @xmath185 be the random variable with law @xmath186 .
then for any sequence of vectors @xmath187 , @xmath188 , we have .
almost surely , @xmath189 equality holds if @xmath190 is the vector with @xmath191 non - zero entries , all equal to @xmath192 . for the proof we refer to section [ sec : worstcase ] which also contains explicit expressions to compute @xmath193 .
we use capital boldface for matrices , e.g. @xmath28 , @xmath138, and lowercase boldface for vectors , e.g. @xmath8 or @xmath194 . the ordinary scalar product between @xmath195 is denoted by @xmath196 .
the @xmath197 norm of a vector is denoted by @xmath198 , and we will occasionally omit the subscript for the case @xmath199 .
the @xmath200 operator norm of the matrix @xmath28 is denoted by @xmath201 . as usual
, we write @xmath202 for the standard gaussian density , and @xmath203 for the gaussian distribution function .
finally we will say that a function @xmath204 is pseudo - lipschitz if there exists a constant @xmath205 such that @xmath206
as mentioned above , theorems [ th : mainsym ] and [ th : mainrec ] are proved in two steps .
we establish an upper bound on the value of the optimization problem by using sudakov - fernique inequality and prove that the bound is tight by analyzing an iterative algorithm that solves the optimization problem .
the first statement concerns the [ eq : symmetricmodel ] .
[ th : upperboundssymmetric ] consider the [ eq : symmetricmodel ] , and let @xmath143 be the [ eq : positivepcasymm ] estimator , with @xmath142 the value of the corresponding optimization problem .
then , under the assumptions of theorem [ th : mainsym ] , we have @xmath207 further , there exists a deterministic function @xmath208 , with @xmath209 such that , almost surely , @xmath210 the second statement concern the ( non - symmetric ) [ eq : spikedmatrix ] . [ th : upperbounds ] consider the [ eq : spikedmatrix ] and let @xmath143 be the [ eq : positivepca ] estimator , with @xmath173 the value of the corresponding optimization problem .
then , under the assumptions of theorem [ th : mainrec ] , we have @xmath211 further , there exists a deterministic function @xmath208 , with @xmath209 such that , almost surely , @xmath212 the proof of lemma [ th : upperbounds ] can be found in section [ sec : proofupperbound ] . the proof for the case of symmetric matrices , cf .
lemma [ th : upperboundssymmetric ] , is completely analogous and we omit it . while the above upper bounds are stated in asymptotic form , the proofs in section [ sec : proofupperbounds ] imply non - asymptotic upper bounds . roughly speaking ,
the above upper bounds hold non - asymptotically up to an additive correction of order @xmath213 .
we use an algorithmic approach to prove a lower bound that matches the upper bound in lemmas [ th : upperboundssymmetric ] , [ th : upperbounds ] .
the algorithm is close in spirit to the usual power method that computes the leading eigenvector of a symmetric matrix @xmath28 by iterating @xmath214 from an arbitrary initialization @xmath215 .
of course the power method is not well suited for the present problem , since it does not enforce the non - negativity constraint @xmath216 .
we will enforce this constraint iteratively by projecting on the feasible set .
similar non - linear power methods were studied previously , for instance in the context of sparse pca @xcite and a statistical analysis of a method of this type was developed in @xcite .
our approach differs substantially from this line of work .
we develop an approximate message passing ( amp ) algorithm that builds on ideas from statistical physics and graphical models @xcite .
remarkably , exact high - dimensional asymptotics for these algorithms have been characterized in some generality using a method known as _ state evolution _ @xcite .
we establish the desired lower bounds by applying this theory to our problem . as before
, we will start by considering the case of symmetric matrices and then move to rectangular matrices .
the amp algorithm is iterative and , after @xmath50 iterations , mantains a state @xmath217 .
we initialize it with @xmath218 , @xmath219 , and use the update rule , for @xmath220 , @xmath221 where @xmath222 and @xmath223 is the normalized projection on the positive orthant : @xmath224 ( the factor @xmath225 is introduced here for future convenience . )
if we neglect the memory term @xmath226 , the algorithm [ eq : ampsym ] is extremely simple : it alternates between a power iteration , and an orthogonal projection onto the constraint set @xmath227 . as proved in @xcite the memory term ( ` onsager term ' ) plays a crucial role in allowing for an exact high - dimensional characterization .
note that @xmath228 does not satisfy in general the positivity constraint .
indeed it is not the algorithm estimate of @xmath15 . after any number @xmath50 of iteration
we construct the estimate @xmath229 state evolution @xcite is a mathematical technique that provides an exact distributional characterization of a class of algorithms that includes [ eq : ampsym ] , under suitable probabilistic models for the matrix @xmath28 . in the present case , we will assume the [ eq : symmetricmodel ] , with @xmath230 converging in empirical distribution to a random variable @xmath104 .
informally , state evolution predicts that as @xmath231 , for any fixed @xmath232 , the state vector @xmath228 is approximately normal with mean @xmath233 and covariance @xmath234 .
in other words , it can be viewed as a noisy version of the signal @xmath15 : @xmath235 the signal - to - noise ratio @xmath236 is determined recursively by letting @xmath237 and for all @xmath232 , @xmath238 . explicitly : @xmath239 with @xmath240 independent of @xmath104 . a formal statement is given below .
[ prop : ampsym ] consider the [ eq : symmetricmodel ] , and assume that @xmath241 converges in empirical distribution to a random variable @xmath242 .
further , let @xmath243 be defined by the state evolution recursion ( [ eq : symmetricstateevolution ] ) .
then , for any pseudo - lipschitz function @xmath244 and any @xmath232 we have , almost surely , @xmath245 where @xmath246 is independent of @xmath242 .
further , the convergence in eq .
( [ eq : limitpsipseudolipsym ] ) also holds for @xmath247 and any @xmath248 .
the proof of this result is a direct application of the results of @xcite and can be found in appendix [ app : ampsym ] .
a second important result that follows from state evolution is that the sequence @xmath249 converges in the following asymptotic sense .
[ lem : diffgoestozerosymm ] under the assumptions of proposition [ prop : ampsym ] , fix any @xmath250 .
then , we have almost surely @xmath251 the proof of this statement is deferred to appendix [ app : diffgoestozerosymm ] . as @xmath252 , @xmath253 , with @xmath254 the unique positive solution of the fixed point equation @xmath255 .
by using the above two propositions , we then obtain the following lower bound , whose proof can be found in section [ sec : prooflowerbounds ] .
[ th : lowerboundsampsymmetric ] consider the [ eq : symmetricmodel ] , and assume that @xmath241 converges in empirical distribution to a random variable @xmath242 .
further , let @xmath256 be the amp iterates as defined by [ eq : ampsym ] and eq .
( [ eq : amp_estimates_symm ] ) .
finally , let @xmath254 be the unique positive solution of the fixed point equation @xmath257 ( equivalently @xmath258 ) .
then we have , almost surely , @xmath259 this provides the necessary lower bound that complements the upper bound based on sudakov - fernique inequality , cf .
section [ sec : upperboundslepiannonnegative ] . in this case
the algorithm keeps track after @xmath50 iterations of @xmath260 and @xmath51 .
these are initialized by setting @xmath218 , @xmath261 , and updated by letting , for @xmath220 , @xmath262 where @xmath263 and @xmath264 and @xmath265 and @xmath266 are defined by : @xmath267 after any number @xmath50 of iteration we construct the estimates @xmath268 these satisfy the normalization and positivity constraints and are used as estimates of @xmath269 , @xmath15 .
we consider the high dimensional setup where @xmath270 , and @xmath271 with converging aspect ratio @xmath272 .
we assume that @xmath273 converges in empirical distribution to @xmath274 and @xmath275 converges in empirical distribution to @xmath242 .
the high dimensional asymptotics of @xmath276 , @xmath228 is characterized as in the symmetric case through state evolution .
we introduce the real - valued _ state evolution sequences _ @xmath277 and @xmath278 through the following recursion for @xmath220 @xmath279 with initial conditions @xmath280 .
we refer to these as to the _ state evolution equations_. roughly speaking , state evolution establishes that @xmath276 is approximately normal with mean @xmath281 and unit covariance , and @xmath228 is approximately normal with mean @xmath282 and unit covariance .
this is formalized below .
[ prop : segeneral ] consider the [ eq : spikedmatrix ] and assume that @xmath273 converges in empirical distribution to a random variable @xmath274 and @xmath283 converges in empirical distribution to a random variable @xmath242 .
further , let @xmath284 , @xmath285 be defined by the state evolution recursion [ eq : seqthetamu ] . then
, for any pseudo - lipshitz function @xmath286 and any @xmath232 we have , almost surely @xmath287 where @xmath246 is independent of @xmath274 and @xmath242 .
further , the convergence in eq .
( [ eq : limitpsipseudoliprec ] ) also holds for @xmath247 and any @xmath248 .
the proof is very similar to the one of proposition [ prop : ampsym ] and is again a direct application of the results of @xcite .
we omit it to avoid redundancy .
we also have an analogous of proposition [ lem : diffgoestozerosymm ] . [
lem : diffgoestozero ] under the assumptions of proposition [ prop : segeneral ] , let @xmath250 be a fixed integer .
then we have , almost surely , @xmath288 we omit the proof , as it is very similar to the one of proposition [ lem : diffgoestozero ] .
in the limit @xmath252 ( and assuming @xmath289 ) , the sequence defined in [ eq : seqthetamu ] converges to a nonzero fixed point @xmath290 satisfying the fixed point equations @xmath291 we will prove that these equations admit a unique positive solution . considering @xmath252 ( after @xmath231 )
we can thus prove the following .
[ th : lowerboundsampnonsymmetric ] consider the [ eq : spikedmatrix ] and assume that @xmath273 converges in empirical distribution to a random variable @xmath274 and @xmath283 converges in empirical distribution to a random variable @xmath242 .
further , let @xmath292 , @xmath293 be the amp estimates as defined by [ eq : ampnonsym ] and eq .
( [ eq : amp_estimates_rec ] ) finally , let @xmath294 be the only positive solution of the fixed point equations ( [ eq : fixedpointrectangular ] ) .
then we have , almost surely , @xmath295 the proof of this theorem can be found in section [ sec : prooflowerbounds ] . as a direct consequence of the characterization of amp established in propositions [ prop : ampsym ] and [ prop : segeneral ] , we can upper bound the number of iterations needed for algorithms [ eq : ampnonsym ] and [ eq : ampsym ] to converge .
we point out that the cost of each step of the amp algorithms is dominated by a matrix vector multiplication .
this operation can easily be parallelized and performed efficiently . to be definite
, we state the next result in the case of symmetric matrices .
a completely analogous statement holds for rectangular matrices .
[ prop : geometricconvergence ] for any law @xmath296 and any @xmath45 there exists a constant @xmath297 such that the following holds true . under the assumptions of proposition
[ prop : ampsym ] , let @xmath293 be the sequence of estimates produced by amp .
then , for all fixed @xmath298 we have @xmath299 the proof of this statement follows immediately from theorem [ th : mainsym ] and [ th : lowerboundsampsymmetric ] .
a more careful treatment of error terms in the latter can be used to show that indeed @xmath300 for some finite constant @xmath301 .
notice that the computational cost of amp is dominated by the one of matrix vector multiplications , call it @xmath302 .
the above discussion indicates that the average - case complexity of the algorithms [ eq : ampnonsym ] and [ eq : ampsym ] is @xmath303 .
we carried out numerical simulations on synthetic data generated following [ eq : symmetricmodel ] .
we use a signal @xmath15 that takes two values : @xmath304 where @xmath305 $ ] is of size @xmath306 .
it is immediate to see that the sequence @xmath307 converges in empirical distribution to a random variable with distribution @xmath308 in other words @xmath146 is the 2-points mixture @xmath309 .
the predictions of theorem [ th : mainsym ] are stated in terms of the function @xmath310 that is rather explicit in this case .
we have @xmath311 we implemented the algorithm [ eq : ampsym ] , and report in figure [ fig : symspik ] the results of numerical simulations with @xmath312 , sparsity level @xmath313 , and signal - to - noise ratio @xmath314 .
in each case we run amp for @xmath315 iterations and plot the empirical average of @xmath316 over @xmath317 instances .
the algorithm convergence is fast and for our purposes this value of @xmath50 is large enough so that @xmath318 and @xmath319 .
( see below for further evidence of this point . )
the results agree well with the asymptotic predictions of theorem [ th : mainsym ] , namely with the curves reporting @xmath320 .
the figure also illustrates that sparse vectors ( small @xmath321 ) correspond to the least favorable signal in small signal - to - noise ratio .
the value @xmath322 corresponds to the phase transition .
[ fig : symspik ] theorem [ th : mainsym ] and proposition [ prop : ampsym ] predict the value of @xmath323 and @xmath324 in the limit @xmath325 .
it is natural to question the validity of the prediction for moderate values of @xmath94 . in order to investigate this point , we performed numerical experiments with amp by generating instances of the problem for several values of @xmath94 and compared the results with the asymptotic prediction of eq .
( [ eq : fvepsilon ] ) . the top left - hand frame in figure [ fig : asymptotic ]
is obtained with @xmath326 , @xmath327 and several value of @xmath328 . for each point
we plot the average of @xmath316 after @xmath329 iterations , over 32 instances . already at @xmath330
the agreement is good , and improving with @xmath94 . in the top - right plot
we plot the deviation between the empirical averages of @xmath331 ( over 32 instances ) and the asymptotic prediction @xmath320 .
the data suggest @xmath332 with @xmath333 . in the bottom frames we plot the theoretical and empirical ( for @xmath334 ) values of @xmath316 for a grid of parameters @xmath335 .
the difference between the two has average @xmath336 and standard deviation @xmath337 .
+ a natural convex relaxation for the [ eq : positivepcasymm ] problem is the semi - definite program @xmath338 it is known @xcite that for @xmath339 the completely positive cone is strictly included in the doubly non - negative cone @xmath340 hence in general this relaxation is not tight .
the solution is a symmetric non - negative matrix @xmath341 .
we extract the leading eigenvector @xmath342 and use its positive part as our approximation for @xmath36 . in simulations we use cvx @xcite to solve [ eq : sdprelaxation ] , and
compare the result to the output of amp stopped after @xmath343 iterations .
the interior point solver of cvx forces us to consider small problems .
we use @xmath344 , @xmath345 , @xmath346 , and average over @xmath347 instances . on a 2.8 ghz core 2 duo with 8 gb of ram
, cvx stops after about @xmath348 seconds and a matlab implementation of amp after @xmath349 ms . on average ,
the convex relaxation method achieves scalar product @xmath350 , while denoting by @xmath351 the output of amp , we obtain @xmath352 . in figure
[ fig : ampvscvx ] we compare the values reached by each algorithm over the 50 instances of the experiment with the predicted asymptotic value value @xmath353 .
the plot suggests that indeed both methods solve to high accuracy the same problem .
given a random variable @xmath242 , with @xmath354 , it is useful to define the function @xmath355 , by @xmath356 in this section we establish several useful properties of the functions @xmath108 , @xmath109 , @xmath121 , @xmath357 introduced in definition [ def : functionsfgrts ] . throughout @xmath242
is a random variable with law @xmath146 supported on @xmath126 and such that @xmath358 .
note that , in particular , @xmath359 with strictly positive probability .
as before , we let @xmath360 .
all statements concern these functions in their domain , namely @xmath108 , @xmath109 , @xmath361 , @xmath362 .
given a function @xmath363 , we will use @xmath364 , @xmath365 to indicate its first and second derivatives .
[ ref : fgvaluesmonotony ] both @xmath366 and @xmath367 are strictly positive , differentiable and upper bounded by 1 .
further @xmath108 is strictly increasing on @xmath97 , with @xmath368 for all @xmath369 , @xmath109 strictly decreasing on @xmath370 , with @xmath371 for all @xmath372 , and @xmath373 is strictly convex on @xmath97 .
finally @xmath374 . and
therefore @xmath375 $ ] and @xmath376 , @xmath377 , @xmath378 , and @xmath379 .
positivity is immediate from the definition .
the upper bound @xmath380 follows cauchy - schwarz inequality . to prove differentiability
, we write @xmath381 with @xmath382 both differentiable ( by dominated convergence ) since @xmath242 and @xmath383 have bounded second moments , and strictly positive .
therefore @xmath108 is differentiable .
a direct calculation yields the following relations @xmath384 using the last expression ( and substituting the previous ones ) , we see that , to prove that @xmath108 is increasing , it is sufficient to prove that @xmath385 which directly follows from cauchy - schwarz inequality , even for @xmath386 , and equality can not hold as @xmath242 and @xmath383 are independent . in order to show that @xmath109 is decreasing on @xmath370 first observe that for any @xmath387 , @xmath388 . differentiating with respect to @xmath389 and using eq .
( [ eq : fderivative ] ) , we get @xmath390 since @xmath108 is strictly increasing , it follows that @xmath391 is strictly decreasing .
finally , the values at @xmath392 are obtained by simple calculus .
the limits as @xmath393 follow by applying dominated convergence both to the numerator and to the denominator of @xmath394 ( or @xmath395 ) , after dividing both by @xmath389 .
[ lem : fginnerproducts ] let @xmath396 , @xmath397 and , for each integer @xmath4 , let @xmath398 be a deterministic vector with @xmath399 and such that @xmath400 converges in empirical distribution to @xmath401 .
similarly , for an integer @xmath94 let @xmath402 be a deterministic vector such that @xmath403 and @xmath404 converges in empirical distribution to @xmath274 with @xmath405 .
then , for any @xmath406 there exists a sequence @xmath407 , with @xmath408 as @xmath231 such that @xmath409 for eq .
( [ eq : firstscalar ] ) note that @xmath410 in order to prove eq .
( [ eq : secondscalar ] ) , first note that @xmath411 we then introduce the notation @xmath412 and @xmath413 , and @xmath414 for the empirical distribution of @xmath415 .
note that we get @xmath416 and @xmath417 since @xmath418 is bounded and lipschitz continuous on @xmath97 , and by assumption @xmath419 converges weakly to @xmath96 , the last expression tends to @xmath39 as @xmath420 , which proves our claim .
[ lem : twelldefined ] each of the equations @xmath421 admits a unique non - negative solution for each @xmath422 , which we denote by @xmath423 ( for eq .
( [ eq : fixpointsymm ] ) ) and @xmath114 ( for eq .
( [ eq : fixpointsymm ] ) ) .
further , we have @xmath424 let us define the function @xmath425 .
we already know ( by lemma [ ref : fgvaluesmonotony ] ) that @xmath426 , so @xmath427 .
also , since @xmath428 , we have @xmath429 .
further @xmath108 is differentiable and hence so is @xmath430 on @xmath431 .
it is therefore sufficient to prove that @xmath430 is strictly decreasing to prove existence and uniqueness of the solution of eq .
( [ eq : fixpointsymm ] ) .
recall that ( cf .
( [ eq : fderivative ] ) ) : @xmath432 we will prove that @xmath433 is concave .
this implies that @xmath430 is decreasing : indeed , by the last equation we have @xmath434 applying the change of variable @xmath435 , we get @xmath436 this shows that the derivative of @xmath437 is strictly negative provided that @xmath438 is non - positive , or @xmath433 concave .
indeed the second term in the last expression is strictly negative because @xmath439 , cf .
lemma [ ref : fgvaluesmonotony ] and eq .
( [ eq : fderivative ] ) we can write @xmath440 as @xmath441 ( where @xmath442 denotes expectation with respect to @xmath443 ) which , since @xmath444 shows that our claim follows from concavity of @xmath445 , see lemma [ lem : ksqrtcccv ] . [
lem : ksqrtcccv ] the function @xmath446 is concave on @xmath126 .
we have @xmath447 and @xmath448 so @xmath449 this concludes the proof that eq .
( [ eq : fixpointsymm ] ) admits a unique positive solution .
consider now existence and uniqueness of solutions of eq .
( [ eq : fixpointrec ] ) .
note that this is equivalent to proving that for every @xmath450 there exists a unique @xmath451 such that @xmath452 we know that @xmath108 is an increasing function , so @xmath453 is a decreasing function taking positive values .
the result follows by using monotonicity of @xmath430 . in order to prove eq .
( [ eq : claimderivative ] ) , notice that the lower bound follows from lemma [ ref : fgvaluesmonotony ] . for the upper bound , observe that @xmath454 . by evaluating it at @xmath112 , and using @xmath455 , @xmath456
, we get @xmath457 .
[ lem : rvariations ] let @xmath112 and @xmath114 be defined as per eq .
( [ lem : twelldefined ] ) .
then the function @xmath458 is strictly increasing on @xmath459 and strictly decreasing on @xmath460 .
similarly @xmath461 is strictly increasing on @xmath462 and strictly decreasing on @xmath463 .
recall that letting @xmath464 we have , for any @xmath465 , @xmath466 and @xmath467 . as a consequence @xmath468 , and
therefore , for all @xmath469 , @xmath470 recall that , by lemma [ ref : fgvaluesmonotony ] , @xmath368 .
further , as per the proof of lemma [ lem : twelldefined ] , @xmath471 is strictly decreasing with @xmath472 .
this immediately implies the claim for @xmath121 . the argument for @xmath473 is completely analogous .
we write the derivative of @xmath474 with respect to @xmath389 : @xmath475 the claim follows again from @xmath476 , and using the properties of @xmath477 already discussed in the proof of lemma [ lem : twelldefined ] .
[ lem : convergencefixedpointsymm ] let the state evolution sequence @xmath478 defined by @xmath479 and @xmath480 for all @xmath232 .
then , for any law @xmath146 , there exist constants @xmath481 such that , for all @xmath232 @xmath482 we proved in lemma [ lem : twelldefined ] that @xmath483 is monotone increasing with @xmath484 if @xmath485 and @xmath486 if @xmath487 .
it follows that @xmath488 if @xmath489 and @xmath488 if @xmath490 .
hence @xmath491 convergence is exponentially fast , i.e. @xmath492 , since , by lemma [ lem : twelldefined ] @xmath493 .
this proves the first second inequality .
note that @xmath112 is the global maximum of @xmath458 and hence , in a neighborhood of @xmath112 , @xmath494 we state without proof the analogous result for the rectangular case .
the argument is exactly the same as for the symmetric case .
[ lem : convergencefixedpoint ] let the state evolution sequence @xmath495 be defined by the recursion [ eq : seqthetamu ] with the initial condition @xmath280 . for any law @xmath146 there exist constants @xmath496 such that , for all @xmath232 , @xmath497 our results are stated in terms of @xmath108 and @xmath109 , and depend on the law of @xmath242 . however , when @xmath498 , interestingly , two different phenomena occur .
first , our results can be stated independently of law of @xmath104 .
second , a phase transition occurs for a specific value of the signal - to - noise ratio @xmath328 .
this is stated formally below using the notion of uniform convergence introduced in definition [ def : uniformconvergence ] .
[ rk : sparseregimefg ] the following limits hold uniformly over the class @xmath125 of probability distributions on @xmath126 with second moment equal to @xmath380 , and over @xmath499 $ ] for any @xmath500 : @xmath501 further , again uniformly over @xmath125 , for any @xmath502 @xmath503 in order to prove eq .
( [ eq : dlimit ] ) note that , by taking first the expectation over @xmath383 in @xmath504 , we get @xmath505+x^2v^2\big[\phi(xv)-1\big]+xv\phi(xv)\big\}\equiv \e\{f(xv)\}\ , , \ ] ] where @xmath506+z^2\big[\phi(z)-1\big]+z\phi(z)$ ] .
note that @xmath507 and @xmath508 is bounded , whence @xmath509 which yields the desired uniform convergence of @xmath373 .
next recall that @xmath510 , cf .
( [ eq : fderivative ] ) and , by lemma [ ref : fgvaluesmonotony ] , @xmath511 is strictly convex .
we hence have , for all @xmath45 , @xmath512 \le \inf_{\mu_v\in\cp_{\eps}}\f_v(x)\le \sup_{\mu_v\in\cp_{\eps}}\f_v(x)\le \frac{1}{\delta}\sup_{\mu_v\in\cp_{\eps}}\big[\sqrt{\d_v(x+\delta)}-\sqrt{\d_v(x)}\big]\ , . \ ] ] the claim ( [ eq : flimit ] ) follows by taking the limit @xmath513 ( using eq .
( [ eq : dlimit ] ) ) followed by @xmath514 .
the expression of @xmath515 follows by taking the limit on the identity @xmath516 . in order to prove eq .
( [ eq : tepstozero ] ) , let @xmath517 denote the function on the right - hand side and assume by contradiction that there exists a sequence @xmath518 , probability measures @xmath519 such that @xmath520 for some @xmath45 . as shown in the proof of lemma [ lem : twelldefined ] , @xmath521 is monotone increasing . using the definition we have , for all @xmath94 large enough @xmath522 taking the limit @xmath231 , and using eq .
( [ eq : flimit ] ) , we get @xmath523 that yields a contradiction by the definition of @xmath524 .
hence @xmath525 .
the matching lower bound is proved in the same way . finally , the proof of eq .
( [ eq : sepstozero ] ) follows along the same lines . in this section
we prove lemma [ th : upperbounds ] . as mentioned before ,
the proof of lemma [ th : upperboundssymmetric ] is completely analogous and omitted . for @xmath526 $ ] ,
we define @xmath527 note that @xmath528}m_{\bx}(\mu)= m_{\bx}(\<\bv^+,\bvz\ > ) \ , .
\label{eq : sigmaplus } \ ] ] the function @xmath529 is lipschitz continuous with lipschitz constant @xmath380 ( namely @xmath530 ) .
hence , by gaussian isoperimetry , we have @xmath531 further we claim that @xmath532 is uniformly continuous for @xmath526 $ ] . indeed if @xmath533 realizes the maximum over @xmath534 ( with @xmath535 ) , we have @xmath536 recall that @xmath537 for some constants @xmath538 @xcite . hence , with probability at least @xmath539 we have , for all @xmath540 $ ] @xmath541 let @xmath542 $ ] be a grid . by the above uniform continuity , we have , with probability at least @xmath539 , @xmath543}\big|m_{\bx}(\mu ) - \om(\mu)\big| \le \sup_{\mu\in \ci_n } \big|m_{\bx}(\mu ) - \om(\mu)\big| + c''n^{-1/2}\ , . \ ] ] using eq .
( [ eq : isoperimetry ] ) and union bound over @xmath544 , we conclude that @xmath545 } \big|m_{\bx}(\mu ) - \om(\mu)\big| \ge t \big\ } & \le 2\ , n\ , \exp\big\{-\frac{n}{2}(t - c''n^{-1/2})^2\big\ } + c_2e^{-n / c_2 } \le cn \ , e^{-nt^2/4}\ , , \ ] ] where the last inequality holds for all @xmath546 with @xmath547 a suitable constant .
in particular , by borel - cantelli we have , almost surely and in expectation , @xmath548 } \big|m_{\bx}(\mu ) - \om(\mu)\big| = 0\ , .
\label{eq : mom } \ ] ] in order to upper bound @xmath549 , we apply vitale s extension of sudakov - fernique inequality ( see e.g. ( * ? ? ?
* theorem 1 ) and ( * ? ? ?
* theorem 1 ) for a quantitative version ) to the two processes @xmath550 , @xmath551 indexed by @xmath552 defined as follows : @xmath553 for independent random vectors @xmath554 .
it is easy to see that @xmath555 and @xmath556 ^ 2\big\ } & = \left \ { \e\cx(\bu_1,\bv_1 ) - \cx(\bu_2,\bv_2)\right \}^2 + \frac{2}{n } \big(1-\<\bu_1,\bu_2\>\<\bv_1,\bv_2\>\big)\ , , \\ \e\big\{\big[\cy(\bu_1,\bv_1)-\cy(\bu_2,\bv_2)\big]^2\big\ } & = \left \
{ \e\cy(\bu_1,\bv_1 ) - \cy(\bu_2,\bv_2)\right \}^2 + \frac{2}{n } \big(2-\<\bu_1,\bu_2\>-\<\bv_1,\bv_2\>\big)\ , . \ ] ] hence @xmath557 ^ 2\big\}\le \e\big\{\big[\cy(\bu_1,\bv_1)-\cy(\bu_2,\bv_2)\big]^2\big\}$ ] ( this follows from @xmath558 for @xmath559 $ ] ) .
we conclude that @xmath560 where last inequality holds for any @xmath561 , setting @xmath562 .
the maximum in the last expression is achieved for @xmath563 hence , by lemma [ lem : fginnerproducts ] , there exists a deterministic sequence @xmath564 independent of @xmath526 $ ] , such that @xmath565 for any @xmath561 and @xmath566 where we recall that @xmath504 .
we next fix @xmath567 , which is also the unique maximizer of @xmath568 , as shown in lemma [ lem : rvariations ] .
note that eq .
( [ eq : upperboundlimit ] ) is strictly concave in @xmath569 $ ] , with unique maximum at @xmath570 .
substituting in eq .
( [ eq : upperboundlimit ] ) , we get @xmath571}\om(\mu)&\le \sqrt{1-\vartheta_*^2/\beta}+ \sqrt{\alpha\d_v(\vartheta_*/\sqrt{\alpha})}\\ & = \r\rec(\vartheta_*,\alpha ) + \delta_n\ , , \label{eq : maxm } \ ] ] where the last equality follows from the identity @xmath572 , and from the equation @xmath573 with @xmath574 that holds by definition of @xmath575 . from eq .
( [ eq : sigmaplus ] ) , ( [ eq : mom ] ) and ( [ eq : maxm ] ) we finally get @xmath576}\om(\mu ) \\ & \le \r\rec(\vartheta _ * , \alpha ) = \max_{\vartheta\in\reals}\r\rec(\vartheta , \alpha ) \ , , \label{eq : lastupperboundsigma } \ ] ] which coincides with claim ( [ eq : upperboundgeneralproposition ] ) . next reconsidering eq .
( [ eq : upperboundlimit ] ) with @xmath577 , we see that since the right - hand side is strictly concave in @xmath526 $ ] , we can strengthen eq .
( [ eq : maxm ] ) to @xmath578 for some @xmath579 .
we call @xmath580 . by eq .
( [ eq : sigmaplus ] ) and ( [ eq : mom ] ) we have , almost surely , @xmath581 we then use eq .
( [ eq : strictconvm ] ) to deduce that @xmath582 this implies immediately eq . ( [ eq : upperboundnninnerproductnonsymdense ] ) with @xmath583 , since ( as shown above ) @xmath575 , and @xmath584 . in this section
we prove lower bounds on the non - negative eigenvalue ( singular value ) that follows from the analysis of the amp algorithm , namely theorem [ th : lowerboundsampsymmetric ] for symmetric matrices and theorem [ th : lowerboundsampnonsymmetric ] for rectangular matrices .
the proofs are very similar in the two cases , hence we will provide details only in the case of rectangular matrices , and limit ourselves to pointing out differences arising in the symmetric setting . define @xmath585 and observe , using [ eq : ampnonsym ] , @xmath586 where @xmath587 the last step follows from triangular inequality . by proposition [ prop : segeneral ]
applied to @xmath588 , we have , almost surely , @xmath589 and therefore @xmath590 by applying the same proposition to @xmath591 we have @xmath592 further , letting @xmath593 we get @xmath594 where the last equality follows from stein s lemma @xcite .
using together eq .
( [ eq : vp-2norm ] ) and eq .
( [ eq : vp-0norm ] ) , we get @xmath595 using eq .
( [ eq : u-2norm ] ) and eq .
( [ eq : limbt ] ) in the upper bound ( [ eq : e - upperbound ] ) , we get @xmath596 finally , substituting this result together with eq .
( [ eq : gu ] ) and ( [ eq : limbt ] ) in eq .
( [ eq : rt ] ) , we obtain @xmath597 the claim ( [ eq : limamprayleighnonsym ] ) follows by lemma [ lem : convergencefixedpoint ] .
consider next eq .
( [ eq : ampscalprodurect ] ) .
we have @xmath598 where the second equality follows by applying proposition [ prop : segeneral ] to @xmath599 ( for the numerator ) and using eq .
( [ eq : u-2norm ] ) ( for the denominator ) . finally , the claim ( [ eq : ampscalprodurect ] ) follows by taking @xmath252 , and using lemma [ lem : convergencefixedpoint ] .
the proof of claim ( [ eq : ampscalprodvrect ] ) follows by the same argument and we omit it .
the proof in the symmetric case is very similar to the one for rectangular matrices , see theorem [ th : lowerboundsampnonsymmetric ] .
we limit ourselves to sketching the first steps .
we have , using [ eq : ampsym ] , @xmath600 and we are left with a term @xmath601 which we treat similarly to @xmath602 of theorem [ th : lowerboundsampnonsymmetric ] .
namely , by using proposition [ prop : segeneral ] and proposition [ lem : diffgoestozero ] , we prove that @xmath603 in addition , it follows from proposition [ prop : segeneral ] that @xmath604 this terminates the proof sketch . in this section
we prove that the least favorable vectors @xmath15 are asymptotically of the following form : @xmath605 for all @xmath606 , and @xmath607 otherwise , for some support @xmath608 $ ] .
further , we characterize the least favorable size of the support @xmath609
. the proofs proceed by analyzing the expression in theorem [ th : mainsym ] and applying strong duality to a certain linear program over probability distributions , that is related to the function @xmath610 .
we start with some preliminary facts and definitions in section [ subsec : preliminary ] .
the key step is to reduce ourselves to two points mixtures : this is achieved in section [ subsec : reduction ] .
finally , in sections [ subsec : worstsymm ] and [ subsec : worstrec ] , we use these results to prove theorems [ th : fworstcase ] and [ th : fworstcaserec ] .
since the proof of theorem [ th : fworstcaserec ] is completely analogous to the one of theorem [ th : fworstcase ] , we will limit ourselves to mentioning the main differences . for @xmath611 $ ] and @xmath612 , we define the 2-points mixture @xmath613 in particular , when @xmath614 ( and hence the above distribution has second moment equal to @xmath380 ) , we write @xmath615 . we also write with a slight abuse of notation @xmath616 instead of @xmath394 when @xmath617 .
explicitly @xmath618 even more explicitly @xmath619 we will also adopt the shorthand @xmath620 when @xmath621 .
we wil next establish two calculus lemmas that are useful for the following .
[ lemma : twosols ] for any given @xmath622 , the equation @xmath623 in the unknown @xmath624 has at most two solutions @xmath625 .
let @xmath626 denote the left - hand side of eq .
( [ eq : simpleeq ] ) .
then @xmath627 if @xmath628 , then we conclude that @xmath629 for all @xmath630 and hence the equation @xmath631 has at most one positive solution . if on the other hand @xmath632 , then @xmath633 for @xmath634 and @xmath635 for @xmath636 .
it follows that the equation @xmath631 has at most one solution in @xmath637 $ ] and at most one in @xmath638 .
[ lemma : elementaryinequality ] let @xmath639 be defined as @xmath640 .
then , for every @xmath641 , we have @xmath642 by simple calculus , we get @xmath643 , and the derivatives @xmath644 let us further recall the inequalities ( valid for @xmath451 ) @xmath645 which immediately imply for all @xmath451 @xmath646 therefore the left - hand side of eq .
( [ eq : inequality ] ) is always strictly positive .
consider the right - hand side . by consulting special values of the normal distribution
, we see that @xmath647 . by a change of variables we know that @xmath648 or , equivalently @xmath649\big\}\ , . \ ] ] since the term in curly brackets is decreasing in @xmath389 , and is negative at @xmath650 , we have @xmath651 for all @xmath652 . therefore the right - hand side of eq .
( [ eq : inequality ] ) is non - positive for @xmath653 .
this proves the claim for @xmath652 , and we will assume hereafter @xmath654 . next notice that @xmath655 for @xmath656 .
therefore , by taylor expansion and intermediate value theorem , we get , for @xmath654 , @xmath657 the right - hand side is negative for @xmath658 where @xmath659 in particular @xmath660 , and @xmath661 .
it follows that the right - hand side of eq .
( [ eq : inequality ] ) is non - positive for @xmath662
. we will therefore restrict ourselves to considering @xmath663 .
note that our claim can be equivalently written as @xmath664 we will next develop , for @xmath665 , a lower bound on the left - hand side , to be denoted by @xmath666 , and an upper bound on the right - hand side , to be denoted by @xmath667 and prove that @xmath668 . for the left hand side note that @xmath669 for @xmath665 and hence , again by taylor expansion @xmath670 for the right hand side note that @xmath671 .
further @xmath672 is monotone decreasing .
we therefore define @xmath673 and obtain the upper bound @xmath674 .
hence @xmath675 it is a straightforward exercise to check that indeed @xmath676 thus completing the proof .
the main theorem of this section shows that @xmath394 is minimized by probability measures @xmath146 that are mixture of at most two point masses .
[ thm : leastfavorablef ] fix @xmath372 .
then for any random variable @xmath242 with probability distribution @xmath677 , we have @xmath678 } \f_{\eps } ( x)\ , . \ ] ] the proof of this theorem is presented at the end of the section . before getting to it
, we ll introduce a related problem .
note that @xmath679 where @xmath680 is the value of a constrained optimization problem : @xmath681 here it is understood that @xmath682 if this problem is unfeasible .
[ lemma : constrained ] let @xmath683 be such that the problem ( [ eq : constrainedproblem ] ) is feasible . then there exist @xmath684 such that @xmath685 is feasible and @xmath686 $ ] , such that , letting @xmath687 , we have @xmath688 by a rescaling of the objective function , and letting @xmath689 , we can rewrite the problem ( [ eq : constrainedproblem ] ) as @xmath690 now , for fixed @xmath691 , let @xmath692 and write @xmath693 with @xmath96 a measure on @xmath126
. then we can rewrite the optimization problem as the following ( with decision variable @xmath96 ) @xmath694 the corresponding value is @xmath695 .
note that each of the functions @xmath696 , @xmath697 , @xmath698 is bounded and lipschitz continuous , with a finite limit as @xmath699 .
this implies that the value @xmath695 is achieved by a measure @xmath700 on the completed real line @xmath701 $ ] , with total mass @xmath702 .
indeed the family of normalized distributions on @xmath701 $ ] is tight and both the objective and the constraints are continuous in the weak topology .
hereafter , we shall assume this holds .
functions on @xmath703 are extended by continuity to @xmath704 . by introducing lagrange multipliers , we obviously have , for any @xmath705 @xmath706 where the infimum is over measures @xmath96 on @xmath126 . by strong duality
( which follows , for instance , from the kneser - kuhn minimax theorem @xcite , see also ( * ? ? ?
* theorem a.1 ) ) , there exists . in the present case
, it is not had to show that , letting @xmath707 denote the right hand side of eq .
( [ eq : duality ] ) , the @xmath708 $ ] is actually achieved at finite @xmath709 , by showing that the sequence must remain bounded and using standard compactness arguments . ]
@xmath710 such that the above holds with equality .
note that for such choice @xmath711 for all @xmath712 $ ] , because otherwise the infimum is @xmath713 . under this condition , the infimum term in eq .
( [ eq : duality ] ) is zero , and hence we must have @xmath714\ , , \label{eq : gammastar } \ ] ] because otherwise we could increase the lower bound eq .
( [ eq : duality ] ) by increasing @xmath715 .
further , since the right - hand side is an analytic function of @xmath716 , the infimum in eq .
( [ eq : gammastar ] ) is achieved on a finite set @xmath717 $ ] , and the minimizer @xmath700 of problem ( [ eq : muproblem ] ) has support @xmath718 because otherwise the infimum in the lower bound ( [ eq : duality ] ) would not be achieved .
next we claim that @xmath719 for some finite @xmath720 .
indeed , let @xmath721 .
it follows from eqs .
( [ eq : f ] ) and ( [ eq : g ] ) that @xmath722 assume @xmath723 .
we then have @xmath724 for some finite @xmath725 if and only if @xmath726 by lemma [ lemma : twosols ] this has at most two solutions @xmath727 .
if on the other hand @xmath728 , then the above equation reduces to @xmath729 which has at most one solution . in both cases , at most one solution call it @xmath730 is a local minimum of @xmath731 . we conclude that @xmath719 , and therefore the value of the problem ( [ eq : muproblem ] ) is achieved by a measure of the form @xmath732 the three constraints imply the following relations @xmath733 and the value is @xmath734 the proof is completed by the change of variables @xmath735 , @xmath736 , @xmath737 , @xmath738 , @xmath739 . with these substitutions eq .
( [ eq : yproof ] ) yields ( [ eq : yf ] ) , and eq .
( [ eq : cfproof ] ) yields ( [ eq : calf ] ) .
we are now in position to prove theorem [ thm : leastfavorablef ] , that is the main result in this section . by lemma [ lemma : constrained ] and eq .
( [ eq : fcf ] ) , we have , for any @xmath740 , @xmath741 where the infimum is over @xmath686 $ ] , @xmath742 $ ] , @xmath743 .
our claim is equivalent to saying that the infimum on the right hand side is achieved when @xmath744 .
since @xmath451 is given , we will can regard the right - hand side as a function of @xmath745 and @xmath321 , and substitute @xmath746 .
we then define the function @xmath747 more explicitly @xmath748 which needs to be optimized over @xmath742 $ ] , and @xmath749 $ ] .
our claim is equivalent to saying that the minimum can not be in the interior of this domain .
since @xmath383 is analytic in the mentioned domain , a minimum in the interior must satisfy @xmath750 .
simple calculus shows that these two conditions are equivalent respectively to : @xmath751 & = \big[x^2+\eps\ , b(w)\big]\,\eps\big[b'(w)+\phi(w)\big]\ , , \\
2b(w ) \,\big[x^2+\frac{1}{2}+\eps \big(b(w)+\phi(w)-\frac{1}{2}\big)\big ] & = \big[x^2+\eps\ , b(w)\big]\,\big[b(w)+\phi(w)-\frac{1}{2}\big]\ , . \ ] ] taking the ratio of these equations , we obtain the necessary condition @xmath752 or equivalently @xmath753 lemma [ lemma : elementaryinequality ] establishes that this equation does not have any solution in @xmath754 and hence @xmath755 does not have stationary points in domain @xmath756 $ ] , @xmath749 $ ] .
this finishes our proof .
we conclude with a corollary of theorem [ thm : leastfavorablef ] .
( figure [ fig : symworstcase ] provides an illustration of the argument used in the proof . )
[ coro : leastfavorablet ] fix @xmath757 . then for any random variable @xmath242 with probability distribution @xmath677
, we have @xmath758 } \t_{\eps } ( \beta)\ , .
\label{eq : worstcaset } \ ] ] further , for any @xmath759 , the infimum on the right - hand side is achieved at some @xmath760 $ ] .
assume the claim ( [ eq : worstcaset ] ) does not hold .
then there exists @xmath761 such that @xmath762 for all @xmath742 $ ] .
now , on the one hand , by definition we have @xmath763 on the other hand , by theorem [ thm : leastfavorablef ] , there exists @xmath764 $ ] such that @xmath765 for @xmath766 . using this fact , the contradiction assumption @xmath767 , and the fact that @xmath768 is strictly decreasing on @xmath754 as shown in the proof of lemma [ lem : twelldefined ] ,
we get @xmath769 we therefore reached a contradiction , which proves the claim ( [ eq : worstcaset ] ) . in order to prove that the infimum is achieved at some @xmath770 $ ] , note that @xmath771 is clearly continuous and , by lemma [ rk : sparseregimefg ] ,
@xmath772 it is therefore sufficient to show that @xmath773 is decreasing for @xmath321 small enough . by an argument similar to the above , this follows if we show that @xmath774 is decreasing for @xmath775 and @xmath321 small enough . indeed using the definition ( [ eq : fepsdefinition ] ) and recalling that @xmath776 as @xmath699 , we get , for every fixed @xmath451 @xmath777 which is of course decreasing in @xmath321 for @xmath778 $ ] with @xmath779
. first let @xmath780 $ ] .
we then set @xmath781 and
@xmath782 then of course @xmath783 converges in empirical distribution to @xmath685 and , by theorem [ th : mainsym ] @xmath784 with @xmath785 the only non - negative solution of @xmath786 . by lemma [ rk : sparseregimefg ] ( cf .
( [ eq : flimit ] ) , ( [ eq : tepstozero ] ) ) , we have @xmath787 , and hence @xmath788 the claim ( [ eq : worstcaselowsnr ] ) then follows by replacing @xmath321 , by sequence @xmath789 with @xmath790 sufficiently slowly .
the limit vanishes in this case as well by a standard argument .
next consider the claim ( [ eq : worstcasehighsnr ] ) .
we let @xmath791 be the value achieving the infimum in eq .
( [ eq : worstcaset ] ) , which exists by corollary [ coro : leastfavorablet ] .
it is obvious ( by another application of theorem [ th : mainsym ] ) that equality holds for the stated choice of @xmath166 .
assume by contradiction that the inequality ( [ eq : worstcasehighsnr ] ) does not hold for some sequence @xmath792 .
then , by tightness , there exists a subsequence along which the limit on the left hand side exists , and that converges in empirical distribution to a certain probability measure @xmath677 . hence , using theorem [ th : mainsym ] , it follows that ( using the definition of @xmath112 ) @xmath793}\t_{\eps}(\beta)\ , . \ ] ] this contradicts corollary [ coro : leastfavorablet ] , hence proving our claim .
the proof of theorem [ th : fworstcaserec ] is very similar to the proof of theorem [ th : fworstcase ] , and therefore we will only sketch the first steps .
first if @xmath794 , we set @xmath795 @xmath796 then @xmath797 converges in empirical distribution to @xmath685 and , by theorem [ th : mainrec ] , @xmath798 with @xmath799 for @xmath617 is given by definition [ def : functionsfgrts ] , i.e. is the only positive solution @xmath389 of @xmath800 by lemma [ rk : sparseregimefg ] ( cf .
( [ eq : flimit ] ) , ( [ eq : sepstozero ] ) ) , we have @xmath801 , and hence @xmath802 the claim follows by taking @xmath803 slowly enough . next consider @xmath804 . by the same argument as in corollary [ coro : leastfavorablet ]
, we have , for any @xmath677 , @xmath805 } \s_{\eps } ( \beta,\alpha)\ , .
\label{eq : worstcases } \ ] ] further , for any @xmath759 , the infimum on the right - hand side is achieved at some @xmath760 $ ] .
we then take @xmath806 . assuming that the claim ( [ eq : worstcaserecs ] ) is false , we can construct by the same tightness argument used in the previous section , a probability distribution @xmath146 , such that @xmath807 .
this contradicts eq .
( [ eq : worstcases ] ) , which proves our claim .
this work was partially supported by the nsf grant ccf-1319979 and the grants afosr / darpa fa9550 - 12 - 1 - 0411 and fa9550 - 13 - 1 - 0036 .
in this appendix we characterize the high - dimensional behavior of amp as per proposition [ prop : ampsym ] and proposition [ lem : diffgoestozerosymm ] .
the analogous results for rectangular matrices ( namely , propositions [ prop : segeneral ] and [ lem : diffgoestozero ] ) follow from very similar arguments which we omit here .
it is convenient to first state two simple facts .
the first one allows to control small perturbations of a given iterative scheme .
[ lemma : perturbedamp ] let @xmath28 be as in the statement of proposition [ prop : ampsym ] , and the sequences @xmath808 , @xmath809 be defined by the recursions @xmath810 where @xmath811 and @xmath812 .
assume that @xmath813 , @xmath814 and , for every @xmath815 , we have the following , almost surely 1 .
2 . @xmath817 .
@xmath818 is lipschitz continuous with bounded lipschitz constant .
namely , there exists @xmath819 independent of @xmath94 such that @xmath820 for all @xmath821 . then , for all @xmath822 we have @xmath823 the proof is immediate by induction over @xmath50
. we will prove eq .
( [ eq : perturbedamp ] ) : eq .
( [ eq : boundedamp ] ) follows by a similar argument . the case @xmath824 holds by assumption . in order to prove the induction step , note that @xmath825 with probability larger that @xmath826 for some @xmath827 @xcite . by triangular inequality @xmath828 where the second inequality holds with probability at least @xmath826 .
the induction claim follows by dividing the above inequality by @xmath225 .
the second remark allows to establish limit results as in proposition [ prop : ampsym ] , once they have been established for a perturbed sequence .
[ lemma : perturb ] assume that the sequences of vectors @xmath829 , @xmath830 satisfy @xmath831 and further assume @xmath832 be such that @xmath833 .
if @xmath834 exists for some pseudo - lipschitz function @xmath835 , then @xmath836 using the pseudo - lipschitz property of @xmath835 , and cauchy - schwartz , we get @xmath837 by eqs .
( [ eq : perturbassumption1 ] ) and ( [ eq : perturbassumption2 ] ) , @xmath838 . using this fact together with the other assumptions , we get from the last inequality @xmath839 which proves our claim .
the proof consists in modifying the amp sequence @xmath249 as to reduce ourselves to the setting of @xcite .
the first step consists in introducing a sequence @xmath840 defined by @xmath841 , @xmath842 and letting , for all @xmath220 , @xmath843 where @xmath844 . the relation between this recursion and the original one is quite direct : they differ only by a normalization factor .
[ lemma : vvsw ] let @xmath845 be defined per eq .
( [ eq : simplifiedamp ] ) and @xmath249 be the amp sequence , as per ( [ eq : ampsym ] ) .
then , for all @xmath232 we have @xmath846 the proof is by induction over the number of iterations .
let us first assume that it holds for all iterations until @xmath50 , and prove it for iteration @xmath847 .
multiplying eq .
( [ eq : simplifiedamp ] ) by @xmath848 , we get @xmath849 note that the induction hypothesis implies @xmath850 for some constant @xmath851 , and hence @xmath852 by the same argument and using @xmath853 , we get @xmath854 using eqs .
( [ eq : firstwv ] ) and ( [ eq : secondwv ] ) in eq .
( [ eq : inductionwv ] ) , we obtain @xmath855 the induction step is completed by comparing this with ( [ eq : ampsym ] ) . the base case follow easily by a similar argument . as a second step ,
we introduce a sequence @xmath856 defined as follows .
first , we let @xmath857 , @xmath858 be scalars given by @xmath859 with initial conditions @xmath860 and @xmath861 .
note that by cauchy - schwartz @xmath862 and @xmath863 , whence @xmath864 for all @xmath50 .
further , since @xmath865 with probability @xmath866 , we also have @xmath867 , @xmath868 , whence @xmath869 for all @xmath50 . using these quantities ( and recalling that @xmath870 with @xmath871 , i.i.d . for @xmath72 ) , we define @xmath872 as usual , here @xmath873
is interpreted as the component - wise application of @xmath874 .
the initial condition is @xmath875 .
this iteration is in the form of ( * ? ? ?
* theorem 1 ) ( and analogous to ( * ? ? ? * theorem 4 ) ) , which implies immediately the following .
[ lemma : stateevolutions ] for any @xmath232 and any pseudo - lipshitz function @xmath876 , we have , almost surely @xmath877 where expectation is with respect to @xmath443 independent of @xmath242 .
the sequences @xmath856 and @xmath845 are in fact closely related as we show next .
[ lemma : stateevolutionw ] for any @xmath232 and any pseudo - lipshitz function @xmath876 , we have @xmath878 where expectation is with respect to @xmath443 independent of @xmath242 .
define @xmath879 . then eq .
( [ eq : simplifiedamp ] ) implies immediately @xmath880 next note that our claim is equivalent to the following holding for every pseudo - lipshitz @xmath835 and every iteration number @xmath881 : @xmath882 we prove this by induction over the iteration number .
assume that the claim indeed holds for all @xmath883 .
then comparing eq .
( [ eq : iterations ] ) and eq .
( [ eq : iterationstilde ] ) we obtain by lemma [ lemma : perturbedamp] that eq .
( [ eq : svsstilde ] ) holds for all @xmath884 provided we can prove that @xmath885 for all @xmath886 .
now we have @xmath887 now , using @xmath888 in eq .
( [ eq : seinduction ] ) we get for all @xmath886 , almost surely @xmath889 in other words @xmath890 almost surely . using again eq .
( [ eq : seinduction ] ) we have @xmath891 almost surely . therefore ,
since @xmath892 is finite for all @xmath50 , we get @xmath893 for some constant @xmath894 bounded uniformly in @xmath94 . finally , fix @xmath45 and let @xmath895 then @xmath896\\ & \le \frac{1}{n}\sum_{i=1}^n\big[\psi_{\delta}(\bs^{\ell}_i+\mu_{\ell}(\bvz)_i\sqrt{n}+\delta)- \psi_{\delta}(\bts^{\ell}_i+\mu_{\ell}(\bvz)_i\sqrt{n})\big]\\ & \le \frac{1}{n}\sum_{i=1}^n\big[\psi_{\delta}(\bs^{\ell}_i+\mu_{\ell}(\bvz)_i\sqrt{n}+\delta)- \psi_{\delta}(\bs^{\ell}_i+\mu_{\ell}(\bvz)_i\sqrt{n } ) + \frac{1}{\delta}\|\bs^{\ell}_2-\bts^{\ell}_i|\big]\\ & \le \frac{1}{n}\sum_{i=1}^n\big[\psi_{\delta}(\bs^{\ell}_i+\mu_{\ell}(\bvz)_i\sqrt{n}+\delta)- \psi_{\delta}(\bs^{\ell}_i+\mu_{\ell}(\bvz)_i\sqrt{n})\big ] + \frac{1}{\delta\sqrt{n}}\|\bs^{\ell}-\bts^{\ell}\|_2\ , . \ ] ] taking @xmath231 and using eqs .
( [ eq : svsstilde ] ) , ( [ eq : seinduction ] ) , we conclude that @xmath897 and since @xmath383 has a density with respect to lebesgue measure , this implies , by letting @xmath514 , @xmath898 . a lower bound
is obtained by a similar argument yielding @xmath899 and hence @xmath900 . by eq .
( [ eq : bounddelta ] ) we have @xmath901 and hence eq .
( [ eq : svsstilde ] ) holds for @xmath902 .
finally , eq . ( [ eq : seinduction ] ) follows for @xmath902 by lemma [ lemma : perturb ] , lemma [ lemma : stateevolutions ] and eq .
( [ eq : svsstilde ] ) ( for @xmath902 ) . finally , the proof of proposition [ prop : ampsym ] follows immediately from lemma [ lemma : stateevolutionw ] , using lemma [ lemma : vvsw ] .
indeed , by applying [ lemma : stateevolutionw ] to @xmath903 , we get , almost surely @xmath904 hence , for any pseudo - lipshitz function @xmath835 , almost surely @xmath905 we conclude by noting that by comparison of eq .
( [ eq : symmetricstateevolution ] ) with eqs .
( [ eq : stateevolutionsplit1 ] ) and ( [ eq : stateevolutionsplit2 ] ) it follows that @xmath906 for all @xmath50 . finally the claim @xmath247 , follows by a standard argument already used above .
namely , we use the bounds @xmath907 , with the definition in eq .
( [ eq : psidelta ] ) , apply the previous result to the lipschitz functions @xmath908 , @xmath909 and eventually let @xmath514 . the proof is based on a version of state evolution that describes the asymptotic joint distribution of @xmath228 , @xmath910 for two distinct times @xmath911 . for this purpose ,
we define a function @xmath912\times\reals_+\times \reals_+\to\reals$ ] as follows @xmath913 where expectation is with respect to the centered gaussian vector @xmath914 with @xmath915 , @xmath916 , independent of @xmath242 .
let the state evolution sequence @xmath917 be given as per eq .
( [ eq : symmetricstateevolution ] ) , and define recursively @xmath918 by letting @xmath919 with initial condition @xmath920 and , for @xmath921 , @xmath922 then we have the following extension of state evolution .
[ lemma : twotimes ] with the above definitions , let @xmath923 be a pseudo - lipschitz function .
then , for any @xmath924 , we have , almost surely @xmath925 where expectation is with respect to the centered gaussian vector @xmath926 with @xmath927 , @xmath928 , independent of @xmath242 .
very similar statements were proven , for instance in ( * ? ? ?
* theorem 4.2 ) or ( * ? ? ?
* lemma c1 ) .
the construction is always the same , and we will only sketch the first steps .
thanks to lemma [ lemma : stateevolutionw ] , it is sufficient to prove that , for @xmath856 defined per eq .
( [ eq : iterations ] ) , we have @xmath929 where @xmath930 is a centered gaussian vector with @xmath931 , @xmath932 , @xmath933 . in order to prove the last claim
, we fix a maximum time @xmath934 , and consider all @xmath935 .
we then define @xmath936 , @xmath937 that we can think of either as a vector of lenfth @xmath94 , with entries in @xmath938 , or as a matrix with dimensions @xmath939 . with the last interpretation in mind ,
@xmath940 is defined as a matrix whose first @xmath847 columns ar @xmath941 , @xmath942 , , @xmath943 , and the others vanish , namely @xmath944\ , .
\ ] ] define @xmath945 by letting @xmath946 then it is easy to see that eq .
( [ eq : iterations ] ) implies @xmath947 for a certain sequence of matrices @xmath948 .
the proof then follows by applying ( * ? ? ?
* theorem 1 ) to @xmath949 .
the next lemma provides the basic tool for applying the state evolution method to prove our claim .
[ lemma : twotimestinfty ] let @xmath950 be defined as above using the two times state evolution recursion ( [ eq : twotimesse ] ) . then @xmath951 before proving this lemma , we state a useful general fact ( which appeared already in specific forms in @xcite .
[ lemma : convexitytwotimes ] let @xmath952 be a borel function , @xmath953 random variables , and @xmath954 a probability distribution such that under @xmath954 @xmath955 is a centered gaussian vector independent of @xmath956 , with covariance given @xmath957 and @xmath958 .
assume @xmath959 and define @xmath960 then @xmath961 is non - decreasing and convex on @xmath962 $ ] .
further , unless @xmath963 is affine in @xmath389 , it is strictly convex . finally assuming @xmath731 is weakly differentiable , and denoting by @xmath964 its derivative with respect to the first argument , we have @xmath965 ^ 2\}\ ,
: derivativetwotimes}\end{aligned}\ ] ] first consider the case of @xmath966 independent of the second argument .
let @xmath967 be the stationary ornstein uhlenbeck process with covariance @xmath968 .
then @xmath969 then we have the spectral representation ( for @xmath970 and @xmath971 , @xmath972 the @xmath881-th eigenfunction of the ornstein uhlenbeck generator @xmath973 whence the @xmath974 is non - decreasing and convex .
strict convexity follows since @xmath975 for some @xmath976 as long as @xmath977 is non - linear .
finally , if @xmath731 depends on its second argument as well , we have @xmath978 , with @xmath979 .
using independence of @xmath955 and @xmath956 , the previous proof applies to @xmath980 for almost every @xmath956 and , by linearity , to @xmath981 .
equation ( [ eq : derivativetwotimes ] ) follows by writing @xmath982
^ 2\big\}\ , , \ ] ] with @xmath983 .
the claim follows by using the representation @xmath984 , @xmath985 , with @xmath986 independent standard normal , @xmath987 , @xmath988 , and taylor expanding the right hand side in @xmath989 .
we are now in position to prove lemma [ lemma : twotimestinfty ] .
recall that @xmath990 , cf .
lemma [ lem : convergencefixedpointsymm ] .
letting @xmath991 , we define @xmath992 , i.e. @xmath993 with @xmath914 a centered gaussian vector with @xmath994 and @xmath995 . by lemma [ lemma : convexitytwotimes ] , the function @xmath996 is strictly convex and monotone increasing in @xmath962 $ ] .
further we have @xmath997 and , for @xmath443 , @xmath998 note that @xmath999 where the last inequality follows since @xmath1000 , and applying stein s lemma to the second term .
we therefore have @xmath1001 and therefore , by convexity , @xmath1002 for all @xmath1003 .
now , for ease of notation , let @xmath1004 .
note that @xmath1005 $ ] for all @xmath1006 $ ] : indeed , for @xmath1007 , the random variables @xmath1008 and @xmath1009 are non - decreasing functions of positively correlated ones , and hence are positively correlated . therefore @xmath1010 $ ] for all @xmath50 .
assume by contradiction that @xmath1011 does not converge to @xmath380 , and let @xmath1012 .
let @xmath1013 be a subsequence with @xmath1014 .
since @xmath1015 , @xmath1016 is continuous and @xmath1017 is non - decreasing , we have @xmath1018 this contradicts the previous remark that @xmath1002 for all @xmath1019 , and hence proves the claim that @xmath1020 .
we are now in position to prove our claim ( [ lem : diffgoestozerosymm ] ) .
first note that by triangular inequality it is sufficient to consider the case @xmath1021 . using lemma [ lemma : twotimes ] for @xmath1022 and @xmath1023 , we get , almost surely @xmath1024 since by lemma [ lem : convergencefixedpointsymm ] the sequence @xmath236 converges to a finite limit as @xmath252 , we have @xmath1025 . hence taking the limit @xmath252 in the last expression and using lemma [ lemma : twotimestinfty ]
, we obtain the desired result .
for the sake of clarity , we will note the dimension index @xmath94 for a matrix @xmath1026 , distributed according to the [ eq : symmetricmodel ] . in order to prove eq .
( [ eq : lambdaplussim ] ) ( i.e. @xmath1027 almost surely ) , we need to prove : @xmath1028 = 1~~\text{and}~~ \prob\left [ \limsup_{n \to \infty } \lambda^+(\bx_n ) \leq \r\sym(\t_v(\beta ) ) \right ] = 1~. \ ] ] * theorem [ th : lowerboundsampsymmetric ] states that there exists a deterministic sequence @xmath1029 such that @xmath1030 and @xmath1031 = 1~~.\ ] ] it follows , using @xmath1032 , and taking the intersection of these events for @xmath1033 , that @xmath1034 = 1~~.\ ] ] * since the function @xmath1035 is 1-lipschitz continuous , then using the upper bound of lemma [ th : upperboundssymmetric ] and gaussian isoperimetry , for any @xmath1036 we have , with probability at least @xmath1037 , @xmath1038 taking @xmath1039 , with probability at least @xmath1040 , @xmath1041 hence @xmath1042 almost surely by borel - cantelli .
this concludes the proof of eq .
( [ eq : lambdaplussim ] ) . equation ( [ eq : innerproductsim ] ) follows immediately from lemma [ th : upperboundssymmetric ] since @xmath1043 , and we know that the sequence @xmath1044 converges almost surely to to @xmath1045 . in order to prove the limit behavior as @xmath1046 of eqs .
( [ eq : rlimitepszerosym ] ) and ( [ eq : flimitepszerosym ] ) we refer to lemma [ rk : sparseregimefg ] in section [ sec : preliminaryproof ] that establish the limit behavior of functions of interest @xmath1047 uniformly over the class of probability distributions @xmath125 . we know ,
thanks to definition [ def : uniformconvergence ] , lemma [ rk : sparseregimefg ] , and uniform continuity on the interval @xmath1048 $ ] of the square function @xmath1049 , and on @xmath370 of @xmath1050 that for any @xmath1051 , one can find @xmath1052 such that for any @xmath1053 and @xmath1054 , we have , for @xmath1055 , @xmath1056 and @xmath1057| \leq \kappa~~.\ ] ] this proves uniform convergence of @xmath1058 to @xmath1059 and of @xmath1060 to @xmath1061 . since we have @xmath1062 and @xmath1063 the result
is proved . in order to prove theorem [ th : mainrec ] we proceed as for theorem [ th : mainsym ] .
we consider a sequence of random matrices @xmath1064 of size @xmath1065 , generated according to the [ eq : spikedmatrix ] .
we use lemma [ th : upperbounds ] and gaussian isoperimetry for the @xmath380-lipschitz function @xmath1066 to conclude that with probability at least @xmath1040 , @xmath1067 this proves , using borel - cantelli lemma , that @xmath1068 almost surely .
by theorem [ th : lowerboundsampnonsymmetric ] there exists a deterministic sequence @xmath1069 such that @xmath1070 = 1~~. \ ] ] since @xmath1071 , and by taking the intersection over @xmath1033 , we get @xmath1072 almost surely .
this concludes the proof of eq .
( [ eq : mainrrec ] ) , i.e. @xmath1073 .
samuel burer , kurt m. anstreicher , and mirjam dur , _ the difference between 5@xmath1077 5 doubly nonnegative and completely positive matrices _ , linear algebra and its applications * 431(9 ) * ( 2009 ) , 15391552 . florent benaych - georges , alice guionnet , and mylne mada , _ large deviations of the extreme eigenvalues of random deformations of matrices _ , probability theory and related fields * 154 * ( 2012 ) , no . 3 - 4 , 703751 .
mireille capitaine , catherine donati - martin , and delphine fral , _ the largest eigenvalues of finite rank deformation of large wigner matrices : convergence and nonuniversality of the fluctuations _ , the annals of probability * 37 * ( 2009 ) , no . 1 , 147 . mireille capitaine , catherine donati - martin , delphine fral , maxime fvrier , et al . , _
free convolution with a semicircular distribution and eigenvalues of spiked deformations of wigner matrices _ , electron .
j. probab * 16 * ( 2011 ) , no .
64 , 17501792 .
mireille capitaine , catherine donati - martin , and delphine fral , _ central limit theorems for eigenvalues of deformations of wigner matrices _ , annales de linstitut henri poincar , probabilits et statistiques , vol .
48 , institut henri poincar , 2012 , pp .
107133 .
michel journe , yurii nesterov , peter richtrik , and rodolphe sepulchre , _ generalized power method for sparse principal component analysis _
, the journal of machine learning research * 11 * ( 2010 ) , 517553 .
yoshiyuki kabashima , florent krzakala , marc mzard , ayaka sakata , and lenka zdeborov , _ phase transitions and sample complexity in bayes - optimal matrix factorization _
, arxiv preprint arxiv:1402.1298 ( 2014 ) .
alexey pavlov , valeri a makarov , ioulia makarova , and fivos panetsos , _ sorting of neural spikes : when wavelet based methods outperform principal component analysis _
, natural computing * 6 * ( 2007 ) , no . 3 , 269281 .
alessandro pizzo , david renfrew , and alexander soshnikov , _ on finite rank deformations of wigner matrices _ , annales de linstitut henri poincar , probabilits et statistiques , vol .
49 , institut henri poincar , 2013 , pp .
6494 .
xing sun and andrew b nobel , _ on the maximal size of large - average and anova - fit submatrices in a gaussian random matrix _
, bernoulli : official journal of the bernoulli society for mathematical statistics and probability * 19 * ( 2013 ) , no . 1 , 275 . charles stein , _ a bound for the error in the normal approximation to the distribution of a sum of dependent random variables _ , proceedings of the sixth berkeley symposium on mathematical statistics and probability , 1972 .
jeremy vila , philip schniter , and joseph meola , _ hyperspectral image unmixing via bilinear generalized approximate message passing _ , spie defense , security , and sensing , international society for optics and photonics , 2013 , pp
. 87430y87430y .
pu - ming zhang , jin - yong wu , yi zhou , pei - ji liang , and jing - qi yuan , _ spike sorting based on automatic template reconstruction with a partial solution to the overlapping problem _ , journal of neuroscience methods * 135 * ( 2004 ) , no . 1 , 5565 .
|
principal component analysis ( pca ) aims at estimating the direction of maximal variability of a high - dimensional dataset .
a natural question is : does this task become easier , and estimation more accurate , when we exploit additional knowledge on the principal vector ?
we study the case in which the principal vector is known to lie in the positive orthant .
similar constraints arise in a number of applications , ranging from analysis of gene expression data to spike sorting in neural signal processing . in the unconstrained case , the estimation performances of pca has been precisely characterized using random matrix theory , under a statistical model known as the ` spiked model . '
it is known that the estimation error undergoes a phase transition as the signal - to - noise ratio crosses a certain threshold .
unfortunately , tools from random matrix theory have no bearing on the constrained problem . despite this challenge
, we develop an analogous characterization in the constrained case , within a one - spike model .
in particular : @xmath0 we prove that the estimation error undergoes a similar phase transition , albeit at a different threshold in signal - to - noise ratio that we determine exactly ; @xmath1 we prove that unlike in the unconstrained case estimation error depends on the spike vector , and characterize the least favorable vectors ; @xmath2 we show that a non - negative principal component can be approximately computed under the spiked model in nearly linear time .
this despite the fact that the problem is non - convex and , in general , np - hard to solve exactly .
|
recently the zeus collaboration at hera presented differential cross section data for the neutron tagged process @xmath2 @xcite .
the cross sections have been measured in the photoproduction region with photon virtuality @xmath3 for @xmath4 center - of - mass energy @xmath5 in the interval @xmath6 , for jets with transverse energy @xmath7 , neutron energy @xmath8 , and neutron production angle @xmath9 mrad .
the cross sections were compared to predictions with the one - pion exchange model and rather good agreement was found .
+ due to the kinematic constraints on the neutron detection the squared momentum transfer @xmath10 between ingoing proton and outgoing neutron is very small @xcite . in this case
it is expected that the @xmath11 transition amplitude , _
i.e. _ the amplitude for the production of neutrons , which carry most of the momentum of the proton , is dominated by the lightest particle in the @xmath10-channel , the pion .
hence the isolation of the one - pion exchange contribution provides the possibility to study the photon - pion interaction @xmath12 @xcite .
+ selecting high @xmath13 jets in the final state @xmath14 allows us to apply the qcd improved parton model to describe the characteristic features of the jet production dynamics .
this is completely analogous to the process @xmath15 , which has been calculated in the past up to next - to - leading order ( nlo ) using different computational techniques @xcite .
the only difference in @xmath16 is that the parton distribution functions ( pdfs ) of the proton are replaced by the pdfs of the pion .
in addition , the pion flux , generated by the @xmath11 transition , has to be supplied .
+ the pion flux can , in principle , be measured in charge - exchange processes in soft hadronic reactions , where an initial - state proton is transformed into a final - state neutron , @xmath11 , with small momentum transfer . a successful phenomenological description of the corresponding data has been obtained in the framework of reggeized isovector exchange , such as @xmath17 , and @xmath18 , with the pion being by far the lightest meson dominating the @xmath11 transition , particularly at small values of the squared momentum transfer @xmath10 between proton and neutron @xcite . + the pdfs of the pion can be measured in deep - inelastic scattering with a forward going neutron detected in the final state : @xmath19 @xcite .
the measurements of the structure functions so far are limited to rather small bjorken-@xmath20 @xcite .
the measured semi - inclusive cross sections for leading neutrons by the h1 collaboration @xcite can be described entirely by @xmath21 exchange , and the bjorken-@xmath20 and @xmath22 dependence of these data are consistent with leading order ( lo ) pdfs of the pion . there
exist several such pion pdfs in the literature @xcite .
they are constrained by dimuon and prompt - photon production data from fixed target experiments that are sensitive to the valence quark distribution in a bjorken-@xmath20 range relevant for dijet photoproduction on pions .
+ in this work we present the results of a calculation of differential cross sections for the reaction @xmath23 with the kinematical constraints as in the zeus analysis @xcite .
we have done these calculations in lo and in nlo .
it is well known , however , that lo predictions are not reliable due to the possibly strong scale dependence of the results .
another problem of lo results for jet cross sections is the fact that in lo only two large @xmath13 jets with opposite transverse momenta can be produced . therefore these cross sections are independent of the jet algorithm which is applied in order to define jets in the analysis of the experimental data . in nlo calculations
there appears an explicit dependence on the jet algorithm due to the possible production of three partons ( in addition to the remnant partons on the photon or pion side , respectively ) in the final state , two of which must be combined in one jet in accordance with the experimental constraints for the construction of jets out of hadrons . therefore our lo results are just for comparison and for demonstrating the reduced scale dependence of the nlo results .
our genuine predictions are the nlo cross sections which will be compared with the zeus data @xcite .
+ in the next section we shall describe the kinematical variables and define our input for the pion flux and the pion pdfs .
section 3 contains our results and the comparison with the experimental data . in the last section
we give a short summary and draw some conclusions .
the event kinematics has been described in detail already in @xcite . here
we repeat those variables which are needed in the calculation of the cross cross sections . the reaction @xmath24 where h is the hadronic system containing the jets , is characterized by the four - momenta @xmath25 and @xmath26 of the initial and scattered positron and by @xmath27 and @xmath28 , the four - momenta of the ingoing proton and outgoing neutron , respectively .
the positron - photon vertex is described by the exchanged photon virtuality @xmath22 and the positron s inelasticity @xmath29 , _
i.e. _ @xmath30 in the zeus experiment @xcite @xmath31 and @xmath32 , corresponding to the @xmath4 center - of - momentum energy range @xmath6 .
the protons at hera have the energy @xmath33 and collide with @xmath34 positrons , corresponding to a center - of - momentum frame energy @xmath35 .
the two variables , which describe the proton - neutron vertex , are the fraction of the energy of the initial - state proton carried by the neutron @xmath36 and the square of the momentum transfer @xmath10 between the proton and the produced neutron : @xmath37 the splitting function or pion flux for the transition @xmath38 is usually parameterized by different forms which can be summarized as @xmath39 ^ 2.\ ] ] here @xmath40 is the coupling constant of the @xmath41 vertex , @xmath42 is the pion mass , and @xmath43 is the regge trajectory of the pion .
@xmath44 is a form factor which describes the off - shell behavior of the virtual pion and/or possible final state rescattering effects of the neutron .
two choices for the form factor @xmath44 used in the analysis of various charge - exchange scattering experiments are @xmath45 \\
\exp [ r^2(t - m_{\pi}^2)/(1-x_l ) ] \end{array } \right.\ ] ] where @xmath46 and @xmath47 are constants .
the first of these choices , the exponential form , is usually taken with the regge trajectory factor in ( 4 ) with @xmath48 .
the second choice , the light - cone form factor , is usually associated with the flux without the regge trajectory factor , _
i.e. _ @xmath49 in ( 4 ) .
these choices are associated with the experimental finding that in soft hadronic reactions the shape of the @xmath36 distribution depends on @xmath10 @xcite .
the pion - nucleon coupling constant is well known from low energy @xmath50 and @xmath51 scattering data .
we take @xmath52 as obtained in a recent analysis @xcite .
+ in the scattering process of @xmath53 massless partons , the fractions of the four - momenta @xmath54 and @xmath55 participating in the hard scattering by the initial state parton are given by @xmath56 where the sums in ( 6 ) and ( 7 ) run over the variables of the two jets in the final state .
here we assumed that @xmath57 .
the energy fraction contributing by the exchanged virtual photon to the production of the dijets is @xmath58 whereas the corresponding contribution of the virtual pion ( or possibly of a reggeized @xmath59 or @xmath18 ) is @xmath1 . in ( 6 ) @xmath60 is the energy of the ingoing virtual photon . in nlo also
three jets can be produced in the final state
. then ( 6 ) and ( 7 ) are no longer valid . in order to estimate the energy fractions @xmath58 and @xmath61 for this case , too
, one uses ( 6 ) and ( 7 ) in the form that the sums in ( 6 ) and ( 7 ) run over the two jets of largest @xmath13 in an event .
+ as is well known two mechanisms contribute to the photoproduction of jets , the direct and the resolved process .
the observable @xmath58 is sensitive to the amount of direct and resolved processes .
the lo direct process , where the photon couples directly to the quarks , contributes at @xmath62 , the resolved and the nlo direct processes contribute in the region @xmath63 .
it must be emphasized , however , that the distinction of direct and resolved processes looses its meaning in nlo , so that the characterization of the contributions to the various @xmath58 regions by the direct and resolved processes is not unique : it becomes scale dependent .
the variable @xmath1 can be used to analyze which region of the pion pdfs and which partonic contributions are most important for dijet production with leading neutrons . + as for the jet definition and the combination of two partons into one jet in the nlo contributions of the direct and the resolved parts , we use the @xmath64 cluster algorithm in the form as introduced in @xcite and as used in the jet analysis of the experimental data @xcite .
the photon flux is calculated with the usual weizscker - williams approximation , including the non - logarithmic correction as calculated in @xcite . for the resolved cross sections we need the pdfs of the photon in lo and nlo .
a popular parameterization is grv @xcite .
we use the parameterization gs96 @xcite , since it leads to a slightly better description of the experimental data . for the comparison with data of inclusive jet production without the leading neutron , which has been simultaneously analyzed in @xcite under similar kinematical conditions ,
we need also the pdfs of the proton , for which we employ the recent cteq5 m @xcite parameterization .
the @xmath65 parameter which we need in the nlo @xmath66 formula is taken from the proton pdf fit .
it is @xmath67 mev . for the pion pdfs we employ two alternatives , smrs3 @xcite and grs @xcite .
unfortunately , the @xmath65 values used for the evolution of these pdfs are somewhat smaller , namely @xmath68 and 200 mev . for lo predictions , we use lo matrix elements with the one - loop formula for @xmath66 and the same values of @xmath69 . since only grs have constructed lo and nlo pion pdfs , we have to use the nlo smrs3 set also with our lo predictions .
the grs pdfs of the pion are constructed for three flavors only .
since we include four flavors throughout , we also need the pion pdf for charm .
this is taken from the earlier work of grv @xcite .
the calculation of the cross sections is based on the formalism fully described in our previous work @xcite . to cancel infrared and collinear singularities in the nlo contributions of the final state and to define the collinear initial state singularities to be removed through the renormalization of the pdfs of the photon or the pion ,
the phase - space slicing method was applied . since the neutron kinematics could not be fixed in detail in the experiment we had to integrate over a finite region in @xmath36 and @xmath10 .
we did this in accordance with the specifications of the zeus experimental analysis .
except for the outgoing positron and the leading neutron the final state consists of two or three jets .
the two - jet sample contains the bare parton jets from the lo and virtual nlo contributions and the two jets originating from the recombination of two partons in the three - parton terms .
the three - jet sample is just given by the uncombined three - parton final states .
+ first we calculated the differential cross section @xmath70 , where @xmath13 and @xmath71 are the transverse energy and the rapidity of one of the jets in the two- or three - jet sample , respectively .
as in the zeus analysis we restricted the selection in the three - jet sample .
we included from this sample only the two jets with the largest @xmath13 . in our earlier work @xcite this cross section
was denoted the inclusive one - jet cross section . there , however , we included all three jets from the three - jet contributions . in @xcite the cross section defined above
is denoted the dijet cross section .
this should be distinguished from the inclusive two - jet cross section defined in @xcite , which depends on @xmath13 , the transverse energy of one of the two jets with the highest @xmath13 , and the rapidities @xmath72 and @xmath73 of these high @xmath13 jets .
the @xmath74 is integrated over @xmath75 as in @xcite .
the result for the full jet sample ( without detection of the leading neutron ) as a function of @xmath13 for @xmath7 turn out to be in reasonable agreement with the experimental data from @xcite ( not shown ) .
the point with the largest cross section at @xmath76 agrees perfectly with the measured value .
the points at medium @xmath13 lie somewhat higher than the theoretical curve .
the equivalent cross section @xmath70 , also integrated over @xmath77 , for the case with the leading neutron is shown as a function of @xmath13 for @xmath78 in fig .
1a . in this figure
three curves are shown : one from lo with pion pdfs smrs3 @xcite and photon pdfs gs96 @xcite , the latter in lo , and two other curves for nlo with pion pdfs smrs3 @xcite and grs @xcite and charm content from grv @xcite for comparison .
we see that all three predictions lie very near to each other in the logarithmic plot .
the lo and nlo curves for smrs3/gs96 differ only very little in this plot , showing that the nlo corrections are small for the scale choice @xmath79 as used in fig .
1a ( and in all the following figures ) for the lo and nlo result .
the grs choice for the pion pdfs leads to a somewhat larger cross section at small @xmath13 s and smaller cross section at the larger @xmath13 s as compared to the smrs3 choice .
we have calculated that the gluon in the pion gives the dominant contribution to the cross section for @xmath80 gev .
resolved photons are more important than direct photons for @xmath81 gev .
the experimental data @xcite with statistical and systematic errors shown separately come with a light shaded band which indicates the additional systematic experimental uncertainty due to the calorimeter energy scale @xcite .
the medium shaded band gives the uncertainty of the lo result when the scale of @xmath66 and of the factorization is varied in the interval between @xmath82 and @xmath83 .
the scale uncertainty of the nlo prediction ( dark shaded band ) coincides with the line thickness of the nlo curve .
taking into account the fairly large experimental errors including the calorimeter energy scale uncertainty the agreement of the zeus data with the theoretical predictions concerning shape and absolute normalization is reasonably good .
it must be emphasized , however , that the absolute normalization depends on the pion form factor ( 5 ) . in fig .
1a we assumed the light - cone form factor with @xmath84 .
recent determinations of @xmath85 from a comparison to the rather accurate neutron production data from flauger and mnning @xcite yield @xmath86 @xcite and @xmath87 @xcite in reasonable agreement with our value for @xmath85 . a change of @xmath47 from @xmath88 to @xmath89 , _ i.e. _ an increase by @xmath90 , leads to a @xmath91 smaller cross section .
+ since we have calculated also @xmath70 with the same kinematical constraints for @xmath92 we can take the ratio of the two cross sections @xmath93 for @xmath94 and @xmath95 .
this is shown in fig .
1b and compared to the zeus data for this ratio .
the agreement with the data is perfect . for low @xmath13 , where the experimental errors are smallest , the smrs3 result agrees better than the ratio obtained with the grs pion pdfs , otherwise the results are the same . in fig .
1b the experimental uncertainty from the calorimeter energy measurements , the systematic error and the scale uncertainty in the theoretical curve are not included since they are supposed to cancel in the ratio @xcite .
thus , this ratio seems to be a good example to get information on the pion pdfs and/or the pion form factor .
+ the next calculated differential cross section is @xmath96 , _ i.e. _ @xmath70 integrated over @xmath13 with @xmath97 .
the result is shown in fig .
2a , where @xmath96 is plotted as a function of @xmath98 for @xmath98 between @xmath99 and @xmath100 , together with the data points from @xcite .
there are three theoretical curves for the three cases as in fig . 1a . here
one can see the difference between smrs3/gs96 in lo and nlo .
the difference is very small , the lo cross section is approximately @xmath101 larger than the nlo cross section for the same scale @xmath79 .
concerning the dependence on the pion pdfs the choice smrs3 agrees best with the data .
the result with the grs pdfs is larger , in particular for negative @xmath98 . in @xcite two other pion pdfs
have been published , smrs1 and smrs2 .
the cross sections for these two pdfs lie between the smrs3 and the grs curves and are therefore not shown explicitly .
the gluons in the pion contribute about twice as much as quarks to the rapidity distribution over the entire @xmath98-range .
direct ( resolved ) photons dominate in the region @xmath102 @xmath103 .
all three curves do not agree well with the data points for @xmath104 . for jet production without leading neutron this region
is known to be influenced by hadronization corrections @xcite .
we expect such corrections also for the cross section in fig .
2a , although somewhat less in the case with leading neutron due to the reduced total energy producing the hard jets .
the theoretical result with such hadronic corrections included ( taken from the case without leading neutron @xcite ) is shown in fig .
2a as the lower full curve .
it leads to a better agreement in the negative @xmath98 region , but it produces also a somewhat lower cross section at positive @xmath98 s , reducing the good agreement wit the data in this region . in fig .
2a we show again three shaded bands : ( i ) the light shaded band shows the additional systematic uncertainty from the calorimeter energy scale , ( ii ) the medium shaded band gives the variation of the lo result ( dotted curve ) with the scale @xmath105 , which was varied as usual in the region @xmath106 , ( iii ) the dark shaded band shows the variation of the nlo cross section with changing the scale in the same interval . here , we see quite clearly that the nlo result has a very much reduced scale dependence compared to the lo result . similarly to fig .
1b we have plotted in fig .
2b the ratio of the cross sections for smrs3/gs96 and grs / gs96 and the cross section @xmath96 for dijet production without leading neutron as a function of @xmath98 and compared them with the same experimental ratio from zeus . within errors the smrs3 and the grs results are consistent with the data , the smrs3 curve agrees slightly better .
+ the last differential cross section which we investigated is @xmath107 as a function of @xmath108 . in order to determine @xmath1 , which is a measure of the exchanged pion s momentum fraction participating in the production of jets as defined in ( 7 ) , we need the transverse momenta and rapidities of the two most energetic jets in the event . the experimental cross section @xmath109 , however , has the constraint that for the two jets @xmath110 _ and _ @xmath111 .
cross sections for @xmath112 are problematic theoretically , since they become infrared sensitive in nlo , _
i.e. _ depend on the slicing cut used to cancel infrared and collinear singularities . in order to avoid this sensitivity and
also possible hadronic and intrinsic @xmath64 effects not considered in our nlo framework we need constraints on @xmath113 or @xmath114 which smear out the problematic region @xmath112 .
this problem is well known and was encountered some time ago also in connection with the calculation of the inclusive dijet cross section in @xmath4 collisions @xcite .
one remedy to remove the infrared sensitivity is the requirement of slightly different lower limits on @xmath115 and @xmath116 .
therefore , we have calculated @xmath117 with the constraint @xmath118 gev , @xmath119 gev if @xmath120 or @xmath115 and @xmath116 interchanged if @xmath121 .
the cross sections calculated with this constraint are plotted in fig .
3 , again for the three cases as in fig .
these cross sections are compared with the corresponding zeus data .
the nlo prediction with the pion pdfs of smrs3 agrees better with the data than the one with grs pion pdfs . except for large @xmath122 the gluons in the pion dominate over quarks , and
the resolved photon contribution is larger than the direct one for @xmath123 .
the agreement with the data is good for @xmath124 , but less good for the smaller @xmath1 .
this @xmath1 region is related to the negative @xmath98 region in the @xmath98 distribution in fig .
2a , where one has hadronization corrections ( see ( 7 ) for the relation between @xmath1 and the @xmath125 s ) .
we note that the k factor between the nlo and lo result is larger than in @xmath96 and is larger than one now in particular for small @xmath1 .
this is connected with the specific constraint on @xmath115 and @xmath116 , which was not necessary for the results in figs . 1 and 2
. this constraint does not exactly correspond to the one applied to the measured @xmath117 . on the other hand
the low @xmath13 calorimeter jets used in the experimental analysis have an uncertainty of @xmath101 which introduces a corresponding uncertainty on the constraint @xmath126 . in order to see the influence of
the @xmath127 cut we reduced the difference @xmath128 between @xmath115 and @xmath116 from @xmath129 to @xmath130 .
the result is shown in fig . 3 as the lower full curve .
the reduction of the difference leads to a lower cross section as we expect , so that the low @xmath1 experimental points are described very well now but not the higher @xmath1 region .
since this region is much less affected by hadronization corrections we prefer the original prediction ( upper full curve in fig .
3 ) . the absolute normalization of the cross section @xmath109 depends also on the pion form factor .
the curves in fig .
3 are calculated with the light - cone form factor with @xmath131 .
@xmath89 instead gives a @xmath91 lower cross section and better agreement for @xmath132 but worse agreement above this value .
again taking into account possible hadronization corrections we consider the value @xmath84 as more realistic .
the result with @xmath89 is almost indistinguishable from the result with the exponential form factor with @xmath133 .
+ in @xcite several uncorrected distributions are presented , in particular event rates as a function of @xmath58 in order to say something about the amount of resolved and direct photoproduction . from these rates ,
the ratio of resolved ( corresponds to @xmath134 ) to the direct cross section ( corresponds to @xmath135 ) has been deduced .
we checked that the theoretical prediction for this ratio is in reasonable agreement with the data .
it decreases with increasing @xmath13 , as one expects it .
+ in this work we considered only the pion exchange in the @xmath10-channel .
other isovector exchanges , such as @xmath59 and @xmath18 , are expected to be small .
recent studies of leading neutron production in @xmath136 collisions show that the rate can be described entirely by @xmath137 exchange @xcite .
the additional exchanges , which increase the rate of neutron production , are offset by absorptive rescattering of the neutron , which decreases the rate by approximately the same amount @xcite .
therefore both effects are neglected in the present analysis .
if the experimental data were more specific concerning the dependence on @xmath10 and @xmath36 , the dominance of the @xmath137 exchange could be demonstrated .
we have presented a next - to - leading order calculation of dijet production in collisions of almost real photons and pions , where the photons are generated through bremsstrahlung from an initial electron or positron beam and the pion is assumed to give the dominant contribution to the forward proton to neutron transition amplitude .
our results have been compared with recent data from zeus for distributions in the transverse energy @xmath13 and rapidity @xmath98 of one of the two jets and for the observed momentum fraction of the partons in the pion @xmath61 .
we found that the nlo corrections to the lo result were generally small , also in the case of inclusive photon - proton scattering with similar kinematic constraints , but lead to a largely reduced dependence on the renormalization and factorization scales .
for the nlo distribution in @xmath138 , we had to relax the experimental condition of @xmath139 gev to allow for a small difference @xmath140 gev of the two minimal transverse energies , corresponding to the non - zero experimental uncertainty in the jet energy scale .
+ the overall normalization of all three distributions is sensitive to the form factor of the @xmath141 transition .
we have used a light - cone form with a value for the parameter @xmath47 which agrees well with recent determinations from neutron production data .
once the normalization is fixed , the shapes of the distributions , particularly the one in @xmath138 , are sensitive to the parton distributions in the pion .
the zeus data for the distribution in @xmath138 as well as in @xmath13 and @xmath98 are best described by the smrs3 pion parameterization , whereas the grs parameterization with grv charm distribution gives larger predictions , especially at small @xmath61 .
we emphasize that for small @xmath13 one is particularly sensitive to the gluon content of the pion .
furthermore , the gs96 photon densities agree better with the data than grv .
hadronization corrections were found to reduce the perturbative prediction considerably in the region of small @xmath98 , which corresponds to small @xmath61 .
we thank m. khakzad , g. levman and j. whitmore for making the zeus data available to us and m. wing for calculating the hadronization corrections with the herwig monte carlo generator .
financial support by the deutsche forschungsgemeinschaft through grant no .
kl 1266/1 - 1 , by the bundesministerium fr bildung und forschung through grant no . 05 ht9gua 3 , and by the european commission through the research training network _
quantum chromodynamics and the deep structure of elementary particles _ under contract no . erbfmrx - ct98 - 0194 is gratefully acknowledged .
j. breitweg _ et al . _
[ zeus collaboration ] , desy-00 - 142 , october 2000 , hep - ex/0010019 .
g. kramer , proc . of the seminar on _ e p and e e storage rings _
, ed . by j.k .
bienlein , i. dammann and h. wiedemann , desy-73 - 66 , december 1973 .
m. klasen and g. kramer , z. phys .
c * 76 * ( 1997 ) 67 ; + m. klasen , t. kleinwort and g. kramer , eur .
j. direct c * 1 * ( 1998 ) 1 .
b. w. harris and j. f. owens , phys .
d * 56 * ( 1997 ) 4007 ; + s. frixione and g. ridolfi , nucl .
b * 507 * ( 1997 ) 315 ; + p. aurenche , l. bourhis , m. fontannaz and j. p. guillet , eur .
j. c * 17 * ( 2000 ) 413 .
u. dalesio and h. j. pirner , eur .
j. a * 7 * ( 2000 ) 109 and the earlier literature given there .
j. d. sullivan , phys .
d * 5 * ( 1972 ) 1732 ; + n. s. craigie and g. schierholz , nucl .
b * 100 * ( 1975 ) 125 .
m. derrick _ et al . _
[ zeus collaboration ] , phys .
b * 384 * ( 1996 ) 388 . c. adloff
_ et al . _
[ h1 collaboration ] , eur .
j. c * 6 * ( 1999 ) 587 .
p. j. sutton , a. d. martin , r. g. roberts and w. j. stirling , phys .
d * 45 * ( 1992 ) 2349 .
j. f. owens , phys .
d * 30 * ( 1984 ) 943 ; + p. aurenche , r. baier , m. fontannaz , m. n. kienzle - focacci and m. werlen , phys .
b * 233 * ( 1989 ) 517 ; + m. glck , e. reya and a. vogt , z. phys .
c * 53 * ( 1992 ) 651 .
m. glck , e. reya and i. schienbein , eur .
j. c * 10 * ( 1999 ) 313 .
j. engler _ et al .
_ , nucl .
b * 84 * ( 1975 ) 70 ; + w. flauger and f. mnnig , nucl .
b * 109 * ( 1976 ) 347 ; + b. robinson _
et al . _ ,
* 34 * ( 1975 ) 1475 ; + g. levman and k. furutani , desy-95 - 142 , july 1995 , + and further references given there .
t. e. ericson , b. loiseau and a. w. thomas , hep - ph/0009312 .
s. catani , y. l. dokshitzer , m. h. seymour and b. r. webber , nucl .
b * 406 * ( 1993 ) 187 ; + s. d. ellis and d. e. soper , phys .
d * 48 * ( 1993 ) 3160 .
s. frixione , m. l. mangano , p. nason and g. ridolfi , phys .
b * 319 * ( 1993 ) 339 .
m. glck , e. reya and a. vogt , phys .
d * 46 * ( 1992 ) 1973 .
l. e. gordon and j. k. storrow , nucl .
b * 489 * ( 1997 ) 405 .
h. l. lai _ et al . _
[ cteq collaboration ] , eur .
j. c * 12 * ( 2000 ) 375 .
h. holtmann , a. szczurek and j. speth , nucl .
a * 596 * ( 1996 ) 631 .
b. w. harris , m. klasen and j. vossebeld , proc . of the workshop on _ monte carlo generators at hera physics _ ,
hamburg , germany , 1998 , hep - ph/9905348 ; + m. wing , priv . communication .
|
differential cross sections for inclusive dijet photoproduction on a virtual pion have been calculated in next - to - leading order qcd as a function of @xmath0 , and @xmath1 .
the cross sections are compared with recent zeus data on photoproduction of dijets with a leading neutron in the final state .
|
three - pion final states of two - photon interactions are restricted to quantum numbers suitable for study of resonance formation .
the @xmath6 channel is known to be dominated by the formation of @xmath4 in the helicity 2 state [ 1 - 10 ] .
the @xmath4 is a ground state of isospin 1 @xmath7 @xmath8 mesons .
observations of higher mass resonances have been reported [ 7 - 10 ] .
study of higher mass states is important for the assignment of nonet members and for the understanding of confinement in the quark model [ 11 - 15 ] .
the l3 collaboration reported the @xmath5 in @xmath6 and @xmath9 decay modes mainly in helicity 2 states @xcite .
the belle observation of a resonance in @xmath10 @xcite at 1737 is listed in the particle data group ( pdg ) @xcite and attributed to the @xmath5 .
experiments on @xmath11 collisions [ 18 - 20 ] and @xmath12 photo - production @xcite have also observed a three - pion resonance near @xmath13 .
the crystal barrel collaboration reported the @xmath5 in the @xmath14 mode in @xmath15 collisions @xcite .
the @xmath16 and @xmath17 in the @xmath18 mode were reported in @xcite .
we present a spin - parity analysis of three - pion events using data collected with the belle detector at kekb @xcite .
resonance formation is investigated in the @xmath6 and @xmath9 modes , including interference between them .
the data sample was taken with similar trigger conditions during 2000 and 2001 at a center - of - mass energy @xmath19 .
the corresponding total integrated luminosity is 26.0 fb@xmath1 .
we first describe the theoretical formulae for spin - parity analysis of three - pion events .
the spin dependence is investigated using the distribution of the vector normal to the three - pion decay plane .
upper limits on two - photon radiative widths are determined for @xmath20 , @xmath21 and @xmath22 .
a neural network method is applied to enhance tensor state selection .
the helicity states of tensor resonances are investigated using @xmath23 distributions of final state pions .
background contamination is evaluated by comparing @xmath24 distributions for data and monte carlo generated two - photon interactions .
the three - pion and di - pion mass spectra are investigated .
the two - photon radiative width of the @xmath4 is measured and compared to the pdg world average .
the mass region above @xmath25 is examined for the @xmath5 and new states .
the two - photon radiative widths are determined .
interference between tensor states and @xmath6 and @xmath9 decay modes is investigated and the coupling amplitudes and phase angles are also determined .
the three - pion final state in two - photon interactions is expressed as the composition of production of di - pion isobars , @xmath26 , and the decay of @xmath27 .
possible di - pion isobars are @xmath28 , @xmath29 and @xmath30 . the assignment of spin - parity ( @xmath31 ) and orbital angular momentum ( l ) are listed in table [ tab : spin ] .
gauge invariance and bose symmetry forbid @xmath32 and @xmath33 .
only the helicity @xmath34 state is allowed for @xmath35 and @xmath36 , and @xmath37 for @xmath38 @xcite .
the helicity 0 fraction of a tensor state is predicted to be zero by the non - relativistic quark model @xcite .
the static quark model also predicts a small value ( 1/7 ) @xcite . .
spin - parity of the three - pion final state in two - photon interactions . listed are the possible di - pion isobars ( @xmath28 , @xmath39 or @xmath30 ) , helicity ( @xmath40 ) states allowed , and the lowest orbital angular momentum ( @xmath41 ) between the di - pion isobar and the third pion . [
tab : spin ] [ cols="<,^,^,^,^,^",options="header " , ]
in the region of low three pion mass the reaction @xmath42 is dominated by the @xmath4 in the helicity 2 state .
events in higher mass regions are also found to be dominated by the spin - parity @xmath3 helicity 2 partial wave .
the contribution of @xmath20 is negligible .
the upper limit is found to be @xmath43 ( 90% c.l ) .
the contributions of the @xmath21 and @xmath22 resonances are also negligible .
the upper limits are determined to be @xmath44 and @xmath45 ( 90% cl ) , respectively .
the observed radiative width of the @xmath4 is consistent with the pdg world average .
the mass spectrum in the region above the @xmath4 consists of a broad distribution attributed to the @xmath5 and is best interpreted with two more radially excited tensor states at 1950 and 2140 decaying to @xmath6 and @xmath9 .
the mass spectrum is well represented by interfering tensor states with coupling amplitudes and phase angles relative to @xmath4 .
the amplitude obtained for @xmath5 is 0.37 , corresponding to the radiative width of @xmath46 .
the higher mass tensor states observed here may be compared to the theoretical predictions of the relativistic quark model @xcite .
the radiative widths obtained in this study are compatible with the relativistic quark model calculations .
|
a high statistics study of the reaction @xmath0 has been performed with the belle detector using a data sample of 26 fb@xmath1 collected at @xmath2 .
a spin - parity analysis shows dominance of the @xmath3 helicity 2 wave for three - pion invariant masses from 1 to 3 .
the invariant mass distribution exhibits @xmath4 , @xmath5 and higher mass enhancements .
|
the main objects of study in this paper are the _ 2-dimensional orthant model _ ( one of the most interesting examples within a class of models called _ degenerate random environments _ ) , and its dual model , a version of _ oriented site percolation_. part of the motivation for studying degenerate random environments is an interest in the behaviour of random walks in random environments that are non - elliptic . indeed , many of the results of this paper and of @xcite have immediate implications for the behaviour ( in particular , directional transience ) of random walks in certain non - elliptic environments ( see e.g. @xcite ) . for fixed @xmath2 , let @xmath3 be the set of unit vectors in @xmath4 , and let @xmath5 denote the power set of @xmath6
let @xmath7 be a probability measure on @xmath5 .
a _ degenerate random environment _ ( dre ) is a random directed graph , i.e. an element @xmath8 of @xmath9 .
we equip @xmath9 with the product @xmath10-algebra and the product measure @xmath11 , so that @xmath12 are i.i.d . under @xmath13 .
we denote the expectation of a random variable @xmath14 with respect to @xmath13 by @xmath15 $ ] .
we say that the dre is _
@xmath16-valued _ when @xmath7 charges exactly two points , i.e. there exist distinct @xmath17 and @xmath18 such that @xmath19 and @xmath20 . as in the percolation setting ,
there is a natural coupling of graphs for all values of @xmath21 as follows .
let @xmath22 be i.i.d .
standard uniform random variables under @xmath13 .
setting @xmath23 it is easy to see that for any fixed @xmath21 , @xmath24 has the correct law , and that the set of @xmath25 sites is increasing in @xmath21 , @xmath13-almost surely .
for the most part , in this paper , we will consider 2-dimensional models .
interesting examples of @xmath16-dimensional , @xmath16-valued models include the following : @xmath26 : [ exa : ne_sw ] let @xmath27 and @xmath28 ( and set @xmath19 , @xmath20 ) .
we call the generalization to @xmath29 dimensions the _ orthant model _ ( so this is the
_ 2-@xmath29 orthant model _ ) .
@xmath30 [ exa : swe_n ] let @xmath31 and @xmath32 .
@xmath33 : [ exa : ns_we ] let @xmath34 and @xmath35 .
we will focus on model [ exa : ne_sw ] in this paper , but we believe extensions of our results to model [ exa : swe_n ] are possible .
a central limit theorem for random walk in the degenerate random environment of model [ exa : ns_we ] is proved in @xcite .
see @xcite for more on these and other examples of @xmath16-dimensional @xmath16-valued models , and their properties . for any set @xmath36 ,
let @xmath37 denote its cardinality .
[ def : connections ] given an environment @xmath24 : * we say that @xmath38 is _ connected _ to @xmath39 , and write @xmath40 if : there exists an @xmath41 and a sequence @xmath42 such that @xmath43 for @xmath44 .
we say that @xmath38 and @xmath39 are _ mutually connected _ , or that they _
communicate _ , and write @xmath45 if @xmath40 and @xmath46 .
* define @xmath47 ( the _ forward _ cluster ) , @xmath48 ( the _ backward _ cluster ) , and @xmath49 ( the _ bi - connected _ cluster ) .
set @xmath50 , @xmath51 , and @xmath52 .
* a nearest neighbour path in @xmath4 is _ open in @xmath24 _ if that path consists of directed edges in @xmath24 .
it is interesting to consider percolation - type questions where @xmath53 , and indeed in the true percolation settings where @xmath54 for some configuration @xmath55 , there is a simple relation between the directed percolation probabilities of the form @xmath56 ( see ( * ? ? ?
* lemma 2.5 ) ) .
our current interest lies in those cases where @xmath7 is such that @xmath57 this is precisely the condition which ensures that a random walk on this random graph visits infinitely many sites .
it is equivalent to the condition that there exists a set of orthogonal directions @xmath58
such that @xmath59 , i.e. that almost surely at every site , some element of @xmath60 occurs ( * ? ? ?
* lemma 2.2 ) .
models [ exa : ne_sw]-[exa : ns_we ] above satisfy this condition ( e.g. by taking @xmath61 ) .
note that @xmath62 if and only if there exists @xmath63 such that @xmath64 , and that none of these three models satisfy this condition .
in fact @xmath65 for each of these models when @xmath18 @xcite .
model [ exa : ne_sw ] exhibits a phase transition ( in fact two phase transitions by symmetry ) when the parameter @xmath21 changes @xcite . while @xmath66 and @xmath67 for all @xmath21 , the geometry of an infinite @xmath68 changes from having a non - trivial boundary to being all of @xmath0 as @xmath21 decreases from @xmath69 to @xmath70 .
moreover the critical point @xmath71 at which this transition takes place is also the critical point @xmath72 for oriented site percolation on the triangular lattice ( otsp ) .
information about the geometry of an infinite @xmath73 is then inferred , based on the geometry of @xmath68 , and crude estimates of @xmath71 are given .
similar results hold for model [ exa : swe_n ] , which is dual to a partially oriented site percolation model on the triangular lattice . however , for model [ exa : ns_we ] , if @xmath74 then @xmath75 almost surely , regardless of @xmath21 @xcite .
simulations indicate that @xmath1 and infinite @xmath76 clusters have similar geometry , except that @xmath1 typically has
holes " whereas @xmath76 does not .
in order to give a clearer description of this weak kind of duality , we study the geometry of @xmath77 , defined by @xmath78 when @xmath79 and of [ @xmath1 and @xmath76 when @xmath80 for model [ exa : ne_sw].,title="fig : " ] when @xmath79 and of [ @xmath1 and @xmath76 when @xmath80 for model [ exa : ne_sw].,title="fig : " ] the results of this paper can be separated into two groups . the first group concerns the possible geometries of the sets @xmath81 for a class of 2-dimensional models .
these appear in section [ sec : cgeneral ] , and include adaptations of the results about the geometries of @xmath68 clusters in @xcite . the second group of results concerns model [ exa : ne_sw ] .
some of these results , included in section [ sec : orthant_main ] , are again adaptations of results in @xcite , and are proved using similar arguments .
others exploit much more deeply the duality between @xmath76 , @xmath82 and otsp .
an analysis of otsp ( see section [ sec : op ] and also e.g. @xcite ) is used to prove the following result ( see also figure [ fig : orthant_c_o_b_o ] ) in sections [ sec : orthant_main ] and [ sec : slopes ] .
[ thm : orthant_main ] for model [ exa : ne_sw ] @xmath26 , the following hold : 1 .
@xmath83 if and only if @xmath84 is infinite with positive probability .
[ thm : mphasetransition ] 2 .
@xmath85 @xmath13-a.s . whenever @xmath86 , under the coupling ( [ eq : coupling ] ) .
[ thm : orthant_monotone ] 3 . when @xmath87 , the northwest - pointing boundary of @xmath82 has an asymptotic slope @xmath88 , with @xmath89 strictly increasing to @xmath90 as @xmath91 .
the southeast - pointing boundary of @xmath82 has asymptotic slope @xmath92 .
[ thm : orthant_slope ] note that results of @xcite already show @xmath93 for ( i ) .
but @xmath94 is only verified there assuming a stronger condition on @xmath84 ( that @xmath84 be gigantic " ) .
in contrast to ( ii ) , the cluster @xmath95 is _ not _ monotone in @xmath21 under the natural coupling ( [ eq : coupling ] ) .
for example , letting @xmath96 , we see that @xmath97 if @xmath98 or @xmath99 , and @xmath100 otherwise ( in fact @xmath101 ) .
even when restricting attention to @xmath102 , holes may open or close in @xmath1 as @xmath21 increases . finally note that a corollary of ( iii ) is that random walks in random environments whose supports are @xmath103 with probability @xmath21 and @xmath104 with probability @xmath105 are transient in direction @xmath106 when @xmath87 ( see @xcite ) . on the other hand , by finding arbitrarily large connected circuits in model [ exa : ne_sw ] when @xmath107 we obtain the following result ( proved in section [ sec : pcbounds ] ) about otsp .
[ thm : otsp_pc ] the critical occupation probability for oriented site percolation on the triangular lattice ( @xmath108 ) is at least @xmath109 .
this improves on the best rigorous bounds that we have found in the literature , namely : @xmath110 ( see @xcite ) .
note that the estimated value is @xmath111 @xcite .
we believe that an adaptation of these arguments to model [ exa : swe_n ] yields a bound on the critical occupation probability @xmath112 for a partially oriented site percolation model , and it may be that one can similarly obtain useful bounds on the critical probability @xmath113 for oriented site percolation on the square lattice via related degenerate random environments on the triangular lattice .
in this section we investigate properties of the random sets @xmath114 .
typically these clusters are rather different from connected clusters in percolation models ( where @xmath115 ) .
for example , in the @xmath16-valued setting the sets @xmath116 are not increasing under the coupling ( [ eq : coupling ] ) unless @xmath117 .
in particular for models [ exa : ne_sw]-[exa : ns_we ] the cluster @xmath1 is not monotone in @xmath21 .
a natural question is to ask whether or not the connection events @xmath118 are positively correlated , i.e. whether @xmath119 .
while such a property is true in the percolation setting , this fails in general for degenerate random environments .
the easiest example is the case @xmath120 and @xmath121 , where @xmath122 .
we believe that it fails for models [ exa : ne_sw]-[exa : ns_we ] ( and many others ) as well . on the other hand , for a fixed @xmath39 , the events @xmath123 are positively correlated .
[ lem : jointc ] for @xmath124 , ( y _ o , y _
it is sufficient to prove that @xmath125 .
let @xmath126 be the set of lattice sites in a ball of radius @xmath127 , centred at @xmath128 .
let @xmath129 be the event that there is a path from @xmath128 to @xmath39 lying entirely in @xmath126 .
let @xmath130 be the event that there is no path connecting @xmath38 to @xmath39 that lies entirely in @xmath126
. then @xmath131 and @xmath132 are increasing and decreasing events respectively , with @xmath133 and @xmath134 .
therefore , @xmath135 let @xmath136 be the set of sites that can be reached from @xmath38 using only sites in @xmath137 . observe that if @xmath138 and @xmath139 in @xmath137 then @xmath40 in @xmath137 .
thus for any @xmath140 with @xmath141 , on the event @xmath142 , we have that @xmath143 occurs if and only if @xmath144 occurs .
this latter event depends only on the random variables @xmath145 , while @xmath142 depends only on the random variables @xmath146 .
thus we have @xmath147 in other words , @xmath148 , and taking the limit as @xmath149 establishes the result .
note that by translation invariance and relabelling of vertices , lemma [ lem : jointc ] is equivalent to saying that @xmath150 so roughly speaking , knowing that something connects to @xmath128 makes it more likely that other things connect to @xmath128 .
let @xmath151 .
we say that @xmath152 has a
_ finite hole _
@xmath153 if @xmath154 , @xmath153 is finite , @xmath153 is connected in @xmath0 , and every @xmath155 that is a neighbour of @xmath153 must belong to @xmath152 .
the following elementary lemma implies that @xmath82 is obtained from @xmath1 by filling in all finite holes , and that the backward cluster of a finite hole @xmath153 is simply @xmath153 .
[ lem : choles ] suppose that @xmath156 .
then @xmath38 belongs to a finite hole @xmath153 of @xmath1 and @xmath157 .
let @xmath156 , and let @xmath158 be the @xmath0-connected cluster of @xmath38 in @xmath159 .
clearly every neighbour @xmath155 of @xmath153 is in @xmath160 .
let @xmath161 .
then there is a finite self - avoiding path in @xmath153 from @xmath162 to @xmath39 , so @xmath163 .
thus @xmath164 .
every infinite connected subset of @xmath0 contains an infinite nearest - neighbour self - avoiding path ( it is easy to construct this iteratively - if @xmath165 is connected to infinity then there exists some neighbour @xmath166 of @xmath165 that connects to infinity off @xmath167 etc ) .
therefore @xmath153 is finite .
thus @xmath153 is a finite hole containing @xmath38 .
finally , since @xmath168 and @xmath169 , it follows that @xmath157 .
the above relationships between @xmath170 and @xmath171 clusters are rather weak
. we can however prove results about the @xmath171 clusters which are dual to results about @xmath170 clusters in @xcite , using modifications of the arguments from @xcite .
to state these results , we need the notion of blocking functions .
[ def : blocked ] given @xmath172 , define @xmath173 and @xmath174 by @xmath175}\le w(y^{[1]})\ } \quad \text { and } \quad w _
> = \{y\in { { { \mathbb z}}}^2 : y^{[2 ] } > w(y^{[1]})\}.\ ] ] we say that @xmath39 is _
below @xmath176 _ if @xmath177}\le w(y^{[1]})$ ] , and _ strictly below @xmath176 _ if @xmath177 } < w(y^{[1]})$ ] .
we define @xmath178 and @xmath179 similarly , and speak likewise of @xmath39 being _ above @xmath176 _ or _ strictly above @xmath176_. we say that @xmath172 is a _ forward lower blocking function _ ( flbf ) for @xmath24 if there is no open path in @xmath24 from @xmath178 to @xmath179 , i.e. if @xmath180 .
similarly , @xmath176 is a _
forward upper blocking function _
( fubf ) for @xmath24 if @xmath181 .
note that these notions are different from the ( backward ) lower blocking function @xmath182 and ( backward ) upper blocking function @xmath183 in @xcite .
in particular , @xmath176 is a flbf if and only if @xmath184 is a bubf , and @xmath176 is a fubf if and only if @xmath185 is a blbf .
we write @xmath186 for @xmath187 and use shorthand such as @xmath188 .
the following propositions are @xmath171-dual versions of the @xmath170 results ( * ? ? ?
* proposition 3.8 and corollary 3.10 ) .
[ prp : ctrichotomy ] fix @xmath108 .
assume that @xmath189 , @xmath190 , @xmath191 , @xmath192 , @xmath193 , and @xmath194 . 1 .
the following @xmath13-a.s .
exhaust the possibilities for @xmath81 : 1 .
@xmath195 ; 2 .
there exists a decreasing flbf @xmath196 such that @xmath197 .
3 . there exists a decreasing fubf @xmath196 such that @xmath198 ; 2 . only one of ( i ) , ( ii ) , ( iii ) can have probability different from 0 .
[ prp : ctrichotomy2 ] fix @xmath108 .
assume that @xmath199 , @xmath189 , @xmath191 , @xmath192 , and @xmath193
. 1 . the following @xmath13-a.s .
exhaust the possibilities for @xmath81 : 1 .
@xmath195 ; 2 .
there exists a flbf @xmath196 such that @xmath197 ; 2 .
only one of ( i ) or ( ii ) can have probability different from 0 .
proposition [ prp : ctrichotomy ] applies to model [ exa : ne_sw ] , while proposition [ prp : ctrichotomy2 ] applies to model [ exa : swe_n ] .
when @xmath197 ( so @xmath200 is a flbf ) we say that @xmath81 is _ blocked below_. similarly when @xmath198 ( so @xmath200 is a fubf ) we say that @xmath81 is _ blocked above_. an important notion that arises in the proofs of these results ( and elsewhere throughout this paper ) is the asymptotic slope of a path .
[ def : slope ] a nearest - neighbour path @xmath201 with @xmath202},x_i^{[2]})\in { { { \mathbb z}}}^2 $ ] is said to have _
asymptotic slope @xmath10 _ if @xmath203}}{x_n^{[1]}}=\sigma.\ ] ] _ proof of proposition [ prp : ctrichotomy ] .
_ take @xmath204 . as in the proof of (
* proposition 3.8 ) , we may construct nw or se paths from any point .
suppose that @xmath205 , @xmath206 , @xmath207 , @xmath208}=\bar y^{[1]}=\bar z^{[1]}$ ] , but @xmath209}<w^{[2]}<\bar z^{[2]}$ ] .
because @xmath210 is either in @xmath1 or it is enclosed by @xmath1 , we can find @xmath211 such that @xmath212}=\bar y^{[1]}$ ] but @xmath177}\le \bar y^{[2]}$ ] .
likewise there is a @xmath213 with @xmath214}=\bar z^{[1]}$ ] and @xmath215}\ge \bar z^{[2]}$ ] .
the se paths from @xmath39 and @xmath216 intersect , by ( * ? ? ?
* lemma 2.3 ) .
so do the nw paths from @xmath39 and @xmath216 .
these four paths enclose @xmath176 , so @xmath217 . letting @xmath218 and
@xmath219 it follows that @xmath220 . case ( i ) corresponds to @xmath221 and @xmath222 .
we can rule out the possibility that @xmath223 or @xmath224 jump from finite to infinite values , or vice versa , just as in ( * ? ? ?
* proposition 3.8 ) . to see that if @xmath223 takes finite values , it must be decreasing , consider @xmath225 . by definition ,
since @xmath190 , it follows that @xmath227 .
this implies that @xmath228 .
similarly , @xmath229 is decreasing if finite .
if @xmath230 then @xmath231 is a flbf , and similarly if @xmath232 then @xmath233 is a fubf .
thus it remains to prove that one of @xmath223 or @xmath224 must be infinite , and it suffices to do this for @xmath234 and @xmath235 . to do this , we make use of a number of paths .
define the @xmath236 path ( @xmath237 ) from @xmath38 to be that path starting from @xmath38 obtained by following @xmath238 whenever possible , and otherwise following @xmath239 . on a set of @xmath13-measure 1 ,
this path exists ( since @xmath189 ) , and has asymptotic slope @xmath240 .
similarly the @xmath241 path from @xmath38 , defined to be that path starting from @xmath38 obtained by following @xmath239 whenever possible and otherwise following @xmath238 , exists and has asymptotic slope @xmath242 .
we can also define the @xmath243 and @xmath244 paths from @xmath38 and find their asymptotic slopes .
except in the case that @xmath245 or @xmath246 ( whence trivially @xmath224 or @xmath223 is infinite ) we have that @xmath247 and @xmath248 .
moreover , since every direction is possible we have that @xmath249 and @xmath250 .
let @xmath251 .
if @xmath252 , we may start from the origin and follow the @xmath244 path until reaching a vertex @xmath253 on this path from which the @xmath236 path includes a vertex @xmath254 for some @xmath255 .
this shows that @xmath235 is almost surely infinite .
if @xmath256 then follow the @xmath243 path from the origin until reaching a vertex @xmath257 on this path from which the @xmath241 path includes a vertex @xmath258 for some @xmath255 .
this shows that @xmath223 is almost surely infinite .
it remains to verify the claim when @xmath259 , which implies that @xmath260 and @xmath261 . in this case
we define a new path , called the @xmath262 path . from a vertex @xmath38 ,
it follows whichever of @xmath239 or @xmath238 is possible ( now only one will be ) , except when at a vertex @xmath39 such that @xmath263 and @xmath264 ( i.e. such that @xmath265 and @xmath266 ) . at such a vertex @xmath39
, it follows the @xmath267 step , followed by the @xmath239 step .
this path has a slope @xmath268 .
we may now proceed as before by following the @xmath243 path from @xmath128 until reaching a vertex @xmath257 from which the @xmath262 path passes through @xmath258 for some @xmath255 .
this shows that @xmath223 is almost surely infinite , and completes the proof of part ( a ) .
part ( b ) now follows as in ( * ? ? ?
* proposition 3.8 ) .
the proof of proposition [ prp : ctrichotomy2 ] is a modification of that of ( * ? ? ?
* corollary 3.10 ) , in exactly the same way as the proof of proposition [ prp : ctrichotomy ] adapts that of ( * ? ? ?
* proposition 3.8 ) .
we therefore omit it .
for any site @xmath269 , let @xmath270 denote the set of sites @xmath271 for which there is some @xmath272 and a sequence @xmath273 such that @xmath274 for @xmath275 and @xmath276 for @xmath277 .
similarly let @xmath278 denote the set of sites @xmath271 for which there is some @xmath272 and a sequence @xmath279 such that @xmath274 and @xmath280 for each @xmath281 .
let @xmath282 i.e. @xmath283 is the set of sites that are in a bi - infinite cluster ( in the sense of oriented site - percolation on the triangular lattice ( otsp ) ) of @xmath103 sites . for model [ exa : ne_sw ]
, the conclusions of proposition [ prp : ctrichotomy ] can be extended to the following .
[ prp : cphasetransition ] for model [ exa : ne_sw ]
@xmath26 : 1 .
@xmath284 is a.s .
blocked above ; 2 . @xmath285 a.s . ; 3 .
@xmath286 is a.s .
blocked below .
we proceed as in the proof of ( * ? ? ?
* theorem 3.12 ) , and will reiterate part of the latter in order to explain the role of the triangular lattice .
let @xmath287 be decreasing . for @xmath176 to be a flbf for @xmath24 , the vertices in @xmath288 that have a ( square - lattice ) nearest - neighbour in @xmath289 can be enumerated naturally as @xmath290 to form a sequence of vertices moving upwards and to the left .
the possible transitions in this sequence of vertices are as follows .
* upwards , e.g. from @xmath291 to @xmath292 .
this happens if @xmath293 .
* leftwards , e.g. from @xmath291 to @xmath294 .
this happens if @xmath295 .
* diagonally to the nw , e.g. from @xmath291 to @xmath296 .
this happens if @xmath297 .
these are three of the six possible transitions in a triangular lattice , whose families of lines are horizontal , vertical , and diagonal with slope @xmath90 ( the set of lattice points is still @xmath0 ) . for @xmath287 to be a flbf ,
it is necessary and sufficient that each vertex @xmath298 in this sequence has local environment @xmath103 .
calling @xmath103 vertices `` open '' and @xmath104 vertices `` closed '' , this sequence defines a bi - infinite oriented ( triangular lattice ) nearest - neighbour path such that @xmath299 .
it follows that each @xmath300 and in particular @xmath301 .
now continue exactly as in the proof of ( * ? ? ? * theorem 3.12 ) to obtain ( c ) , substituting proposition [ prp : ctrichotomy ] for ( * ? ? ?
* proposition 3.8 ) . a similar argument ( or just symmetry ) gives part ( a ) , and part ( b ) follows immediately from the above argument and proposition [ prp : ctrichotomy ] . for @xmath87
define @xmath302 by @xmath303 the above sequence @xmath298 traces out @xmath304 sequentially as a ( triangular lattice ) path .
clearly @xmath128 lies in or above the set @xmath304 , since @xmath305 implies that @xmath306 .
we will need the following later .
[ cor : lbfcriterion ] for model [ exa : ne_sw ] @xmath26 , when @xmath87 the flbf @xmath307 satisfies : @xmath308 define @xmath309 to be the right hand side of . since @xmath310 is finite ( as in the proof of proposition [ prp : cphasetransition ] ) and
since if @xmath311 then @xmath312 for some @xmath313 , it follows that @xmath314 is well defined . our objective is to show that @xmath315 .
note that @xmath316 is decreasing , since e.g. if @xmath317 and @xmath318 then there exists @xmath319 $ ] such that @xmath320 , and then @xmath321 .
clearly then the origin can not connect to anything below @xmath316 , so @xmath322 .
we have seen in the proof of proposition [ prp : cphasetransition ] that @xmath323 for every @xmath324 . since @xmath306 and @xmath325 , we get @xmath326 . therefore @xmath327 . as in the proof of proposition
[ prp : cphasetransition ] , @xmath328 or @xmath329 ( depending on the sign of @xmath324 ) via the path @xmath298 .
since @xmath323 , we get that @xmath330 , and the result follows .
we will need a corresponding result for an infinite @xmath76 cluster . indeed , by (
* theorem 3.12 ) there exists a decreasing function @xmath331 with @xmath332 ( in @xcite , @xmath333 is a bubf , so @xmath60 is a flbf ) .
the proof of the following proceeds just as in that of corollary [ cor : lbfcriterion ] , so will be omitted .
[ cor : lbfcriterionforb ] consider model [ exa : ne_sw ] @xmath26 with @xmath87 .
if @xmath76 is infinite , then @xmath332 for a decreasing flbf @xmath331 satisfying : @xmath334 as @xmath21 increases from 0 to 1 , simulation suggests that the cluster @xmath76 can change from infinite to finite and back again arbitrarily many times , while holes " in @xmath1 can expand and contract .
nevertheless , both clusters have a kind of monotonicity , as in theorem [ thm : orthant_monotone ] .
_ proof of theorem [ thm : orthant_monotone ] .
_ couple the environments for all @xmath21 as in ( [ eq : coupling ] ) so that as @xmath21 decreases we switch @xmath103 to @xmath104 . due to proposition
[ prp : cphasetransition ] , the conclusion of the theorem is trivial if @xmath335 .
so assume @xmath336 , and let @xmath200 be the flbf with @xmath337 .
this implies that every site @xmath338 satisfies @xmath339 and @xmath340 .
therefore @xmath341 for all such @xmath38 , which implies that regardless of @xmath342 , there can be no open path in @xmath343 from @xmath128 to any site below @xmath344 .
in other words , @xmath345 , and by definition of @xmath82 also @xmath346 . this implies that @xmath347 , which establishes the desired conclusion except for showing that the inequality is strict . it remains to prove strictness , ie .
@xmath348 . let @xmath349 .
. then @xmath351 and @xmath352 are @xmath353-measureable random sets , and @xmath354 .
conditional on @xmath353 , for any @xmath353-measureable subset @xmath355 of @xmath356 we have that @xmath357 are i.i.d .
@xmath358 $ ] random variables .
in particular @xmath359 are i.i.d .
@xmath358 $ ] random variables under this conditioning .
thus , @xmath360 almost surely , so @xmath361 .
this says that ( almost surely ) there exists @xmath362 with @xmath363 $ ] .
therefore @xmath364 , so @xmath365 , hence @xmath366 .
the following is a version of theorem [ thm : orthant_monotone ] for the clusters @xmath367 .
the proof is similar , and will be omitted .
[ cor : orthantb_monotone ] consider model [ exa : ne_sw ] @xmath26 , with the coupling ( [ eq : coupling ] ) .
let @xmath86 .
then @xmath368 @xmath13-a.s .
on @xmath369 .
in this section we state without proof a number of results about the otsp model @xmath370 that follow using the methods of @xcite for two dimensional oriented percolation models . an expanded version of this paper , with
the omitted proofs included , is available from the authors upon request . in this model
we have local environment @xmath371 with probability @xmath21 , and @xmath372 with probability @xmath105 , both on the triangular lattice described in section [ sec : orthant_main ] .
recall that forward clusters in this model are denoted @xmath270 , and backward clusters @xmath278 . the natural coupling ( [ eq : coupling ] )
gives a probability space on which the sets @xmath373 are increasing in @xmath21 almost surely , so @xmath374 giving the critical value @xmath375 .
the estimated value is @xmath376 ( see de bell and essam @xcite or jensen and guttmann @xcite ) .
the best rigorous bounds that we have found in the literature are @xmath110 , where the latter comes from the fact that @xmath377 ( the latter , referring to oriented percolation on the square lattice , is in balister et al @xcite ) .
similarly , if @xmath378 denotes the critical threshold for ( un - oriented ) triangular site percolation , then @xmath379 ( see hughes @xcite ) .
the lower bound of @xmath380 comes from @xcite based on estimates of the critical value for the model @xmath381 . in section [ sec : pcbounds ] we improve this lower bound to @xmath382 , by finding arbitrarily large connected circuits in the dual model @xmath381 when @xmath383 in order to describe the shape of an infinite @xmath270 cluster , define @xmath384 and @xmath385 .
the following proposition is proved using subadditivity of quantities related to @xmath386 .
minor modifications arise from the proofs in @xcite , because the latter treats oriented bond percolation on the square lattice , while we need oriented site percolation on the triangular lattice .
[ prp : opslope ] for the percolation model @xmath370 with @xmath387 , there exists @xmath388 such that almost surely on the event @xmath389 , the upper and lower boundaries of @xmath390 have asymptotic slopes @xmath391 and @xmath392 respectively . in other words , @xmath393 and @xmath394 almost surely as @xmath395 .
since @xmath396 is bounded below by a sum of independent geometric@xmath397 random variables , we get the inequality @xmath398 .
the following two additional lemmas can be proved as in @xcite .
[ lem : opslope ] @xmath89 is continuous and strictly decreasing in @xmath399 , with @xmath400 as @xmath401 .
let @xmath402 , which measures the furthest diagonal line reached by the forward cluster of the origin . more generally , if @xmath403 , let @xmath404 .
note that @xmath405 .
[ lem : dual_tail ] if @xmath406 , then there exist constants @xmath152 , @xmath407 such that @xmath408 . on the event @xmath389 ,
let @xmath409}:x_n\in \mathbf{c}_o , |\mathbf{c}_{x_n}|=\infty , x_n^{[1]}=-n\}$ ] for @xmath410 .
the following result says that @xmath411 has the same asymptotic slope as the upper boundary of @xmath390 .
[ cor : a_op_slope ] for the percolation model @xmath370 with @xmath406 , @xmath412 almost surely on the event @xmath389 .
let @xmath406 .
since @xmath413 for every @xmath127 , and @xmath414 , it suffices to prove that for each @xmath415 , @xmath416 let @xmath417 .
then @xmath418 for all sufficiently large @xmath127 , so we may find @xmath419 such that @xmath418 for every @xmath420 .
it will therefore suffice to show that @xmath421 call @xmath422 the _ generation _ of a point @xmath423 . along any open path in this model ,
the generation increases at each step , by 1 or by 2 .
fix @xmath415 , and @xmath424 .
suppose that @xmath425 , and @xmath426 , and @xmath427 .
let @xmath428 be the triangle with vertices @xmath429 , @xmath430 , and @xmath431 , and let @xmath432 be the triangle with vertices @xmath433 , @xmath434 , and @xmath128 .
any open path @xmath435 from @xmath128 to @xmath436 enters @xmath428 along the side @xmath437 .
since @xmath438 , @xmath437 has slope @xmath439 , so entry to @xmath428 occurs no later than generation @xmath127 .
therefore from generation @xmath127 through @xmath440 the path @xmath435 lies entirely within @xmath428 .
consider the lattice point @xmath441 on this path whose generation first exceeds @xmath442 .
then @xmath443 , so @xmath444 , and hence @xmath445 .
the generation of @xmath436 is at least @xmath446 larger than that of @xmath216 , so @xmath447 . on the other hand , @xmath448 , so @xmath449 .
therefore we can not have @xmath450 .
in other words , @xmath451 . we have shown that for @xmath424 , @xmath452 there are at most @xmath453 lattice points in @xmath454 , so by lemma [ lem : dual_tail ] , the probability of this event is at most @xmath455 , which sums
. therefore follows by borel - cantelli .
to complete this circle of results , it simply remains to show that the remaining parts of theorem [ thm : orthant_main ] follows from the results of the previous section .
_ proof of theorem [ thm : orthant_slope ] .
_ by proposition [ prp : cphasetransition ] , when @xmath87 , @xmath82 is bounded below by a flbf @xmath307 . as in corollary
[ cor : lbfcriterion ] , we construct @xmath456 .
so @xmath457 and for @xmath458 , we may define @xmath459}:z_n\in { \bf c}_{y},|{\bf c}_{z_n}|=\infty , z_n^{[1]}=-n\}$ ] .
corollary [ cor : lbfcriterion ] now implies that @xmath460 , so by corollary [ cor : a_op_slope ] and translation invariance , we get that @xmath461 , where @xmath89 is as in proposition [ prp : opslope ] . by that result and lemma [
lem : opslope ] , @xmath400 as @xmath462 .
this establishes the desired statements for the northwest boundary .
the results for the southeast boundary follow by symmetry .
_ proof of theorem [ thm : mphasetransition ] .
_ by ( * ? ?
* theorem 4.9(b ) ) , it simply remains to show that @xmath84 is a.s .
finite when @xmath87 .
let @xmath87 .
then as above , @xmath82 is bounded below by @xmath200 which satisfies @xmath463 and by symmetry in the model @xmath464 also @xmath465 for @xmath458 . similarly by corollary [ cor : lbfcriterionforb ] , @xmath76 is bounded above by @xmath60 , and @xmath466 by symmetry in the dual oriented site percolation model
. then also @xmath467 by symmetry in the model @xmath464 . since @xmath468 it follows that @xmath469 for all but finitely many @xmath127 , so @xmath84 is finite
. see figure [ fig : orthant_c_o_b_o ] .
we define a nearest - neighbour ( self - avoiding ) path @xmath470 to be a _ @xmath471-path _ if @xmath472 and @xmath473 for each @xmath474 . for @xmath475 and @xmath269
we write @xmath476 for the set of @xmath471-paths @xmath477 such that @xmath478}{\rightarrow}\infty$ ] and @xmath478}/x_n^{[1]}{\rightarrow}\sigma$ ] as @xmath395 .
let @xmath479 .
similarly define @xmath480 [ resp .
@xmath481 as the set of @xmath471-paths such that @xmath478}/x_n^{[1]}{\rightarrow}\sigma$ ] and @xmath482}{\rightarrow}-\infty$ ] [ resp .
write @xmath484 and @xmath485 .
the following lemma gives a strategy for generating better one - sided bounds for the critical point for the orthant model , and hence also for oriented site percolation on the triangular lattice .
[ lem : lookforslopes ] consider the model @xmath26 , with @xmath486 .
suppose that for some @xmath487 , there exists a path @xmath488 almost surely
. then @xmath489 .
suppose that @xmath87 . then @xmath82 is bounded below by @xmath200 , where @xmath490 as in proposition [ prp : opslope ] .
it follows that for any @xmath491 , the set @xmath492 } < 0,x_i^{[2]}/x_i^{[1]}>\sigma\}$ ] is finite . since @xmath493
, this implies that there can be no infinite path @xmath488 for @xmath487 .
let @xmath494 [ lem : cslope ] for the model @xmath495 , @xmath1 contains ( self - avoiding ) @xmath496-paths @xmath497 and @xmath498 . moreover , environment - connected components of @xmath499 between these two paths are finite .
define the _ nw path _ from @xmath38 to be the path obtained by always choosing to follow @xmath239 when possible and otherwise following @xmath238 .
the asymptotic slope of the nw path ( call this path @xmath500 ) from the origin is @xmath501 , which establishes the first claim . for the second claim ,
consider the path @xmath502 from the origin that evolves as follows .
whenever the environment at the current location is @xmath239 , the path follows this west arrow .
whenever the environment at the current location @xmath38 is @xmath103 , the path follows the east arrow to @xmath503 if the environment at @xmath503 is also @xmath103 , otherwise the path follows the north arrow . by definition
this path never backtracks , so it is self - avoiding . after each northern step taken by the path ,
the environment thereafter encountered has never been viewed before , hence each northern step constitutes a renewal .
thus the path moves upwards through the set of horizontal bands @xmath504 , @xmath505 . let @xmath506 be the point where our path first enters the @xmath507th band
. then we can represent @xmath508 as follows : @xmath509 then the @xmath510 are i.i.d . with @xmath511
it follows that @xmath512=\sum_{k=0}^\infty kp^{k+1}(1-p)-\sum_{k=1}^{\infty}k(1-p)^kp=\frac{p^2}{1-p}-\frac{1-p}{p}=g_1(p),\ ] ] and that the asymptotic slope of the path @xmath502 is @xmath513 as claimed . by construction ,
a vertex @xmath514 never lies strictly to the left of any vertex @xmath515 ( the paths @xmath502 and @xmath500 may meet , but not cross ) .
an elementary comparison of the asymptotic slopes shows that @xmath516 eventually lies strictly to the right of @xmath517 , in the sense that there exists some @xmath518 such that for all @xmath519 , @xmath520 . trivially also by construction @xmath521}{\rightarrow}\infty$ ] and similarly for @xmath522}$ ] . to prove the last claim observe that the nw path from each @xmath523 eventually hits the path @xmath500 ( since nw paths from any two vertices eventually meet ) , and that these nw paths are all in @xmath1 .
note that it follows immediately from this result that we can find similar paths ( up to symmetry ) in @xmath1 for the models containing this as a submodel .
this implies results for other models , such as the following improvement on the lower bound on @xmath71 in ( * ? ? ?
* theorem 4.12 ) .
[ cor : corthant ] in model [ exa : ne_sw ] @xmath26 , @xmath524 for @xmath525 . therefore @xmath526 .
since model [ exa : ne_sw ] contains @xmath527 , by lemma [ lem : cslope ] and symmetry , @xmath1 contains a self - avoiding path @xmath528 .
so @xmath524 for any @xmath529 such that @xmath530 , by proposition [ prp : cphasetransition ] and lemma [ lem : lookforslopes ] . the condition @xmath530 holds for @xmath531 , where @xmath532 , i.e. for @xmath533 ( the condition is actually equivalent to @xmath534 since @xmath535 ) .
the true value we are aiming for is @xmath72 , which is estimated as @xmath536 .
the path @xmath502 above is defined ( see lemma [ lem : cslope ] ) in terms of expected horizontal displacements that can be achieved for paths consistent with the environment , staying within a horizontal band of width 1 .
in fact , there is a sequence of estimates that in principle should converge to the true value . repeat the above argument , but using bands of width @xmath537 . the resulting bounds @xmath538 should converge to @xmath72 .
we ll content ourselves with computing @xmath539
. _ proof of theorem [ thm : otsp_pc ] .
_ in analogy with the proof of lemma [ lem : cslope ] , we let @xmath540 and @xmath510 , where @xmath541 we set @xmath542/2 $ ] so that the path using horizontal bands has slope @xmath543 .
the strategy of corollary [ cor : corthant ] works with @xmath544 in place of @xmath545 , provided @xmath546 .
therefore we are left to compute @xmath547 .
this is done below , ultimately leading to @xmath548 where @xmath549 using newton s method ( with a starting point of @xmath550 , in the computer algebra system _ maxima _ ) gives @xmath551 , which completes the proof . in the remainder of this section ,
we show how to obtain .
we write @xmath552 for the pair of local environments at @xmath553 and @xmath554 .
first consider the case @xmath555 .
our first goal will be to compute @xmath556.\ ] ] paths are not actually unique , but we take the convention that in this situation the path moves @xmath238 from @xmath553 to @xmath554 , and then considers its next move . if @xmath557 ( the @xmath558 simply means an arbitrary environment ) the path moves @xmath559 to @xmath106 .
if @xmath560 then we have hit an impassable obstacle , and the path has no choice but to exit using @xmath238 , in which case @xmath561 .
the remaining possibility is @xmath562 .
there could in fact be a sequence of such pairs , followed either by a @xmath563 or by @xmath564 .
for example , @xmath565 or @xmath566 the first case also represents an impassable obstacle , so our path chooses to exit using @xmath238 , making @xmath561 , in the second case , we move @xmath559 , @xmath267 , take a sequence of @xmath559 s , and then move @xmath238 to reach the top row of the band again .
this description makes it clear that there is a renewal structure here , with the construction starting afresh every time we reach a new @xmath103 environment on the top row of the band . to formalize this ,
let @xmath567 be the site of the @xmath568th such new @xmath103 environment reached by our path ( with @xmath569 corresponding to the initial environment ) .
if @xmath570 denotes the total number of such environments reached , then we take @xmath571 , @xmath572 .
the dynamics are that @xmath573 let @xmath574 and @xmath575 denote the probability of encountering a @xmath564 before a @xmath563 .
let @xmath576 denote the probability that @xmath577 .
then @xmath578=p+p(1-p)\sum_{i=1}^\infty ( i+1)[p(1-p)]^{i-1}p^2=\theta+\frac{\alpha p(1-p)}{1-p(1-p)}\ ] ] and the renewal structure implies that @xmath5791_{j\le j}\big]=\sum_{j=0}^\infty { \mathbb{e}}[z_1]\theta^j=\frac{{\mathbb{e}}[z_1]}{1-\theta}.\ ] ] the second case we consider is that @xmath580 , so our next goal is to compute @xmath581.\ ] ] if @xmath582 then we travel @xmath568 steps @xmath559 and then @xmath238 and find ourselves back in the situation just considered . while if @xmath583 then we are blocked to the right , and instead travel @xmath238 and then follow @xmath239 s till reaching a @xmath103 on the top row of the band , at which point we exit from the band via that @xmath238 .
this gives us the expression @xmath584^{j-1}p^2 + \alpha \mu_{{\begin{picture } ( , ) \put(1.5,-3){${\scriptstyle \rightarrow}$ } \put(0,.5){${\scriptstyle \uparrow}$ } \end{picture}\hspace{.35 cm } } } ^{{\begin{picture } ( , ) \put(1.5,-3){${\scriptstyle \rightarrow}$ } \put(0,.5){${\scriptstyle \uparrow}$ } \end{picture}\hspace{.35 cm } } } -(1-\alpha)\sum_{i=1}^\infty i(1-p)^{i-1}p = \alpha \mu_{{\begin{picture } ( , ) \put(1.5,-3){${\scriptstyle \rightarrow}$ } \put(0,.5){${\scriptstyle \uparrow}$ } \end{picture}\hspace{.35 cm } } } ^{{\begin{picture } ( , ) \put(1.5,-3){${\scriptstyle \rightarrow}$ } \put(0,.5){${\scriptstyle \uparrow}$ } \end{picture}\hspace{.35 cm } } } + \frac{\alpha}{1-p(1-p)}-\frac{1-\alpha}{p}.\ ] ] the third case of interest is @xmath585.\ ] ] we start out by following @xmath239 along the bottom row , till reaching a @xmath103 site , at which point we go @xmath238 to the top row .
if the site so reached is a @xmath104 then we again proceed @xmath239 till reaching a @xmath103 on the top row , at which point we exit from the band via the @xmath238 . on the other hand ,
if the first site reached in the top row is a @xmath103 then we have the opportunity to regain some lost ground
. we step @xmath559 along the top row as long as possible , and only go @xmath238 just before reaching a @xmath104 .
for example , @xmath586 has @xmath587 .
this leads to the general expression @xmath588\\ & = -\frac{1}{p}-\frac{1-p}{p}+\sum_{j=0}^\infty jp^{j+2}\sum_{k = j+2}^\infty ( 1-p)^k + \sum_{k=1}^\infty ( k-1)p^2[p(1-p)]^{k-1}\\ & = -\frac{2-p}{p}+\frac{\alpha(1-p)}{1-p(1-p)}[p+(1-p)^2].\end{aligned}\ ] ] the fourth and final case of interest is @xmath589.\ ] ] once again , we must go @xmath239 till reaching the first @xmath103 on the bottom row of the band , and then go @xmath238 . if this leads to a @xmath104 vertex , then
all we can do is head @xmath239 on the top row , till reaching a @xmath103 vertex , at which point we may exit via @xmath238 . however ,
if we reach the top row at a @xmath103 vertex then we can regain lost ground by heading @xmath559 .
either this ends as in the previous case , before making it all the way back to @xmath554 .
or we do make it back to @xmath554 this way , in which case we find ourselves back in the first case examined above . for example , @xmath590 has @xmath591 .
this leads to the general expression @xmath592\\ & = -\frac{2-p}{p}+\frac{\alpha}{1-p(1-p)}[1+(1-p)^3]+\alpha \mu_{{\begin{picture } ( , ) \put(1.5,-3){${\scriptstyle \rightarrow}$ } \put(0,.5){${\scriptstyle \uparrow}$ } \end{picture}\hspace{.35 cm } } } ^{{\begin{picture } ( , ) \put(1.5,-3){${\scriptstyle \rightarrow}$ } \put(0,.5){${\scriptstyle \uparrow}$ } \end{picture}\hspace{.35 cm } } } .\end{aligned}\ ] ] thus , @xmath593 is equal to @xmath594=&p^2\mu_{{\begin{picture } ( , ) \put(1.5,-3){${\scriptstyle \rightarrow}$ } \put(0,.5){${\scriptstyle \uparrow}$ } \end{picture}\hspace{.35 cm } } } ^{{\begin{picture } ( , ) \put(1.5,-3){${\scriptstyle \rightarrow}$ } \put(0,.5){${\scriptstyle \uparrow}$ } \end{picture}\hspace{.35 cm } } } + p(1-p)\mu_{{\begin{picture } ( , ) \put(1.5,-3){${\scriptstyle \rightarrow}$ } \put(0,.5){${\scriptstyle \uparrow}$ } \end{picture}\hspace{.35 cm } } } ^{{\begin{picture } ( , ) \put(0.5,3.2){${\scriptstyle \leftarrow}$ } \put(6.3,-0.5){${\scriptstyle \downarrow}$ } \end{picture}\hspace{.35 cm } } } + p(1-p)\mu_{{\begin{picture } ( , ) \put(0.5,3.2){${\scriptstyle \leftarrow}$ } \put(6.3,-0.5){${\scriptstyle \downarrow}$ } \end{picture}\hspace{.35 cm } } } ^{{\begin{picture } ( , ) \put(1.5,-3){${\scriptstyle \rightarrow}$ } \put(0,.5){${\scriptstyle \uparrow}$ } \end{picture}\hspace{.35 cm } } } + ( 1-p)^2\mu_{{\begin{picture } ( , ) \put(0.5,3.2){${\scriptstyle \leftarrow}$ } \put(6.3,-0.5){${\scriptstyle \downarrow}$ } \end{picture}\hspace{.35 cm } } } ^{{\begin{picture } ( , ) \put(0.5,3.2){${\scriptstyle \leftarrow}$ } \put(6.3,-0.5){${\scriptstyle \downarrow}$ } \end{picture}\hspace{.35 cm } } } .\end{aligned}\ ] ] this gives us that @xmath595 here the denominator is equal to @xmath596 solving for @xmath597 is equivalent to finding @xmath598 , where @xmath599 as claimed .
holmes s research is supported in part by the marsden fund , administered by rsnz .
salisbury s research is supported in part by nserc .
both authors acknowledge the hospitality of the fields institute , where part of this research was conducted .
|
we consider connectivity properties and asymptotic slopes for certain random directed graphs on @xmath0 in which the set of points @xmath1 that the origin connects to is always infinite .
we obtain conditions under which the complement of @xmath1 has no infinite connected component . applying these results to one of the most interesting such models
leads to an improved lower bound for the critical occupation probability for oriented site percolation on the triangular lattice in 2 dimensions .
# 1 # 1 # 1 # 1
|
in the last decade , @xmath1010 planets and planet candidates with estimated masses @xmath813 m@xmath6 have been imaged in orbit around young stars and brown dwarfs ( e.g. * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) . in total ,
@xmath1030 companions with estimated masses @xmath825 m@xmath6 have been imaged .
( see http:@xmath11exoplanet.eu for a compilation of these objects . )
these discoveries have provided a wealth of new information about young giant planets , as well as some surprises .
prior to these detections , models predicted that young gas giant planets at moving group ages ( 10 - 300 myr ) would likely have cool photospheres with prominent methane absorption features @xcite , i.e. that these objects would be spectral analogs to t - type brown dwarfs .
however , all known directly - imaged planets at these ages ( specifically 2mass 1207b and the hr 8799 planets , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) have lacked methane absorption and show extremely red colors , likely due to dust clouds and/or non - equilibrium chemistry in their atmospheres @xcite .
additionally , all of these companions except for @xmath0 pic b @xcite , hr 8799e @xcite , and lkca 15b @xcite lie at projected separations greater than 20 au , considerably wider than giant planets in our own solar system .
such widely separated companions pose a challenge for the accepted model of core - accretion formation , which likely formed the closer - in ( @xmath810 au ) population of planets detected to date via radial velocity studies ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
however , given that only @xmath1010 such companions have been imaged to date , it is perhaps premature to make statements based on such a small sample . thus , it is a priority to discover additional companions as well as to constrain on the distributions of their semimajor axes , eccentricities , masses , etc . in the last decade , a number of deep , adaptive - optics aided surveys with sample sizes @xmath420 stars have been completed at 8-m telescopes to search for additional planetary companions .
many of these have been conducted in the 1.6 @xmath12 m @xmath2-band or 2.2 @xmath12 m @xmath13-band @xcite , while others have focused further into the infrared ( 3.5 - 5 @xmath12 m ) in the @xmath14 , @xmath15 , or @xmath16 bands @xcite . a number of very large scale surveys ( @xmath4100 stars ) are ongoing or recently completed , including the nici campaign at gemini - south ( liu et al .
2010 , this publication , wahhaj et al .
2013a , nielsen et al .
2013 , wahhaj et al .
2013b ) , the naco large program using naco at the vlt ( buenzli et al .
2010 ) , seeds ( strategic exploration of exoplanets and disks with subaru ) using hiciao at subaru @xcite , and the international deep planet survey ( idps ) using primarily gemini and keck @xcite .
the host stars of currently known directly - imaged planets fall into three categories : ( 1 ) members of young ( @xmath1710 myr ) star - forming clusters or ob associations ( e.g. taurus and upper sco ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) , ( 2 ) members of nearby young moving groups ( ages of 10300 myr , e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) , and ( 3 ) unassociated nearby young stars ( e.g. * ? ? ?
* ) . of these three categories of targets ,
moving group objects are particularly compelling targets for direct imaging searches .
the extremely young ages of star - forming clusters translates into considerably brighter planets , but at distances @xmath18140 pc the inner working angles of current instruments generally only allow detection of companions at projected separations @xmath1950 au .
unassociated nearby young stars often do not have well constrained ages , a limitation for estimating the mass of any companion detected and for deriving statistics for the survey sensitivities . moving group stars
provide a unique nearby young sample with well - constrained ages and distances .
we have observed 80 young moving group stars as a part of a dedicated science campaign using the near - infrared coronagraphic imager ( nici ) at the 8.1 m gemini south telescope @xcite .
nici is a dedicated adaptive optics ( ao ) instrument tailored expressly for direct imaging of exoplanet companions , combining several techniques to attenuate starlight and suppress speckles for direct detection of faint companions to bright stars : ( 1 ) lyot coronagraphy , ( 2 ) dual - channel imaging for spectral differential imaging ( sdi ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) , and ( 3 ) operation in a fixed cassegrain rotator mode for angular differential imaging ( adi ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) . while each of these techniques has been used individually in large planet - finding surveys ( e.g. *
* ; * ? ? ?
* ) , the nici campaign is the first time all three have been employed simultaneously in a large survey . from 2008 december to 2012 september , the nici planet - finding campaign @xcite obtained deep , high - contrast ao imaging of a carefully selected sample of over 200 young , nearby stars .
over the course of the campaign , we discovered 4 new brown dwarf companions to young stars : pz tel b @xcite , cd -35 2722b @xcite , hd 1160c @xcite , and hip 79797bab @xcite . here
we report results from the subsample of 80 young stars that are members of the @xmath0 pic , tw hya , ab dor , tucana - horologium , and hercules - lyra moving groups .
moving groups are associations of young stars ( 10300 myr ) that are unconnected to regions of ongoing star - formation . these associations were not discovered until the late 1990 s , as moving group members are often dispersed across a wide part of the sky ( e.g. * ? ? ?
* ; * ? ? ?
moving group members are identified by a combination of youth indicators ( li absorption , high x - ray luminosity , etc . ) and space motion coincident with other cluster members . we have focused our survey on 5 young moving groups with members generally within @xmath1060 pc of the earth .
the star tw hya was the first pre - main sequence ( henceforth pms ) star identified outside of a star - forming region , initially identified by @xcite as an isolated t tauri star . @xcite and @xcite identified 4 additional t tauri stars within 10 degrees of tw hya . @xcite
were the first to label these stars as the tw hya association , based on their strong lithium absorption features , x - ray fluxes , and similar hipparcos parallaxes .
since then , @xmath420 tw hya members have been identified ( see * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
based on lithium absorption and x - ray flux strength , the association is assigned an age of @xmath2010 myr @xcite .
the mean distance of the tw hya association is 48@xmath513 pc @xcite .
the circumstellar disk around the nearby a star @xmath0 pic was first imaged by @xcite , leading to the identification of @xmath0 pic as a young star with planet formation having occurred in the recent past . @xcite found two additional m type stars ( gj 799 and gj 803 ) with matching galactic space motions to @xmath0 pic .
@xcite cemented the existence of the @xmath0 pic moving group with the confirmation of 18 additional members on the basis of their galactic space motions . over 60 members of the @xmath0 pic moving group
have been identified to date @xcite . from the color - magnitude diagram placement of these stars as well as their lithium absorption ages , an age of @xmath2012 myr
is estimated for this moving group @xcite , with a mean distance of 31@xmath521 pc @xcite .
young stars are often far - ir excess sources . based on this fact , @xcite searched the hipparcos catalog for stars with similar space motions and within a 6 degree radius of 24 stars detected at 60 @xmath12 m with iras . from this search
, they identified @xmath1010 stars with distances of @xmath1045 pc and ages of @xmath1030 yr , which they named the tucana association .
@xcite found a group of @xmath1010 stars through x - ray emission and ground - based spectroscopy that showed youth indicators and are associated with the previously identified isolated t tauri star ep eri , which they titled the horologium association .
as the stars in these two associations share the same space motions , ages , distances , and volume density , they are now considered to be part of the same association @xcite . over 60 stars have been identified to date in the tucana - horologium association @xcite , with a mean distance of 48@xmath57 pc @xcite .
the star ab dor is notable as an ultrafast rotator which is also extremely x - ray active , at a distance of only 15 pc and an age of @xmath10100 myr @xcite .
ab dor itself , in fact , is a quadruple system with a close m - dwarf companion and a wider separation m - dwarf binary @xcite . @xcite
identified @xmath1030 nearby star systems with similar space motions to ab dor as well characteristics of youth , which they designated the ab dor moving group . over 50 stars
have been identified to date in the ab dor moving group @xcite , with a mean distance of 34@xmath526 pc @xcite .
@xcite first identified 4 young solar analogues with similar space motions towards hercules .
@xcite identified a further 15 late - type stars with similar space motions and gave the whole complex the name hercules - lyra .
the existence of the hercules - lyra association was initially disputed , as the candidate members possessed a wide age spread inconsistent with a single moving group , with some of the initially identified stars possessing ages ( derived from lithium absorption and chromospheric activity ) much younger or older than the average association age of @xmath10200 myr .
@xcite confirmed the existence of the hercules - lyra association and winnowed down the 27 initial candidate members to 10 confirmed members with an average distance of 20@xmath510 pc and age of @xmath10200 myr .
we observed 80 stars in nearby young moving groups as part of the nici campaign 14 stars from the tw hya association , 30 stars from the @xmath0 pic moving group , 12 stars from the tucana - horologium association , 19 stars from the ab dor moving group , 4 stars from the hercules - lyra association , and 1 star ( bd + 1 2447 ) which is either a hercules - lyra or ab dor moving group member .
the survey sample was selected from a larger sample of moving group stars compiled from the literature .
observations were prioritized according to the probability of detecting a planet around a given survey star @xcite , as predicted by monte carlo simulations similar to those described in section 5.1 .
the survey sample is listed in table [ tab : properties ] and is plotted as a function of age , distance , and spectral type in figure [ fig : mg_sample ] .
histograms of the spectral type and distance distributions are presented in figure [ fig : mg_hists ] .
the majority ( 85@xmath9 ) of sample stars have ages less than 100 myr and distances less than 60 pc .
the median distance is 39.8 pc .
we observed 1 b star , 7 a stars , 11 f stars , 5 g stars , 23 k stars , and 33 m stars .
thus , our moving group sample is primarily composed of lower mass stars .
observations of our survey sample are listed in table [ tab : observations ] .
we only report observations which contain at least 10 individual images in the adi or asdi sequences , in order to achieve the field rotation needed by our adi processing pipeline to obtain reliable detections ( see section [ sec : obs_strategy ] for details ) .
nici was specifically designed to provide the high contrasts necessary to directly image young extrasolar giant planet .
nici s 85-element curvature ao system provides ao correction of @xmath1030 - 45@xmath9 strehl in @xmath2 band @xcite .
the ao beam is then reflected into the science camera , where it passes through a partially transparent focal plane mask . the focal plane mask is a flat - topped gaussian , which suppresses @xmath499.5@xmath9 of the incoming starlight ( @xmath1ch@[email protected] mag , @xmath1@[email protected] mag ; * ? ? ?
* ) , thus reducing scattered light from the central star and increasing the attained contrast . a variety of these semi - transparent masks are available for use with nici ; we utilized the 0.32@xmath22 radius mask for nici campaign observations , thus providing an effective inner working angle of 0.32@xmath22 for faint companions , although tight stellar companions can still be detected in the innermost regions .
the partially - transparent mask also allows us to attain very precise photometry and astrometry , as we can simultaneously obtain unsaturated images of both the primary and faint companions .
the beam then passes through a hard - edged pupil stop , which reduces diffracted light from psf artifacts associated with the gemini - south secondary mirror . for observations in dual - channel mode ,
the beam is split using a dichroic and passes into two separate science cameras . for the majority of the campaign ,
a 50/50 beamsplitter was utilized , resulting in the loss of half of the incoming light to each channel , but from the beginning of 2012 , this beamsplitter was replaced by an h / k dichroic , boosting throughput when imaging simultaneously in these two filters .
different filters may be chosen for each science camera ; thus nici s 2-camera capability can provide simultaneous color information .
both cameras have fields of view of 18@xmath2318@xmath22 , with a platescale of 17.96 mas for the science camera using the 1.578 @xmath12 m @xmath24 filter ( henceforth `` blue channel or ' ' off - methane channel `` ) and a platescale of 17.94 for the science camera using the 1.652 @xmath12 m @xmath25 4% filter ( henceforth ' ' red channel `` or ' ' on - methane channel " ) for the science camera using the 1.578 @xmath12 m .
nici campaign observations were conducted in two separate modes : ( 1 ) single channel h - band adi ( angular differential imaging ) mode and ( 2 ) dual - channel methane band combined adi+sdi ( spectral differential imaging ) mode .
both sdi and adi techniques seek to distinguish real objects from speckles .
sdi achieves this by exploiting a spectral feature in the desired target ( e.g. the 1.6 @xmath12 m methane absorption feature observed in substellar objects with a t spectral type * ? ? ?
* ; * ? ? ?
images are taken simultaneously both within and outside the chosen absorption feature .
due to the simultaneity of the observations , the stellar point - spread functions in the two nici channels , including the coherent speckle patterns , are nearly identical .
in contrast , any faint companion with the chosen absorption feature is bright in one filter and faint in the other .
subtracting the two images thus removes the starlight and speckle patterns while a real companion with the chosen absorption feature remains in the image . in other words ,
the absorption band image acts as an ideal reference point spread function ( henceforth psf ) for the off - absorption band image .
utilizing a signature spectral feature of substellar objects can help distinguish between true methanated companions and likely background objects , e.g. a background object will be subtracted out by the sdi subtraction since it will not have methane absorption . however , this mode is sensitive even to companions without this absorption feature , as a real companion will appear fixed in separation relative to the star in both filters , while a speckle will modulate with the airy pattern and appear further from the star in the red filter relative to the blue filter .
adi employs a different strategy in order to decorrelate real companions from speckles . for adi observations ,
the rotator is left off at the cassegrain focus or set to follow the elevation angle at the nasmyth focus , allowing the telescope optics rotate relative to the sky . in a sequence of images taken at different parallactic angles ,
a real companion will move relative to the detector along with the sky , while the speckles will remain fixed . from a series of images ,
a reference psf can thus be constructed for and subtracted from each individual image , attenuating quasi - static speckle structure . combining both sdi and adi techniques ( henceforth asdi ) thus allows an even greater degree of speckle supression . in order to take advantage of both the higher contrast available within 1.5 using the asdi mode ( due to improved speckle suppression from the sdi subtraction ) and the improved sensitivity available outside of 1.5 with the adi mode ( due to the wider bandpass used during our adi observations ) , most nici campaign stars
were observed in both modes . for asdi mode
, we observed simultaneously in the off - methane ( central @xmath26=1.578 @xmath12 m ; width=0.062 @xmath12 m ; @xmath27 4% ) and on - methane ( central @xmath26=1.652 @xmath12 m ; width=0.066 @xmath12 m ; @xmath25 4% ) bands using nici s dual - channel imaging capability .
adi data were taken with the broadband @xmath2 filter in the blue channel ( central @xmath26=1.65 @xmath12 m , width=0.29 @xmath12 m ) stars fainter than @xmath2=8 mag were observed only in single - channel adi mode , as the contrast within 1.5 was similar to that achievable in the asdi mode .
stars close to the galactic bulge were only observed in asdi mode , as adi mode often yielded enormous numbers ( @xmath450 per field ) of background field objects .
typically , we obtained 20 minutes on - sky data in adi mode and 40 minutes on - sky data in asdi mode for each star .
observations were carefully scheduled in order to maximize field rotation while avoiding too much blurring during single exposures .
we aimed to obtain at least 5@xmath28 field rotation in adi mode observations and at least 15@xmath28 field rotation in asdi mode observations .
this ensures on - sky rotations of at least 3@xmath23fwhm of the psf at 5 separation from the primary in adi mode and at least 3@xmath23fwhm of the psf at 1 separation in asdi mode .
typical fwhms of the psf ranged between 3 - 4 pixels .
out of 68 stars with adi mode observations , all but 4 have at least one dataset with sky rotation @xmath45 degrees . out of 56 stars with asdi observations ,
all but 7 have at least one dataset with sky rotation @xmath415 degrees , and only one asdi observation has sky rotation @xmath810 degrees . for asdi
, individual exposure times were chosen to produce high s / n in the speckle halo while avoiding saturation in this region . in adi mode ,
exposure times of 4 to 60 s were used , allowing the halo to saturate if needed . for bright stars that saturate in the adi exposures
, short exposures were interleaved with deep exposures in order to provide unsaturated images of the star behind the partially transparent mask ( henceforth the `` starspot '' ) for accurate photometry .
all observations are processed using a custom pipeline described in @xcite . here
we briefly summarize procedures for both adi and asdi datasets ; some data processing steps pertain only to the asdi mode and are noted as such below . for all data , the pipeline first applies dark , flatfield , and distortion corrections . for adi data ,
all images are centroided and aligned to the first exposure in the sequence . for asdi data , images from the two science cameras
are then centroided and aligned .
datasets where the starspot is unsaturated are aligned using the starspot centroid position in each science exposure . for saturated images ,
the structure of the saturated psf is used to align the images @xcite .
specifically , the peak of the primary is still discernible as a negative image and can be used to centroid .
we have estimated that the centroiding accuracy of the saturated images is 9 mas by comparing these to the centroids of unsaturated short - exposure images obtained right before and after the long exposures .
image filters ( i.e. unsharp masking or catch filtering ) are applied frame - by - frame . in the asdi case ,
the red - channel image is subtracted from the blue - channel image for each science exposure .
a high - fidelity psf is built for the entire observation by median combining the stack of reduced images and then subtracted from each individual science exposure . finally , the reduced psf - subtracted images are registered , rotated to a common sky orientation , and stacked to produce a final image . in the asdi case ,
3 final output images are produced : the full subtracted reduction as well as single - channel adi reductions for the blue and red channel images respectively , which can be added to achieve deeper sensitivity .
this ensures that no planet candidates are missed due to spectral self - subtraction in the asdi mode .
in order to robustly measure the contrast achieved by our pipeline reductions , we generate 95%-completeness contrast curves following the method described in wahhaj et al .
( 2013b ) .
the 95%-completeness technique accounts for self - subtraction losses endemic to adi and sdi data , unlike simple measurements based solely on the noise level of the data .
briefly , the data are first pipeline - processed , rotationally misaligned ( derotated in the opposite direction of the actual parallactic angle rotation ) , and stacked to create a companion - free reduction .
the 1@xmath29 contrast curve is calculated from the standard deviation found in 3 pixel annuli as a function of separation from the primary star .
next , a set of 20@xmath29 simulated companions ( 1340 total simulated companions , at separations of 0.36@xmath22 to 6.3@xmath22 and uniformly distributed in azimuth in 67 concentric rings ) , produced by scaling the image of the primary star behind the partially - transparent mask , is inserted into the individual raw images , and the new data are re - reduced as before .
the 20@xmath29 simulated companions are recovered in the reduced data and used to evaluate the flux losses and artifacts in input contrasts due to the pipeline .
finally , the recovered companions ( now with flux loss effects and other pipeline artifacts incorporated ) are reinserted into the original reduction and scaled in intensity until they meet our detection criteria .
the contrast at which 95% of the simulated companions are detected is presented as the 95%-completeness contrast curve .
the 95% completeness contrast curves for the moving group sample are presented in figures [ fig : twhya_contrasts ] to [ fig : herlyr_contrasts ] .
tables of measured contrast are presented for the asdi subtracted reductions in table [ tab : contrastsasdi ] and for the adi reductions in table [ tab : contrastsadi ] . for our adi contrast curves , we convert measured contrast to maximum detectable apparent magnitude in table [ tab : appmagadi ] and minimum detectable mass in tables [ tab : minmassadi_dusty ] and tables [ tab : minmassadi_cond ] .
we interpolated from both the dusty and cond models of @xcite and @xcite using the maximum detectable apparent magnitudes , distance , and age of each stars to estimate the minimum detectable mass curves . at some point as they cool and dust condenses from their atmospheres , directly imaged exoplanets are predicted to transition from red , dusty l dwarf spectra ( dusty ) to t dwarf spectra with methane absorption features ( cond ) .
however , no directly imaged planet to date has yet to show strong methane absorption in the near - ir , with only weak methane absorption observed at longer wavelengths @xcite .
thus as this transition has not been observed , we choose here to present minimum detectable masses according to both of these models .
minimum detectable masses as a function of spectral type at 0.5@xmath22 , 1@xmath22 , 2@xmath22 , and 4@xmath22 are presented in figure [ fig : minmass_plots_dusty ] using the dusty models @xcite and in figure [ fig : minmass_plots_cond ] using the cond models @xcite . for the more conservative dusty model case , at 0.5@xmath22
we are sensitive to companions of @xmath3013 m@xmath6 for all but one star . at 2@xmath22
we are sensitive to companions with masses @xmath3010 m@xmath6 for all stars .
the minimum detectable mass varies by star ( according to spectral type , magnitude , distance , etc . )
but we are generally sensitive to @xmath185 m@xmath6 companions at 2@xmath22 around all sample stars
. we do not present minimum detectable masses in asdi subtracted mode here , as this requires knowledge of a potential companion s @xmath2-band spectrum .
for an example of such an analysis of adi self - subtraction as a function of radius , see nielsen et al .
( 2013 ) .
we found numerous candidate companions in our images .
candidates were first identified using an automated finding algorithm and then verified by eye . for the entire nici campaign sample
, candidate companions were found for @xmath1050@xmath9 of observed stars .
the vast majority of these objects are not expected to be true co - moving companions .
to test whether a candidate companion is co - moving with its parent star requires reobserving after enough time has elapsed for significant proper motion and/or parallactic motion of the star in the sky , ideally at the @xmath183 pixel ( @xmath1850 mas ) or greater level .
after identifying candidate companions in our reduced images , we first checked if any older archival data from vlt , hst or gemini were available .
in this manner , we were able to immediately identify a number of bright candidates as background objects .
astrometry for candidates observed at multiple epochs with nici as well as other telescopes is presented in tables [ table_comp2a ] and [ table_comp2b ] . for objects with _
nicmos observations , we retrieved data from the _ hst _ mast archive and used the mosaic files .
images taken at different telescope roll angles were subtracted to remove the slowly changing speckle pattern ( henceforth roll subtraction ) . for datasets with images taken at only one roll angle ,
images were rotated by 180@xmath28 and subtracted from themselves .
we typically performed roll subtraction without any subpixel alignment as most of the candidates were well outside the region where psf subtraction was important .
@xcite found the position of the star behind the nicmos coronagraph using acquistion images and slew vectors gleaned from hst engineering telemetry , and claim that the difference image diffraction spikes do not give an accurate measure of the star s position .
our candidate companions followed up with nicmos archival data are generally at wide separations ( @xmath42 ) and with large time baselines ( usually @xmath183 years ) relative to the nici epoch ; thus , we often did not require an extremely accurate knowledge of the central star position in order to determine if they were background objects . to see if the simpler method of using the diffraction spikes could be used , we tested this method on 10 stars in the @xcite sample by measuring the position of the same companions they reported .
we found a mean difference of 1.2 pixels from their positions . taking this to be entirely due to our centroiding method ,
we combine it in quadrature with their reported 1.05 pixel ( 0.08 `` ) uncertainty to calculate a total uncertainty of 1.6 pixels , or 0.12 '' .
data from gemini - niri were reduced using a custom adi script @xcite .
due to saturation of the primary stars , we estimate our astrometric uncertainty to be @xmath102 pixels , or 0.044 " .
candidates within 400 au from the star and not in dense stellar fields that were not confirmed or rejected as common - proper motion companions using archival data were reobserved with nici .
nici astrometry was measured relative to the unsaturated starspot position in either the science or short exposures .
the uncertainties in the separation and pa are estimated to be 0.009(0.5 pixel ) and 0.2@xmath28 respectively , when the primary is unsaturated , and 0.018(1 pixel ) and 0.5@xmath28 when the primary is saturated ( wahhaj et al .
2013b ) . from the proper motions and parallaxes of our mg sample stars and pinning to the nici first epoch position
, we can calculate the expected motion relative to the primary star for each candidate companion , assuming that the candidate is a background object . on - sky
plots presenting background ephemerides and the actual on - sky motion of each candidate companion relative to the primary are presented in figures [ fig : twhya_skyplots ] to [ fig : abdor_skyplots ] .
we compute the @xmath31 value for the expected background track position relative to the actual sky position for each candidate ( see nielsen et al . 2013 ) .
@xmath31 values are shown in table [ table_comp2a ] .
candidates with reduced @xmath31 values close to 1 are confirmed to be background objects . in total ,
81 candidate companions were tested for common proper motion with either archival or 2nd epoch gemini nici data .
of these candidates , 77 were background objects ; however , four co - moving brown dwarf or stellar companions ( discussed in more detail in section 7 ) were detected for the first time in the moving group sample : pz tel b @xcite , cd -35 2722b @xcite , hd 12894b ( this work ) and bd+07 1919c ( this work ) .
we also retrieve the known stellar companion to hd 82688 @xcite , as well as the brown dwarf companions ab pic b @xcite and hr 7329b @xcite . a number of stars ( hd 139084 b , v343 nor , cd-54 7336 , cd-31 16041 , hd 159911 , gj 560 a , and tyc 7443 - 1102 - 1 ) were near the galactic bulge and often possessed extremely dense starfields ( @xmath420 objects in the nici images ) .
as we expect almost all of these candidates to be background objects , we assigned these stars lower priority for second epoch nici followup and consequently they were not observed before the end of the nici campaign .
astrometry for candidates observed at only one epoch and thus unconfirmed as background or common proper motion is presented in table [ tab:1epoch ] .
in the course of the survey , we discovered two new low mass stellar companions , hd 12894b and bd+07 1919c ( figure [ fig : binaryimages ] ) .
nici and archival datasets analyzed are tabulated in table [ tab : binarydata ] .
both companions have been confirmed to be common proper motion with their primary using vlt naco archival data .
sky plots are shown in figure [ fig : binaryskyplots ] and astrometry is presented in tables [ tab : hd12894 ] and [ tab : bd+07 ] .
archival images were sky - subtracted and flat - fielded .
bad pixels identified from a dark image were removed .
images at different dither positions were registered and stacked .
astrometry was derived from both nici and naco archival datasets using the star and companion centroids measured from the final reduced stacked images . since the nici datasets for these binaries either had the starspot saturated or were taken in the narrow methane filters , we calculated broadband photometry from the vlt naco archival images .
both companions sit on the wings of the primary psf . for these datasets ,
the psf shape was generally azimuthally symmetric . to obtain photometry , we thus subtracted out a psf radial profile generated from the azimuthal median of the star image , excluding the position angle range within @xmath520 degrees of the detected companion .
aperture photometry was performed using 2 , 3 , 4 , 5 , and 6-pixel apertures .
all apertures produced consistent results ; we adopt the results using the 4-pixel aperture here . to estimate photometric errors , photometry was calculated both for individual reduced frames and the final reduced image .
we adopt the rms of the values from the individual reduced frames as the photometric error .
our photometry is presented in tables [ tab : hd12894 ] and [ tab : bd+07 ] .
we estimate companion masses based on the models of @xcite .
we adopt monte carlo methods to account for the photometric uncertainties as well as the range of possible distances and ages for these binaries .
we simulate an ensemble of 10@xmath32 realizations of the system , drawing from gaussian distributions in age , parallax , and photometry with 1@xmath29 widths taken from the measured uncertainties on these parameters . for each realization , we then interpolate with age and single - band absolute magnitude to estimate the mass of the companion from the models of @xcite .
the adopted mass is then the peak of the output distribution of simulated realizations , with error bars drawn from the 68% confidence limits of the output distribution .
results using _ j _ , _ h _ , and _
k_@xmath33 single band absolute magnitudes yielded consistent results ; _ k_@xmath33 band results are presented in tables [ tab : hd12894 ] and [ tab : bd+07 ] .
no estimate was made for the _ l _ band observations of hd 12894 , as we could not find an apparent _ l _ magnitude for hd 12894 in the literature .
we find best mass estimates of 0.46@xmath5 0.08 m@xmath7 for hd 12894b and [email protected] m@xmath7 for bd + 07 1919c .
these relatively massive ( 0.20.5 m@xmath7 ) companions have , unsurprisingly , shown some orbital motion between the archival and nici epochs .
thus , these orbits may yield dynamical mass measurements on a 10 - 20 year timescale . to determine the necessary timescales to measure these orbits
, we estimate their semimajor axes and periods . assuming a uniform eccentricity distribution between 0 @xmath8 e @xmath8 1 and random viewing angles , @xcite compute a median correction factor between projected separation and semimajor axis of 1.10@xmath34 ( 68.3% confidence limits ) . using this correction factor , we derive a semimajor axis of 16.9@xmath35 au for hd 12894ab and a semimajor axis of 13.8@xmath36 au for bd + 07 1919bc ( neglecting the presence or influence of a , which lies several arcsec and @xmath4200 au away ) . to convert from semi - major axis to period
requires an estimate of the total system mass .
we estimate the primary masses using the same monte carlo method as described above for the secondary masses , giving a mass of [email protected] m@xmath7 for hd 12894 and [email protected] m@xmath7 and [email protected] m@xmath7 for bd+07 1919b and c respectively . combining with the previously estimated companion masses , we estimate periods of 56@xmath37 yr for hd 12894ab and 55@xmath38 yr for bd+07 1919bc .
further orbital monitoring will thus be necessary to better constrain the semi - major axes and periods of these orbits . in @xcite , we reported the detection of a 36@xmath56 m@xmath6 companion to the young solar analogue pz tel , a member of the @xmath0 pic moving group . due to the considerable on - sky motion of pz tel b , we were able to constrain the eccentricity of the pz tel b orbit to @xmath40.6 through monte carlo orbital simulations with just two epochs of nici astrometry .
recently , this result has been confirmed by @xcite .
pz tel had previously been reported to have 70 @xmath12 m excess emission and hence a debris disk @xcite .
the existence of a debris disk is hard to reconcile with the highly eccentric orbit of the brown dwarf companion , which would likely disrupt the outer debris disk as it moves through it .
however , recent analysis of both _ spitzer _ 24 and 70 @xmath12 m data as well as _
70 , 100 , and 160 @xmath12 m data yield no detection of excess in any band at the location of pz tel ab ( g. bryden , private communication ) . there is a very red source @xmath1025@xmath22 north of pz tel ab which is likely extragalactic .
the centroiding algorithm used by @xcite allows for the centroid position to move from the target position in order to account for pointing errors and as a result likely mis - identified the extragalactic source as pz tel .
( l. rebull , private communication ) .
thus , pz tel does not possess a debris disk .
@xcite reported the discovery of a faint companion to the tuc - hor association star ab pic , with an estimated mass of 13 - 14 m@xmath6 .
compared to other objects of its spectral type , ab pic b s published @xmath39-band absolute magnitude is anomalously faint for its spectral type of l0.5 @xcite .
the published colors of this object are also quite red for its spectral type . during the nici campaign , we acquired new @xmath39 and @xmath40 photometry for ab pic b , presented here in table [ tab : abpic ] . while we measure a similarly red @[email protected] mag ( vs. [email protected] mag from *
* ) , we find a considerably brighter @xmath39 magnitude of [email protected] mag for b ( vs. [email protected] mag from * ? ? ?
* ) . in figure
[ fig : abpic ] , we plot spectral type vs. @xmath39 magnitude for ab pic b and a number of comparison objects .
the @xcite photometry places ab pic b fainter than expected for its spectral type . assuming the measured difference in photometry is not due to true variability , our brighter @xmath39-band magnitude places ab pic b firmly into the expected position for its spectral type .
here we present limits on the frequency of wide giant extrasolar planets based on two different statistical analyses of our achieved sensitivities for the mg sample .
two of our sample stars have confirmed planetary or planet - brown dwarf boundary companions , specifically @xmath0 pic and ab pic .
the bona fide planet around @xmath0 pic was not detected in our first epoch campaign data while ab pic b was clearly detected . including two stars with known @xmath820 m@xmath6 companions
poses issues for determining an unbiased estimate of planet frequency from our survey .
specifically , it is unclear how much we bias our estimate of planet frequency towards higher values by including a priori known companions .
thus , for the purposes of this analysis , we exclude these two stars from the sample . in section [ bayesdet ]
, we consider the effect of adding these two stars and their companions . following the method of @xcite and @xcite
, we use monte carlo methods to constrain our sensitivity to planets around each target star and combine these results to place constraints on planet fraction across our entire moving group sample .
first , we simulate 10000 planets with a given semimajor axis and mass , as well as randomly selected orbital parameters and eccentricity drawn from the eccentricity distribution of radial velocity planets @xcite . the ensemble of simulated planets in mass , semimajor axis variables
are then converted to equivalent contrasts and projected separations using the cond models of @xcite and the simulated orbital parameters .
this simulation was repeated at masses of 0.5 - 16.9 m@xmath6 , in steps of 0.164 m@xmath6 , and at semi - major axes of 0 - 4200 au , with step size varying as a function of distance ( 0.286 au out to 20 au , 5.333 au from 20 - 100 au , 7.333 au from 100 - 210 au , 10 au from 210 - 500 au , 20 au from 500 - 1000 au , 40 au from 1000 - 2000 au , and 100 au from 2000 to 4200 au ) . the converted ensemble
is then compared with the attained asdi and adi contrast curves for the star to derive the percentage of simulated planets detected at the particular combination of semimajor axis and mass . in cases where a candidate companion was observed in only a single epoch and
thus not confirmed as background or common proper motion , we cut off the contrast curve at the separation of the unconfirmed candidate companion or utilized a shallower contrast curve from an earlier epoch where the unconfirmed candidate was not detected .
a number of stars near the galactic bulge have been dropped from this analysis due to numerous unconfirmed candidate companions , specifically : cd -54 7336 , cd -31 16041 , hd 159911 , v343 nor , and hd 139084b . in total ,
73 stars were used for this analysis . for the asdi contrast curve comparison ,
the fluxes of simulated planets are modified to simulate the effect of asdi self - subtraction using the spex prism library of ultracool dwarfs to partition flux between the on- and off - methane absorption images ( see nielsen et al . 2013 and nielsen & close 2010 ) .
this contrast curve comparison procedure is then repeated along a grid of semimajor axes and masses . after calculating the detection probability grid for each star in the sample
, we use these values to place constraints on the planet frequency over the entire sample as a function of semi - major axis and mass . for a given bin in \{semi - major axis , mass } , the number of planets we expect to detect is given by : @xmath42 where @xmath43 is the fraction of planets with semimajor axis and mass ( @xmath44 ) we could detect given the achieved contrast for star @xmath45 ( i.e. the quantity calculated in our monte carlo simulations ) and @xmath46 is the fraction of stars that have such a planet to detect , hereafter referred to as `` planet fraction '' . according to radial velocity studies
, higher mass stars may preferentially host giant planets compared to lower mass stars @xcite . to account for this variation , we introduce a mass correction to adjust the probability that a given star hosts a planet based on that star s mass : @xmath47 where @xmath48 is the relative probability of hosting giant planets as a function of mass , based on the linear fit of planet frequency as a function of mass for rv planets from @xcite .
the mass - corrected version of equation [ eqn : mcprob ] is then : @xmath49 we normalize this correction at 1 @xmath50 since our sample is composed primarily of fgk stars . to estimate the mass of each of our sample stars , we interpolated from the models of @xcite .
first , we converted v and v - k to m@xmath51 and t@xmath52 using the lookup table developed for pre - main - sequence stars in @xcite .
then we used the @xcite solar metallicity tracks for 0.1 - 7 m@xmath7 stars to find the stellar mass which best reproduces the observed m@xmath51 and t@xmath52 .
in the zero - detection case , we use poisson statistics to set an upper limit on the planet fraction for our entire ensemble . assuming that planet fraction at a given semi - major axis and mass is the same for all survey stars , we remove @xmath53 from the sum . the 95% confidence level upper limit on planet fraction
, @xmath54 is then : @xmath55 where 3 is the poisson expectation value to set a 95% confidence upper limit on planet fraction in the null result case .
many of our sample stars have binary companions , which may disrupt the formation of planets in that system . to account for the effect of binary companions , we have followed the approach detailed in nielsen et al .
( 2013 ) and define an `` exclusion zone '' around each of the binaries in our sample in which we do not expect planets to form and thus where we do not simulate planets .
binaries in our sample are listed in table [ tab : binaries ]
. we also account for nonuniform position angle coverage of our observations at large angular separations .
the nici detector is square , with the focal plane mask and target star placed offset from the center . as a result , while we image 360@xmath28 in position angle at small separations , at larger separations ( @xmath196.3@xmath22 ) our coverage declines as some position angles are off the edge of the detector . in our monte carlo simulations we account for this effect by generating a uniform random variable between 0 and 1 for each simulated planet .
if that random variable is greater than the fractional angular coverage at the projected separation of the simulated planet , then that planet is considered undetectable even if it is brighter than the contrast curve .
this parameter is similar to the position angle of nodes ( rotation of the orbit on the plane of the sky ) , which follows a uniform distribution . when multiple contrast curves are available for a single target star , this random variable is also preserved across all epochs so that the same set of simulated planets are compared to each contrast curve for the same star .
figure [ fig : mg_ct ] gives the upper limit on planet fraction @xmath54 as a function of semimajor axis and planet mass for our entire moving group sample , using the models of @xcite to convert between achieved survey contrast and predicted detectable planet mass .
upper limits on planet fraction as a function of semi - major axis for this analysis ( i.e. single mass cuts from figure [ fig : mg_ct ] ) are presented in table [ table_pf ] .
giant planets are rare at wide separations ; for instance we expect less than 10@xmath9 of stars to possess a 2 m@xmath6 planet at semi - major axes of 49 to 290 au .
note that this analysis does not assume a particular distribution of planets as a function of mass and semi - major axis .
bayesian methods provide a powerful complement to frequentist monte - carlo methods for interpreting large - scale direct imaging surveys for exoplanets ( see e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
frequentist monte - carlo methods produce useful star - by - star constraints but sometimes have difficulties interpreting positive detections .
in contrast , bayesian methods produce less useful star - by - star constraints but can seamlessly handle both null and positive detections , as well as data analysis using multiple parameter models . here
we apply the bayesian statistical analysis method pioneered by @xcite to the nici mg sample .
our goal is to estimate the frequency of planets based on the observational constraints produced by our survey , given the limits of our sample size and sensitivity .
bayes theorem can be simply derived from the basic rules of probability and provides a powerful means to analyze and interpret data ( e.g. * ? ? ?
at the end of our experiment , the quantity we would like to determine is : @xmath56 which is the probability that a given model is correct given the data in hand as well as any other prior information @xmath57 .
this quantity is the posterior probability distribution function ( henceforth posterior pdf ) .
the power of bayes theorem is it allows us to relate the posterior pdf to other , more easily calculated quantities : @xmath58 the quantity @xmath59 is known as the likelihood function
it is the probability of obtaining the data on hand given a specific model and additional prior information .
the quantity @xmath60 is known as the prior probability ( or simply the prior ) and includes any additional prior information we know about the problem .
thus , by formulating reasonable likelihood functions and priors for a direct imaging planet detection survey , we can derive the posterior pdf and constrain models for the underlying planet population .
we adapt the method established in @xcite and @xcite for studying stellar binarity in the context of a direct imaging survey of exoplanets .
@xcite model the distributions of substellar and stellar binary mass ratios and semi - major axis as a power law in mass ratio and a gaussian in semi - major axis ( henceforth @xmath61 ) .
for exoplanet companions to stars , we adopt instead the form of the power law distributions derived for rv planets by @xcite , and consider only the planet mass ( henceforth @xmath62 ) rather than the mass ratio adopted for binaries : @xmath63 @xmath64 following the procedure of @xcite and @xcite , we extend the semi - major axis power - law out to a limiting cutoff value , since earlier studies already rule out a significant population of giant planets at very wide separations@xcite . while such planets do exist ( e.g. * ? ? ?
* ; * ? ? ?
* ) , they are much less common than planets detected via radial velocity at closer separations @xcite . thus , we are left with 4 parameters to our models : the two power - law indices @xmath65 and @xmath0 , the outer cutoff of the semi - major axis distribution ( henceforth @xmath66 ) , and @xmath67 , the fraction of stars with planets .
we define @xmath67 such that the planet fraction over a given range of semimajor axes and masses is : @xmath68 where c@xmath69 is a normalization constant ( and thus a function of f , @xmath65 , @xmath0 , and a@xmath70 ) .
the probability to find a planet around a star in a given \{semi - major axis , mass } bin , for a particular set of values for @xmath65 , @xmath0 , @xmath66 , and @xmath67 , is then the planet fraction within that bin : @xmath71 @xmath72 where c@xmath69 can be determined from equation [ eqn : f ] . to compute the likelihood ,
we calculate how many planets we expect to detect with this model in each \{semimajor axis , mass } bin and compare with the actual number of planets ( generally 0 ) detected in each bin , accounting for projection effects between semi - major axis and projected separation . for
a given \{semimajor axis , mass } bin and set of model parameters , the number of planets predicted will be : @xmath73 where n@xmath74 is the number of times this \{semimajor axis , mass } bin was observed in our survey ( derived from the contrast curves and stellar properties of each survey star ) .
for instance , if we observe 50 stars in our survey and 30 of the observed stars have contrasts deep enough to image a 10 m@xmath6 planet at a semimajor axis of 10 au , then @xmath75(10 au , 10 @xmath76 = 30 .
we then wish to compare @xmath77 with @xmath78 , the number of planets detected for a given \{semimajor axis , mass } bin .
to compare data and model , we need to adopt a likelihood estimator .
since we expect to detect only small numbers of planets , our survey can be treated as a counting experiment .
thus , we adopt poisson statistics to calculate the likelihood : @xmath79 to derive the posterior pdf for this bin , we must multiply the likelihood by any prior probability distribution for our parameters . for now , we adopt the simple uniform priors for @xmath65 , @xmath0 , f , and a@xmath70 : @xmath80 multiplying the likelihood and prior then yields the posterior pdf for this \{semimajor axis , mass } bin . in the last section
, we showed how to calculate the posterior pdf for one \{semimajor axis , mass } bin .
this can be generalized across all \{semimajor axis , mass } bins for the survey fairly easily .
we generalize @xmath75 and @xmath78 into 2d arrays for each \{projected separation , mass } bin observed , which we will henceforth call the window function and detection array , respectively .
to build the window function , we use the contrast curve for each survey star to define the ranges in separation and mass where planets can be detected and the ranges where the contrast is insufficient to do so .
often stars were observed in both adi and asdi modes ; in these cases , we adopt the best contrast value from the available curves at each given separation .
when asdi contrast curves are used , they are corrected for spectral self - subtraction , assuming no methane absorption ( i.e. the most conservative contrast case ) . in cases where a candidate companion was observed in only a single epoch and
thus not confirmed as background or common proper motion , we cut off the contrast curve at the separation of the unconfirmed candidate companion or utilized a shallower contrast curve from an earlier epoch where the unconfirmed candidate was not detected .
a number of stars near the galactic bulge have been dropped from this analysis due to numerous unconfirmed candidate companions , specifically : cd -54 7336 , cd -31 16041 , hd 159911 , v343 nor , and hd 139084b . in total ,
73 stars were used for this analysis .
bins where a planet can be detected are assigned a value of 1 and bins where no planet can be detected are assigned a value of 0 .
we account for nonuniform position angle coverage of our contrast curves by multiplying the window function for each star by the fractional coverage at each separation .
we then convert the window function expressed in projected angular separation and contrast to projected physical separation and estimated mass using the known distance and age of each star and either the dusty models of @xcite or the cond models of @xcite .
the detection array is set up in a similar manner as a simple array with the number of objects detected in each \{separation , mass } bin .
as exoplanets cool with age , dust should condense from their atmospheres , producing a transition from red , dusty spectra ( dusty ) to bluer spectra characterized by methane absorption ( cond ) .
however , no directly imaged planet to date has yet to show methane absorption in the near - ir , so we choose here to present results using both of these models .
we then calculate the posterior pdf for each \{separation , mass } point .
this calculation is accomplished using a small scale monte - carlo simulation . at each physical separation point , we simulate 10@xmath81 planetary orbits , drawing eccentricity , orbital phase , and other orbital elements randomly .
we solve for the semi - major implied for each simulated orbit , then produce a histogram of the result with a 5 au binsize .
the posterior pdf is calculated for each semi - major axis bin in this histogram and then weighted according to the number of simulated orbits falling into that bin to produce the posterior pdf at each given \{separation , mass } point .
we calculate the posterior in this manner at each \{separation , mass } point and then multiply the posterior pdfs across all these points to get the full posterior pdf across observable space for this set of model parameters .
this process is repeated for all sets of model parameters of interest to derive the full posterior pdf as a function of the four model parameters . to determine
what section of parameter space can be ruled out by our exoplanet non - detection around mg stars , we ran the bayesian analysis with all four parameters allowed to vary .
since our contrast curves are only 95% complete , we systematically estimate a slightly low planet fraction , but this effect is likely minor .
additionally , we have adopted hot - start models here , which predict considerably brighter planets at these young ages compared to cold - start models ( for instance * ? ? ?
thus , we predict systematically more stringent upper limits on planet fractions than would be found with cold start models . for window functions and the detection function , we considered a linear grid in separation ( in au ) running from 10.5 to 1015.5 au , with points every 5 au and a linear grid in mass ( in jupiter masses ) running from 0.2 to 19.2 @xmath82 , with points every 1 @xmath82 , thus fully covering the mass range of possible planets as well as low mass brown dwarfs which could plausibly form via core accretion @xcite .
the grids for @xmath65 and @xmath0 were centered on the values @xmath65=-1.16 and @xmath0=-0.61 , derived from radial velocity planet distributions @xcite and converted from the logarithmic units used in @xcite to linear units here .
we allowed @xmath65 to run from -2.09 to -0.16 in increments of 0.066 , @xmath0 to run from -1.54 to 0.39 , in increments of 0.066 , @xmath67 to run from 0.005 to 0.972 in increments of 0.033 , and semi - major axis cutoff @xmath66 to run from 12.5 au to 152.5 au in increments of 5 au .
we choose to investigate this range of semi - major axis cutoff values as a value of @xmath83 10 au is ruled out by radial velocity studies @xcite and a value of @xmath84 150 au is ruled out by previous directly imaging studies @xcite .
planet fraction @xmath67 , and hence also the normalization constant @xmath85 , are calculated over the range 10 150 au .
obviously , the complete 4-dimensional posterior pdf can not be fully visualized , so to present the results , we have calculated 1-d and 2-d marginalized posterior pdfs by integrating over some of the parameters .
1-d marginalized posterior pdfs are presented in figure [ fig:1d_4param_dusty ] for both dusty @xcite and cond models @xcite .
the 2-d marginalized posterior pdfs are presented in figure [ fig:2d_4param_dusty ] for the dusty models and in figure [ fig:2d_4param_cond ] for the cond models .
all pdfs are plotted in logarithmic units .
non - detection of planets with such a large sample and deep contrasts places the strongest constraints to date on the planet fraction @xmath67 for directly imaged exoplanets .
we derive upper limits on planet fraction @xmath67 by normalizing our 1-d marginalized posterior pdf for @xmath67 .
upper limits on planet fraction @xmath67 are tabulated in table [ table_pf_bayesian ] . for the dusty models ,
a semi - major axis range of 10 - 150 au , and companion masses of 1 - 20 m@xmath6 , our 95.4@xmath9 confidence limit on f is @xmath3018@xmath9 , and at a 99.7@xmath9 confidence level , @xmath86 .
for the same parameter ranges and the cond models , at a 95.4@xmath9 confidence level , @xmath876@xmath9 , and at a 99.7@xmath9 confidence level , @xmath8714@xmath9 .
this is consistent with the results from our monte carlo simulations as well ( see table [ table_pf ] ) and is valid for a wide range of possible planet distributions .
our results strongly constrain the frequency of planets within semi - major axes of 50 au as well . for the dusty models , a semi - major axis range of 10 - 50 au , and companion masses of 1 - 20 m@xmath6 , at a 95.4@xmath9 confidence level , @xmath8721@xmath9 , and at a 99.7@xmath9 confidence level , @xmath88 . for
the same parameter ranges and the cond models , at a 95.4@xmath9 confidence level , @xmath877@xmath9 , and at a 99.7@xmath9 confidence level , @xmath8717@xmath9 .
the similar constraints obtained for 10 - 50 au as for 10 - 150 au suggests that the 50 - 150 au semi - major axis range is quite devoid of planets .
other than for @xmath67 , however , our marginalized posterior pdfs remain unconstrained
( i.e. no clear peak or trailing off to 0 ) and do not cover a wide range in ln(pdf ) . while the marginalized 1-d posterior pdf for planet fraction @xmath67 varies by over 10 orders of magnitude ( figure [ fig:1d_4param_dusty ] ) , the marginalized 1-d posteriors for the other 3 parameters vary by @xmath81.5 orders of magnitude ( a factor of 4.5 at most ) .
thus , we do not place confidence intervals on parameters other than the planet frequency @xmath67 . while we can place strong limits on @xmath67 by marginalizing over the other three parameters , determining the best - fit power law parameters for directly imaged planet populations
must be deferred until there is a statistically significant population of such objects to fit .
the choice of a power - law model for directly imaged planet distributions is based on fits to the properties of radial velocity planet @xcite ; it is not known yet whether this is the best model to describe directly imaged planet distributions .
our bayesian approach can seamlessly handle both planet detections and non - detections . here
we rerun the bayesian analysis described above , this time adding in ab pic and @xmath0 pic , the two stars with already known planetary or low mass brown dwarf ( @xmath820 m@xmath6 ) companions in our sample .
we adopt a mass estimate of 8 m@xmath6 ( i.e. the middle of the range found by * ? ? ?
* ) and a projected separation of 8.5 au @xcite for @xmath0 pic b. for ab pic b , we adopt a mass estimate of 13.5 m@xmath6 @xcite and a projected separation of 275 au @xcite . the bayesian analysis described above was rerun with the 73 original stars , and including 1 ) only the @xmath0 pic b detection , 2 ) only the ab pic b detection , and 3 ) both detections .
results are presented in fig .
[ fig:1d_det ] . including the companions
affects the shape of the marginalized pdf for planet fraction @xmath67 and also provide a significant constraint on @xmath66 , the semi - major axis cutoff . the marginalized pdfs for planet fraction
@xmath67 with @xmath0 pic b only and for both detections now show a peak at @xmath104@xmath9 , as a clear detection of a close companion rules out a zero value for planet fraction @xmath67 in the 10 - 150 au range . for the @xmath0 pic b
only case , planet fraction @xmath67=0.04@xmath89 , with 95.4@xmath9 confidence level error bars , so , as expected , detection of a single object does not highly constrain @xmath67 .
the ab pic b detection is at much larger separation than the @xmath0 pic b detection , so it provides much less of a constraint on planet fraction in this range , as there is only a very slight chance that this companion has a semi - major axis @xmath8 150 au ( i.e. a highly eccentric orbit ) .
the marginalized pdf for planet fraction @xmath67 in this case shows some flattening at small values of @xmath67 but no clear peak . as in the non - detection case ,
@xmath65 and @xmath0 do not cover enough range in ln(pdf ) to yield useful constraints .
given that the two companions were detected in very different separation regimes , they provide contradictory constraints on cutoff , with the ab pic b detection strongly ruling out @xmath66 @xmath8 100 au and the @xmath0 pic b weakly ruling @xmath66 @xmath4 100 au . while it is informative to investigate how single detections with varying estimated masses and separations affect the shape of the posterior pdf , it is dangerous to draw conclusions based on such a small sample of detections . detection of a larger cohort of similar companions is necessary to put consistent constraints on the properties of such objects
while a number of giant planets and giant planet candidates have now been imaged at separations @xmath420 au , large - scale surveys illustrate that such planets are comparatively rare around main - sequence solar analogues and low mass stars . of the ensemble of directly - imaged planets known to date , most have been discovered around a stars ( hr 8799bcde , fomalhaut b , @xmath0 pic b , wd 0806 - 661 ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) , with a few also discovered around very young solar analogues ( 1rxs j160929.1@xmath90210524b , lkca 15b , gsc 06214@xmath9000210b , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
only one planet has been directly - imaged to date around a main - sequence solar analogue ( gj 504b ; * ? ? ?
two companions right at the deuterium burning limit have recently been reported around m stars @xcite , but no companion with estimated mass @xmath810 m@xmath6 has yet been imaged around a low mass star . the small number of detected planets is not due to a lack of stars surveyed .
our nici survey of 80 stars is the largest single sample of mg stars observed .
significant numbers of mg stars have also been observed as part of the gemini deep planet survey @xcite , international deep planet survey @xcite , sdi survey @xcite , deep imaging survey of young , nearby austral stars @xcite , a survey of young , nearby , and dusty stars @xcite , naco large program ( vigan et al .
2013 ) , and seeds ( brandt et al . 2013 ) . based on a sample of 118 stars compiled from the surveys of @xcite , @xcite , and @xcite
, @xcite found that planets more massive than 4 m@xmath6 are found around @xmath820@xmath9 of fgkm stars in orbits between 22 and 507 au , at 95@xmath9 confidence .
@xcite find a qualitatively similar result based on a sample of 88 stars ( 51 of which are members of young moving groups ) , constraining the fraction of stars with giant planets to @xmath810@xmath9 at semi - major axes @xmath440 au for a planet distribution extended from radial velocity power laws . with considerably higher contrasts and better inner working angles ( 0.3 vs. typically 0.5 - 0.7 ) ,
our work here directly extends these results to lower masses and smaller separations .
our results are qualitatively similar to those of @xcite but at a considerably higher confidence level . as discussed in section [ bayesnodet ] , we confirm and extend the result of @xcite : @xmath45 m@xmath6 companions to fgkm stars are rare at separations @xmath410 au .
@xcite and @xcite find that for rv planets , host star mass and planet mass are related .
higher mass stars seem to preferentially host more high mass rv planets ( @xmath41 m@xmath6 ) than lower mass stars , attributable to the fact that more massive stars also likely possessed more massive primordial circumstellar disks .
indeed , @xcite find that @xmath41 m@xmath6 radial velocity planets are quite rare around m stars at semi - major axes @xmath85 au .
the fact that the majority of directly - imaged planets to date have been found around higher mass stars qualitatively suggests a similar conclusion may hold for the wide planet population probed by direct imaging .
we examine here whether the statistics from direct imaging surveys to date supports this assertion . the first constraints on directly imaged planet fraction for high mass ab stars
have only recently been published .
@xcite found that @xmath830@xmath9 of massive stars have giant planet ( @xmath41 m@xmath6 ) or brown dwarf companions that formed via gravitational instability with mass @xmath8100 m@xmath6 within 300 au at the 99@xmath9 confidence level for a sample of 18 high - mass stars in the solar neighborhood , however this work does not place limits on core - accretion planets around these hosts .
for a 42-star sample , @xcite found that the fraction of a stars with 1 massive planet ( 3 - 14 m@xmath6 ) from 5 - 300 au was 5.9 - 18.8@xmath9 at the 68@xmath9 confidence level ( assuming power law distributions for mass and semi - major axis appropriate for core - accretion planets ) , however the age determination for their survey stars may be overly optimistic ( nielsen et al .
2013 ) . our current sample
is comprised of 70@xmath9 stars with spectral type of k or later and contains 33 m stars , and thus can be directly compared to the samples of @xcite to test whether planet fraction indeed falls with stellar mass . using the dusty models , we limit the planet fraction @xmath67 of our sample to @xmath303.5@xmath9 at the 68@xmath9 confidence level result for 120 m@xmath6 companions at semi - major axes of 10150 au ; this is lower than the 5.9 - 18.8@xmath9 planet fraction at a 68@xmath9 confidence level found by @xcite .
thus , the current set of direct imaging surveys may hint that directly imaged giant planets are less common around lower mass gkm stars compared to ab stars .
as part of the gemini nici planet - finding campaign , we imaged 80 members of nearby young moving groups , with ages from 10200 myr and within 100 pc . in asdi mode
, we attain median contrasts of @xmath1(mag)=12.4 , 13.9 , and 14.5 mag at 0.5@xmath22 , 1@xmath22 , and 2@xmath22 respectively in the narrow band methane filters ( @xmath26=1.58 @xmath12 m ) , with a typical standard deviation of 0.9 mag . in adi mode ,
we attain median contrasts of @xmath1(mag)=10.4 , 13.2 , and 15.1 mag at 0.5@xmath22 , 1@xmath22 , and 2@xmath22 respectively in @xmath2 band .
we achieve median minimum detectable masses of 11 , 5 , and 3 m@xmath6 at 0.5@xmath22 , 1@xmath22 , and 2@xmath22 using the dusty models @xcite .
candidate companions within 400 au from the star and not in dense stellar fields that could not be confirmed or rejected as common - proper motion companions using archival data were reobserved with nici .
a total of 77 candidate companions were detected and eliminated as background contaminants .
four comoving brown dwarf or substellar companions were discovered in the moving group sample : pz tel b @xcite , cd -35 2722b @xcite , hd 12894b ( this work ) and bd+07 1919c ( this work ) .
pz tel b and cd-35 2722b are both 30 - 40 m@xmath6 brown dwarf companions , while hd 12894b and bd+07 1919c are stellar companions with estimated masses of [email protected] m@xmath7 and [email protected] m@xmath7 respectively .
we also retrieved the substellar companions ab pic b @xcite and hr 7329 b @xcite as well as the known stellar companion to hd 82688 @xcite . to compare to previous published surveys ,
we have adopted hot start models in our statistical analysis , which predict considerably brighter planets at these young ages compared to cold start models ( for instance , * ? ? ?
thus , earlier surveys as well as our own predict systematically more stringent upper limits on planet fraction than would be found with cold start models .
nonetheless , our constraints on planet fraction are consistent with and more stringent than previous work . from a bayesian analysis for a wide range of parameters and power - law models of planet distributions ,
we restrict the frequency of 120 m@xmath6 companions at semi - major axes from 10150 au to @xmath818% at a 95.4@xmath9 confidence level using dusty models @xcite and to @xmath86% at a 95.4@xmath9 confidence level using cond models @xcite .
we thank geoff bryden and luisa rebull for clarification on the spitzer detection of pz tel .
we thank the referee for useful suggestions which helped strengthen this work .
b.a.b . was supported by hubble fellowship grant hst - hf-01204.01-a awarded by the space telescope science institute , which is operated by aura for nasa , under contract nas 5 - 26555 .
this work was supported in part by nsf grants ast-0713881 and ast- 0709484 awarded to m. liu , nasa origins grant
nnx11 ac31 g awarded to m. liu , and nsf grant aag-1109114 awarded to l. close .
the gemini observatory is operated by the association of universities for research in astronomy , inc .
, under a cooperative agreement with the nsf on behalf of the gemini partnership : the national science foundation ( united states ) , the science and technology facilities council ( united kingdom ) , the national research council ( canada ) , conicyt ( chile ) , the australian research council ( australia ) , cnpq ( brazil ) , and conicet ( argentina ) .
based on observations made with the european southern observatory telescopes obtained from the eso / st - ecf science archive facility .
this publication makes use of data products from the two micron all sky survey , which is a joint project of the university of massachusetts and the infrared processing and analysis center / california institute of technology , funded by the national aeronautics and space administration and the national science foundation .
this research has made use of the simbad database , operated at cds , strasbourg , france .
this research has made use of the vizier catalogue access tool , cds , strasbourg , france .
the digitized sky survey was produced at the space telescope science institute under u.s .
government grant nag w-2166 .
the images of these surveys are based on photographic data obtained using the oschin schmidt telescope on palomar mountain and the uk schmidt telescope .
the plates were processed into the present compressed digital form with the permission of these institutions .
allen , p. r. 2007 , , 668 , 492 allers , k. n. , & liu , m. c. 2013 , , 772 , 79 apai , d. , janson , m. , moro - martn , a. , et al .
2008 , , 672 , 1196 barrado y navascus , d. , stauffer , j. r. , song , i. , & caillault , j .-
p . 1999 , , 520 , l123 barrado y navascus ,
d. 2006 , , 459 , 511 baraffe , i. , chabrier , g. , allard , f. , & hauschildt , p. h. 1998 , , 337 , 403 baraffe , i. , chabrier , g. , allard , f. , & hauschildt , p. h. 2002 , , 382 , 563 baraffe , i. , chabrier , g. , barman , t. s. , allard , f. , & hauschildt , p. h. 2003 , , 402 , 701 barman , t. s. , macintosh , b. , konopacky , q. m. , & marois , c. 2011 , , 733 , 65 barman , t. s. , macintosh , b. , konopacky , q. m. , & marois , c. 2011 , , 735 , l39 biller , b. a. , close , l. m. , masciadri , e. , et al .
2007 , , 173 , 143 biller , b. , artigau , .
, wahhaj , z. , et al .
2008 , , 7015 , biller , b. a. , liu , m. c. , wahhaj , z. , et al .
2010 , , 720 , l82 bonavita , m. , chauvin , g. , desidera , s. , et al .
2012 , , 537 , a67 bonnefoy , m. , chauvin , g. , rojo , p. , et al .
2010 , , 512 , a52 bonnefoy , m. , boccaletti , a. , lagrange , a .- m .
, et al . 2013 , arxiv:1302.1160 bowler , b. p. , liu , m. c. , dupuy , t. j. , & cushing , m. c. 2010 , , 723 , 850 bowler , b. p. , liu , m. c. , shkolnik , e. l. , & dupuy , t. j. 2013 , arxiv:1307.2237 buenzli , e. , thalmann , c. , vigan , a. , et al . 2010 , , 524 , l1 burrows , a. , sudarsky , d. , & lunine , j. i. 2003 , , 596 , 587 carson , j. , thalmann , c. , janson , m. , et al .
2012 , arxiv:1211.3744 chabrier , g. , baraffe , i. , allard , f. , & hauschildt , p. 2000 , , 542 ,
464 chauvin , g. , lagrange , a .- m . , dumas , c. , et al .
2005 , , 438 , l25 chauvin , g. , lagrange , a .- m . , zuckerman , b. , et al . 2005 , , 438 , l29 chauvin , g. , lagrange , a .- m . , bonavita , m. , et al .
2010 , , 509 , a52 chauvin , g. , lagrange , a .- m . , beust , h. , et al .
2012 , , 542 , a41 close , l. m. , lenzen , r. , guirado , j. c. , et al .
2005 , , 433 , 286 close , l. m. , thatte , n. , nielsen , e. l. , et al .
2007 , , 665 , 736 close , l. m. , & males , j. r. 2010 , , 709 , 342 chun , m. , toomey , d. , wahhaj , z. , et al .
2008 , , 7015 , currie , t. , thalmann , c. , matsumura , s. , et al .
2011 , , 736 , l33 cumming , a. , butler , r. p. , marcy , g. w. , et al .
2008 , , 120 , 531 cushing , m. c. , rayner , j. t. , & vacca , w. d. 2005 , , 623 , 1115 da silva , l. , torres , c. a. o. , de la reza , r. , et al .
2009 , , 508 , 833 de la reza , r. , torres , c. a. o. , quast , g. , castilho , b. v. , & vieira , g. l. 1989 , , 343 , l61 delorme , p. , gagn , j. , girard , j. h. , et al .
2013 , , 553 , l5 dodson - robinson , s. e. , veras , d. , ford , e. b. , & beichman , c. a. 2009 , , 707 , 79 dommanget , j. , & nys , o. 2002 , vizier online data catalog , 1274 , 0 dupuy , t. j. , liu , m. c. , bowler , b. p. , et al .
2010 , , 721 , 1725 dupuy , t. j. & liu , m. c. 2011 , , 733 , 122 dupuy , t. j. & liu , m. c. 2012 , , 201 , 19 fernndez , d. , figueras , f. , & torra , j. 2008 , , 480 , 735 fischer , d. a. , & valenti , j. 2005 , , 622 , 1102 fuhrmann , k. 2004 , astronomische nachrichten , 325 , 3 gaidos , e. j. 1998 , , 110 , 1259 geballe , t. r. , knapp , g. r. , leggett , s. k. , et al .
2002 , , 564 , 466 gizis , j. e. 2002 , , 575 , 484 gray , r. o. , corbally , c. j. , garrison , r. f. , et al .
2006 , , 132 , 161 gregorio - hetem , j. , lepine , j. r. d. , quast , g. r. , torres , c. a. o. , & de la reza , r. 1992 , , 103 , 549 guenther , e. w. , neuhuser , r. , hulamo , n. , brandner , w. , & alves , j. 2001 , , 365 , 514 guirado , j. c. , reynolds , j. e. , lestrade , j .- f . , et al .
1997 , , 490 , 835 heinze , a. n. , hinz , p. m. , sivanandam , s. , et al .
2010 , , 714 , 1551 heinze , a. n. , hinz , p. m. , kenworthy , m. , et al .
2010 , , 714 , 1570 hg , e. , fabricius , c. , makarov , v. v. , et al . 2000 , , 355 , l27 ireland , m. j. , kraus , a. , martinache , f. , law , n. , & hillenbrand , l. a. 2011 , , 726 , 113 janson , m. , bonavita , m. , klahr , h. , et al .
2011 , , 736 , 89 janson , m. , bonavita , m. , klahr , h. , & lafrenire , d. 2012 , , 745 , 4 jayawardhana , r. , hartmann , l. , fazio , g. , et al .
1999 , , 521 , l129 johnson , j. a. , butler , r. p. , marcy , g. w. , et al .
2007 , , 670 , 833 johnson , j. a. , aller , k. m. , howard , a. w. , & crepp , j. r. 2010 , , 122 , 905 kalas , p. , graham , j. r. , chiang , e. , et al .
2008 , science , 322 , 1345 kasper , m. , apai , d. , janson , m. , & brandner , w. 2007 , , 472 , 321 kastner , j. h. , zuckerman , b. , weintraub , d. a. , & forveille , t. 1997 , science , 277 , 67 kastner , j. h. , zuckerman , b. , & bessell , m. 2008 , , 491 , 829 kenyon , s. j. , & hartmann , l. 1995 , , 101 , 117 kiss , l. l. , mor , a. , szalai , t. , et al .
2011 , , 411 , 117 koen , c. , kilkenny , d. , van wyk , f. , & marang , f. 2010 , , 403 , 1949 kraus , a. l. , ireland , m. j. , martinache , f. , & hillenbrand , l. a. 2011 , , 731 , 8 kraus , a. l. , & ireland , m. j. 2012 , , 745 , 5 kuzuhara , m. , tamura , m. , kudo , t. , et al .
2013 , accepted to , arxiv:1307.2886 lafrenire , d. , marois , c. , doyon , r. , nadeau , d. , & artigau , .
2007 , , 660 , 770 lafrenire , d. , doyon , r. , marois , c. , et al .
2007 , , 670 , 1367 lafrenire , d. , jayawardhana , r. , & van kerkwijk , m. h. 2008 , , 689 , l153 lagrange , a .- m . , gratadour , d. , chauvin , g. , et al .
2009 , , 493 , l21 lagrange , a .- m . , bonnefoy , m. , chauvin , g. , et al .
2010 , science , 329 , 57 landolt , a. u. 2009 , , 137 , 4186 lpine , s. , & simon , m. 2009 , , 137 , 3632 liu , m. c. 2004 , science , 305 , 1442 liu , m. c. , wahhaj , z. , biller , b. a. , et al .
2010 , , 7736 , looper , d. l. , burgasser , a. j. , kirkpatrick , j. d. , & swift , b. j. 2007 , , 669 , l97 looper , d. l. , bochanski , j. j. , burgasser , a. j. , et al .
2010 , , 140 , 1486 looper , d. l. , mohanty , s. , bochanski , j. j. , et al .
2010 , , 714 , 45 lpez - santiago , j. , montes , d. , crespo - chacn , i. , & fernndez - figueroa , m. j. 2006 , , 643 , 1160 lowrance , p. j. , schneider , g. , kirkpatrick , j. d. , et al .
2000 , , 541 , 390 lowrance , p. j. , becklin , e. e. , schneider , g. , et al .
2005 , , 130 , 1845 luhman , k. l. , stauffer , j. r. , & mamajek , e. e. 2005 , , 628 , l69 luhman , k. l. , burgasser , a. j. , & bochanski , j. j. 2011 , , 730 , l9 mamajek , e. e. 2005 , , 634 , 1385 malo , l. , doyon , r. , lafrenire , d. , et al .
2013 , , 762 , 88 marois , c. , doyon , r. , nadeau , d. , et al .
2005 , , 117 , 745 marois , c. , lafrenire , d. , doyon , r. , macintosh , b. , & nadeau , d. 2006 , , 641 , 556 marois , c. , macintosh , b. , barman , t. , et al .
2008 , science , 322 , 1348 marois , c. , zuckerman , b. , konopacky , q. m. , macintosh , b. , & barman , t. 2010 , , 468 , 1080 masciadri , e. , mundt , r. , henning , t. , alvarez , c. , & barrado y navascus , d. 2005 , , 625 , 1004 mason , b. d. , wycoff , g. l. , hartkopf , w. i. , douglass , g. g. , & worley , c. e. 2013 , vizier online data catalog , 1 , 2026 messina , s. , desidera , s. , turatto , m. , lanzafame , a. c. , & guinan , e. f. 2010 , , 520 , a15 metchev , s. a. , & hillenbrand , l. a. 2009 , , 181 , 62 montes , d. , lpez - santiago , j. , fernndez - figueroa , m. j. , & glvez , m. c. 2001 , , 379 , 976 mordasini , c. , alibert , y. , & benz , w. 2009 , , 501 , 1139 mor , a. , brahm , p. , derekas , a. , et al . 2006 , , 644 , 525 mugrauer , m. , rll , t. , ginski , c. , et al .
2012 , , 424 , 1714 nielsen , e. l. , close , l. m. , guirado , j. c. , et al .
2005 , astronomische nachrichten , 326 , 1033 nielsen , e. l. , close , l. m. , biller , b. a. , masciadri , e. , & lenzen , r. 2008 , , 674 , 466 nielsen , e. l. , & close , l. m. 2010 , , 717 , 878 nielsen , e. l. , liu , m. c. , wahhaj , z. , et al .
2012 , , 750 , 53 nielsen , e. l. , liu , m. c. , wahhaj , z. , et al .
2013 , accepted to , arxiv:1306.1233 quanz , s. p. , amara , a. , meyer , m. r. , et al .
2013 , , 766 , l1 racine , r. , walker , g. a. h. , nadeau , d. , doyon , r. , & marois , c. 1999 , , 111 , 587 rameau , j. , chauvin , g. , lagrange , a .- m . , et al .
2013 , , 553 , a60 rameau , j. , chauvin , g. , lagrange , a .- m . , et al .
2013 , arxiv:1305.7428 rebull , l. m. , stapelfeldt , k. r. , werner , m. w. , et al .
2008 , , 681 , 1484 riaz , b. , gizis , j. e. , & harvin , j. 2006 , , 132 , 866 rice , e. l. , faherty , j. k. , & cruz , k. l. 2010 , , 715 , l165 rodriguez , d. r. , bessell , m. s. , zuckerman , b. , & kastner , j. h. 2011 , , 727 , 62 rucinski , s. m. , & krautter , j. 1983 , , 121 , 217 schlieder , j. e. , lpine , s. , & simon , m. 2010 , , 140 , 119 schlieder , j. e. , lepine , s. , & simon , m. 2012 , arxiv:1208.0679 shkolnik , e. l. , liu , m. c. , reid , i. n. , dupuy , t. , & weinberger , a. j. 2011 , , 727 , 6 shkolnik , e. l. , anglada - escude , g. , liu , m. c. , et al .
2012 , arxiv:1207.5074 schneider , j. , dedieu , c. , le sidaner , p. , savalle , r. , & zolotukhin , i. 2011 , , 532 , a79 scholz , r .- d . , mccaughrean , m. j. , zinnecker , h. , & lodieu , n. 2005 , , 430 , l49 siess , l. , dufour , e. , & forestini , m. 2000 , , 358 , 593 sivia , d.s . & skilling , j. `` data analysis : a bayesian tutorial '' , 2006 , oxford science publications skemer , a. j. , close , l. m. , szcs , l. , et al . 2011 , , 732 , 107 skemer , a. j. , hinz , p. m. , esposito , s. , et al .
2012 , , 753 , 14 smith , b. a. , & terrile , r. j. 1984 , science , 226 , 1421 song , i. , bessell , m. s. , & zuckerman , b. 2002 , , 385 , 862 song , i. , zuckerman , b. , & bessell , m. s. 2003 , , 599 , 342 spiegel , d. s. , & burrows , a. 2012 , , 745 , 174 sterzik , m. f. , alcal , j. m. , covino , e. , & petr , m. g. 1999 , , 346 , l41 thalmann , c. , carson , j. , janson , m. , et al .
2009 , , 707 , l123 todorov , k. , luhman , k. l. , & mcleod , k. k. 2010 , , 714 , l84 torres , c. a. o. , da silva , l. , quast , g. r. , de la reza , r. , & jilinski , e. 2000 , , 120 , 1410 torres , c. a. o. , quast , g. r. , da silva , l. , et al .
2006 , , 460 , 695 torres , c. a. o. , quast , g. r. , melo , c. h. f. , & sterzik , m. f. 2008 , handbook of star forming regions , volume ii , 757 van leeuwen , f. 2007 , , 474 , 653 vigan , a. , patience , j. , marois , c. , et al .
2012 , , 544 , a9 vigan , a. , chauvin , g. , bonavita , m. et al .
2013 , proceedings of the iau 299 symposium wahhaj , z. , liu , m. c. , biller , b. a. , et al .
2011 , , 729 , 139 wahhaj , z. , liu , m. c. , nielsen , e. l. , et al .
2013 , , 773 , 179 wahhaj , z. , liu , m. c. , nielsen , e. l. , et al .
2013 , submitted to apj webb , r. a. , zuckerman , b. , platais , i. , et al . 1999 , , 512 , l63 weinberger , a. j. , anglada - escud , g. , & boss , a. p. 2013
, , 762 , 118 white , r. j. , gabor , j. m. , & hillenbrand , l. a. 2007 , , 133 , 2524 zuckerman , b. , webb , r. a. , schwartz , m. , & becklin , e. e. 2001 , , 549 , l233 zuckerman , b. , song , i. , & webb , r. a. 2001 , , 559 , 388 zuckerman , b. , song , i. , bessell , m. s. , & webb , r. a. 2001 , , 562 , l87 zuckerman , b. , song , i. , & bessell , m. s. 2004 , , 613 , l65 zuckerman , b. , & song , i. 2004 , , 42 , 685 zuckerman , b. , & webb , r. a. 2000 , , 535 , 959 zuckerman , b. , rhee , j. h. , song , i. , & bessell , m. s. 2011 , , 732 , 61 lcccccccc tw hya , 10 myr + twa 6 & 10:18:28.8 & -31:50:02 & ( 77 ) & m0 & 12.00 & 8.18 & 8.04 & + twa 7 & 10:42:30.1 & -33:40:16 & ( 28 ) & m2 & 11.06 & 7.12 & 6.90 & + tw hya & 11:01:52.0 & -34:42:17 & 53.7 & k6 & 11.34 & 7.56 & 7.30 & + twa 14 & 11:13:26.5 & -45:23:43.0 & 95.9 & m0 & & 8.73 & 8.49 & + twa 13a & 11:21:17.2 & -34:46:46 & 55.6 & m1 & 11.46 & 7.73 & 7.49 & + twa 13b & 11:21:17.4 & -34:46:50 & 59.7 & m1 & 11.96 & 7.68 & 7.46 & + twa 8a & 11:32:41.2 & -26:51:56 & 47.0 & m3 & 12.23 & 7.66 & 7.43 & + twa 9b & 11:48:23.7 & -37:28:48 & 46.8 & m1 & 14.00 & 9.38 & 9.15 & maybe not member + twa 9a & 11:48:24.2 & -37:28:49 & 46.8 & k5 & 11.32 & 8.03 & 7.85 &
maybe not member + twa 25 & 12:15:30.8 & -39:48:42.0 & ( 51 ) & m1 & 11.62 & 7.50 & 7.31 & + twa 20 & 12:31:38.1 & -45:58:59 & 77.5 & m3 & 13.00 & 8.69 & 8.41 & + twa 10 & 12:35:04.2 & -41:36:39 & ( 52 ) & m2 & 12.96 & 8.48 & 8.19 & + hr 4796 b & 12:36:00.56 & -39:52:16 & 72.8 & m2.5 & 13.3 & 8.53 & 8.35 & +
hr 4796 a & 12:36:01.2 & -39:52:10 & 72.8 & a0 & 5.78 & 5.79 & 5.77 & + + beta pic , 12 myr + hr 9 & 00:06:50.1 & -23:06:27 & 39.4 & f3 & 6.22 & 5.33 & 5.24 & + hip 10679 & 02:17:24.7 & + 28:44:31 & 27.3 & g2 & 7.83 & 6.36 & 6.26 & + hd 15115 & 02:26:16.2 & + 06:17:33 & 45.2 & f4 & 6.83 & 5.86 & 5.82 & + 51 eri & 04:37:36.1 & -02:28:25 & 29.4 & m1 & 5.24 & 4.77 & 4.54 & + gj 182 & 04:59:34.8 & + 01:47:00.7 & 25.9 & m0 & 10.26 & 6.45 & 6.26 & + hip 23309 & 05:00:47.1 & -57:15:25.5 & 26.8 & m0 & 10.25 & 6.43 & 6.24 & + bd-21 1074a & 05:06:49.9 & -21:35:09 & ( 18 ) & m1 & 10.60 & 6.39 & 6.12 & + hip 25486 & 05:27:04.8 & -11:54:04 & 27.0 & f7 & 6.36 & 5.09 & 4.93 & + @xmath0 pic & 05:47:17.1 & -51:03:59 & 19.4 & a3 & 3.87 & 3.54 & 3.53 & + ao men & 06:18:28.2 & -72:02:41.4 & 38.6 & k4 & 10.11 & 6.98 & 6.81 & + hd 139084 b & 15:38:56.9 & -57:42:18 & 38.5 & m5 & 14.80 & 9.45 & 9.19 & + v343 nor & 15:38:57.6 & -57:42:27 & 38.5 & k0 & 8.25 & 5.99 & 5.85 & + hd 155555 c & 17:17:31.3 & -66:57:05 & 31.4 & m3 & 12.82 & 7.92 & 7.63 & + cd-54 7336 & 17:29:55.1 & -54:15:49 & ( 66 ) & k1 & 9.62 & 7.46 & 7.36 & + hd 164249b & 18:03:04.1 & -51:38:56 & 48.0 & m2 & 12.50 & 8.52 & 8.27 & + hd 172555 a & 18:45:26.9 & -64:52:16.5 & 28.5 & a7 & 4.79 & 4.25 & 4.30 & + tyc 9073 - 0762 - 1 & 18:46:52.6 & -62:10:36 & ( 54 ) & m1 & 12.31 & 8.05 & 7.85 & + cd-31 16041 & 18:50:44.5 & -31:47:47 & ( 51 ) & k8 & 11.41 & 7.67 & 7.46 & + pz tel & 18:53:05.9 & -50:10:50 & 51.5 & g9 & 8.50 & 6.49 & 6.37 & + hr 7329 & 19:22:51.2 & -54:25:24 & 48.2 & a0 & 5.02 & 5.15 & 5.01 & + hip 95270 & 19:22:58.9 & -54:32:15 & 51.8 & f6 & 7.09 & 5.98 & 5.91 & + tyc 7443 - 1102 - 1b & 19:56:02.94 & -32:07:18.7 & ( 57.7 ) & m4 & 13.3 & 8.34 & 8.11 & + tyc 7443 - 1102 - 1a & 19:56:04.37 & -32:07:37.7 & ( 57.7 ) & m0 & 11.93 & 8.03 & 7.85 & + hd 191089 & 20:09:05.2 & -26:13:27 & 52.2 & f5 & 7.23 & 6.09 & 6.08 & + gj 799a & 20:41:51.2 & -32:26:06.6 & 10.7 & m4 & 11.49 & 5.20 & 4.94 & + gj 803 & 20:45:09.5 & -31:20:27.1 & 9.9 & m1 & 8.93 & 4.83 & 4.53 & + cp-72 2713 & 22:42:49.0 & -71:42:21 & ( 36 ) & k7 & 10.69 & 7.12 & 6.89 & + hip 112312a & 22:44:57.8 & -33:15:01.0 & 23.3 & m4 & 12.10 & 7.15 & 6.93 & + tx psa & 22:45:00.0 & -33:15:26 & ( 20 ) & m5 & 13.36 & 8.06 & 7.79 & + bd-13 6424 & 23:32:30.9 & -12:15:52 & ( 28 ) & m0 & 10.90 & 6.77 & 6.57 & + + tuc / hor , 30 myr + hip 1481 & 00:18:26.1 & -63:28:39.0 & 41.5 & f8 & 7.52 & 6.25 & 6.15 & + hip 2729 & 00:34:51.2 & -61:54:58 & 43.9 & k4 & 9.66 & 6.72 & 6.53 & + hd 12894 & 02:04:35.0 & -54:52:54 & 47.8 & f4 & 6.49 & 5.49 & 5.45 & + hip 12394 & 02:39:35.2 & -68:16:01 & 46.6 & b9 & 4.09 & 4.43 & 4.25 & + cd-53 544 & 02:41:46.8 & -52:59:52 & ( 42 ) & k6 & 10.45 & 6.93 & 6.76 & + af hor & 02:41:47.3 & -52:59:31 & ( 42 ) & m2 & 12.21 & 7.85 & 7.64 & + cd-58 553 & 02:42:33.0 & -57:39:37 & ( 50 ) & k5 & 11.00 & 7.97 & 7.78 & + hip 24947 & 05:20:38.0 & -39:45:18 & 48.3 & f6 & 7.44 & 6.22 & 6.14 & + ab pic & 06:19:12.9 & -58:03:15 & 46.1 & k1 & 7.58 & 7.09 & 6.98 + hip 104308 & 21:07:51.2 & -54:12:59 & 70.9 & a5 & 6.72 & 6.12 & 6.07 & + hip 107345 & 21:44:30.1 & -60:58:38 & 43.6 & m0 & 11.69 & 8.09 & 7.87 & + hip 118121 & 23:57:35.0 & -64:17:53 & 47.4 & a1 & 5.00 & 4.95 & 4.82 & + + ab dor , 100 myr + hip 5191 & 01:06:26.1 & -14:17:46 & 47.3 & k1 & 9.54 & 7.43 & 7.34 & + hd 19668 & 03:09:42.29 & -09:34:46.59 & 37.4 & g0 & 8.58 & 6.79 & 6.70 & + hip 17695 & 03:47:23.3451 & -01:58:19.927 & 16.1 & m3 & 11.53 & 7.17 & 6.93 & + hd 25457 & 04:02:36.7449 & -00:16:08.123 & 18.8 & f6 & 5.44 & 4.34 & 4.18 & + lp 776 - 25 & 04:52:24.4 & -16:49:22 & ( 16 ) & m3 & 11.76 & 7.15 & 6.89 & + gj 2036b & 04:53:30.5 & -55:51:32 & 11.1 & m3 & 13.92 & 7.24 & 6.89 & + gj 2036a & 04:53:31.2 & -55:51:37 & 11.1 & m3 & 11.15 & 6.62 & 6.34 & + hip 25283 & 05:24:30.1 & -38:58:10 & 18.0 & k6 & 9.29 & 6.11 & 5.92 & + hip 26369 & 05:36:55.1 & -47:57:48 & 25.6 & k6 & 9.87 & 6.83 & 6.61 & + uy pic & 05:36:56.8 & -47:57:52.9 & 25.1 & k0 & 8.01 & 5.93 & 5.81 & + bd-13 1328 & 06:02:21.9 & -13:55:33 & ( 39 ) & k4 & 10.71 & 7.89 & 7.77 & + cd-35 2722 & 06:09:19.2 & -35:49:31 & ( 24 ) & m1 & 11.16 & 7.28 & 7.05 & + hd 45270 & 06:22:30.9 & -60:13:07.1 & 23.8 & g1 & 6.59 & 5.16 & 5.05 & + gsc 8894 - 0426 & 06:25:56.1 & -60:03:27 & ( 22 ) & m3 & 11.7 & 7.47 & 7.21 & + bd+07 1919 a & 08:07:09.09 & + 07:23:00.13 & 35.1 & k8 & 10.02 & 7.32 & 7.26 & + hd 92945 & 10:43:28.3 & -29:03:51.4 & 21.4 & k1 & 7.81 & 5.77 & 5.66 & + hip 81084 & 16:33:41.6081 & -09:33:11.954 & 30.7 & m0 & 11.40 & 7.78 & 7.55 & + hip 82688 & 16:54:08.2 & -04:20:24 & 46.7 & g0 & 7.89 & 6.48 & 6.36 & + hd 159911 & 17:37:46.5 & -13:14:47 & ( 45 ) & k4 & 10.22 & 7.02 & 6.84 & + + her / lyr , 200 myr + hd 70573 & 08:22:49.95 & + 01:51:33.55 & ( 45.7 ) & g1 & 8.74 & 7.28 & 7.19 & + dx leo & 09:32:43.7 & + 26:59:18.7 & 17.8 & k0 & 7.14 & 5.24 & 5.12 & + gj 560 a & 14:42:30.42 & -64:58:30.50 & 16.6 & a7 & 3.19 & 2.47 & 2.42 & + hd 139664 & 15:41:11.38 & -44:39:40.34 & 17.4 & f5 & 4.68 & 3.73 & 3.80 & + + possible ab dor or her / lyr , 200 myr + bd + 1 2447 & 10:28:55.5 & + 00:50:28 & 7.1 & m2 & 9.65 & 5.61 & 5.31 & likely ab dor , malo+2012 + [ tab : properties ] lccccc tw hya , 10 myr + twa 6 & 2009 - 01 - 17 & adi & 45 & 2718 & 153.0 + twa 7 & 2009 - 02 - 11 & asdi & 45 & 2701 & 119.3 + twa 7 & 2009 - 02 - 11 & adi & 20 & 1208 & 14.2 + twa 7 & 2010 - 02 - 28 & adi & 20 & 1208 & 12.0 + tw hya & 2009 - 01 - 18 & adi & 20 & 1208 & 7.0 + tw hya & 2009 - 01 - 18 & asdi & 55 & 3302 & 100.7 + tw hya & 2009 - 02 - 12 & adi & 20 & 1208 & 7.1 + tw hya & 2009 - 02 - 12 & asdi & 45 & 2701 & 68.7 + twa 14 & 2009 - 01 - 14 & adi & 45 & 2701 & 12.7 + twa 14 & 2010 - 01 - 10 & adi & 90 & 2701 & 48.2 + twa 14 & 2012 - 03 - 19 & adi & 45 & 2701 & 39.1 + twa 13a & 2009 - 03 - 10 & adi & 20 & 1208 & 3.4 + twa 13a & 2009 - 03 - 10 & asdi & 45 & 2701 & 13.1 + twa 13b & 2009 - 03 - 11 & asdi & 45 & 2701 & 26.7 + twa 8a & 2009 - 02 - 08 & adi & 20 & 1208 & 3.8 + twa 8a & 2009 - 02 - 08 & asdi & 52 & 3122 & 121.1 + twa 8a & 2010 - 04 - 08 & adi & 20 & 1208 & 1.0 + twa 8a & 2010 - 04 - 08 & asdi & 45 & 2701 & 17.2 + twa 8a & 2011 - 04 - 03 & asdi & 45 & 2701 & 116.7 + twa 9b & 2009 - 03 - 13 & adi & 45 & 2701 & 16.8 + twa 9a & 2009 - 03 - 09 & adi & 45 & 2701 & 28.7 + twa 9a & 2010 - 12 - 26 & adi & 20 & 1208 & 5.0 + twa 25 & 2009 - 03 - 07 & adi & 20 & 1200 & 4.3 + twa 25 & 2009 - 03 - 07 & asdi & 45 & 2701 & 14.8 + twa 25 & 2011 - 04 - 25 & adi & 20 & 1200 & 6.2 + twa 20 & 2009 - 02 - 06 & adi & 45 & 2718 & 21.0 + twa 20 & 2011 - 05 - 11 & adi & 45 & 2718 & 16.2 + twa 20 & 2011 - 05 - 16 & asdi & 45 & 2718 & 27.5 + twa 10 & 2009 - 01 - 15 & adi & 43 & 2581 & 39.1 + twa 10 & 2012 - 04 - 07 & adi & 20 & 1200 & 4.9 + hr 4796 b & 2009 - 02 - 12 & adi & 45 & 2718 & 34.8 + hr 4796 a & 2009 - 01 - 14 & asdi & 45 & 2821 & 25.2 + hr 4796 a & 2009 - 01 - 14 & adi & 20 & 1208 & 21.8 + hr 4796 a & 2012 - 04 - 06 & adi & 64 & 3793 & 82.4 + + beta pic , 12 myr + hr 9 & 2009 - 12 - 05 & adi & 20 & 1208 & 2.8 + hr 9 & 2009 - 12 - 05 & asdi & 45 & 2736 & 22.8 + hip 10679 & 2011 - 10 - 16 & asdi & 45 & 2701 & 13.3 + hd 15115 & 2009 - 12 - 04 & adi & 20 & 1208 & 6.2 + hd 15115 & 2009 - 12 - 04 & asdi & 45 & 2718 & 17.4 + hd 15115 & 2011 - 11 - 07 & adi & 20 & 1185 & 8.6 + hd 15115 & 2011 - 11 - 22 & adi & 40 & 2371 & 13.7 + 51 eri & 2008 - 11 - 20 & asdi & 60 & 3830 & 39.3 + gj 182 & 2009 - 02 - 08 & adi & 20 & 1208 & 6.9 + gj 182 & 2009 - 02 - 08 & asdi & 35 & 2686 & 18.9 + hip 23309 & 2009 - 02 - 07 & asdi & 40 & 3009 & 26.3 + hip 23309 & 2009 - 02 - 07 & adi & 20 & 1208 & 7.8 + bd-21 1074a & 2008 - 12 - 16 & asdi & 45 & 2701 & 48.0 + hip 25486 & 2009 - 01 - 14 & asdi & 45 & 2479 & 34.6 + hip 25486 & 2009 - 01 - 14 & adi & 20 & 1208 & 12.5 + beta pic & 2008 - 11 - 22 & asdi & 90 & 5472 & 68.6 + beta pic & 2009 - 12 - 03 & asdi & 131 & 7964 & 66.8 + beta pic & 2009 - 12 - 04 & adi & 146 & 8654 & 72.7 + ao men & 2009 - 02 - 08 & adi & 20 & 1208 & 7.0 + ao men & 2009 - 02 - 08 & asdi & 43 & 2581 & 16.2 + ao men & 2010 - 01 - 10 & adi & 20 & 1208 & 6.9 + hd 139084 b & 2009 - 02 - 11 & adi & 19 & 1141 & 7.0 + v343 nor & 2009 - 03 - 07 & asdi & 45 & 2718 & 16.4 + hd 155555 c & 2009 - 04 - 08 & adi & 20 & 1208 & 6.7 + hd 155555 c & 2009 - 04 - 08 & asdi & 45 & 2701 & 16.7 + hd 155555 c & 2010 - 04 - 10 & adi & 20 & 1208 & 5.7 + cd-54 7336 & 2009 - 04 - 09 & asdi & 45 & 2701 & 17.2 + cd-54 7336 & 2009 - 04 - 09 & adi & 20 & 1147 & 6.8 + hd 164249b & 2009 - 04 - 10 & asdi & 36 & 2161 & 23.3 + hd 164249b & 2009 - 04 - 10 & adi & 20 & 1200 & 10.8 + hd 164249b & 2010 - 04 - 09 & adi & 20 & 1200 & 12.2 + hd 172555 a & 2009 - 04 - 09 & adi & 20 & 1208 & 6.8 + hd 172555 a & 2009 - 04 - 09 & asdi & 47 & 2786 & 22.0 + tyc 9073 - 0762 - 1 & 2009 - 04 - 09 & adi & 45 & 2701 & 20.9 + tyc 9073 - 0762 - 1 & 2010 - 05 - 09 & adi & 45 & 2701 & 14.2 + cd-31 16041 & 2009 - 04 - 08 & adi & 45 & 2701 & 14.0 + pz tel & 2009 - 04 - 11 & asdi & 45 & 2701 & 16.7 + pz tel & 2009 - 04 - 11 & adi & 20 & 1208 & 5.8 + pz tel & 2010 - 05 - 09 & asdi & 45 & 2701 & 23.2 + hr 7329 & 2009 - 04 - 11 & adi & 20 & 1208 & 8.1 + hr 7329 & 2009 - 04 - 11 & asdi & 45 & 2565 & 22.4 + hip 95270 & 2009 - 04 - 13 & adi & 20 & 1208 & 7.9 + hip 95270 & 2009 - 04 - 13 & asdi & 40 & 2416 & 25.8 + tyc 7443 - 1102 - 1a & 2010 - 05 - 09 & adi & 45 & 2701 & 9.8 + tyc 7443 - 1102 - 1b & 2010 - 05 - 10 & adi & 45 & 2701 & 14.4 + hd 191089 & 2010 - 05 - 11 & asdi & 45 & 2718 & 24.1 + hd 191089 & 2010 - 05 - 11 & adi & 20 & 1208 & 73.1 + gj 799a & 2011 - 05 - 16 & asdi & 33 & 2006 & 23.2 + gj 803 & 2010 - 05 - 09 & asdi & 59 & 3609 & 65.6 + gj 803 & 2010 - 08 - 28 & adi & 53 & 3021 & 36.9 + cp-72 2713 & 2011 - 09 - 13 & adi & 40 & 2416 & 14.9 + hip 112312a & 2010 - 11 - 22 & asdi & 45 & 2701 & 8.8 + hip 112312a & 2010 - 11 - 22 & adi & 20 & 1208 & 2.7 + tx psa & 2010 - 08 - 27 & adi & 47 & 2821 & 30.0 + tx psa & 2011 - 09 - 14 & adi & 45 & 2701 & 76.4 + bd-13 6424 & 2010 - 07 - 27 & adi & 20 & 1208 & 15.3 + bd-13 6424 & 2010 - 07 - 27 & asdi & 44 & 2641 & 29.6 + + tuc / hor , 30 myr + hip 1481 & 2010 - 11 - 25 & asdi & 45 & 2642 & 20.8 + hip 1481 & 2010 - 12 - 26 & adi & 25 & 1510 & 14.7 + hip 2729 & 2011 - 11 - 22 & adi & 40 & 2416 & 20.1 + hd 12894 & 2011 - 11 - 03 & adi & 40 & 2371 & 24.5 + hip 12394 & 2008 - 12 - 15 & asdi & 42 & 2633 & 21.3 + hip 12394 & 2008 - 12 - 15 & adi & 20 & 1200 & 7.9 + cd-53 544 & 2011 - 09 - 15 & adi & 20 & 1208 & 13.1 + af hor & 2011 - 11 - 22 & adi & 32 & 1921 & 21.2 + cd-58 553 & 2009 - 01 - 17 & adi & 45 & 2718 & 18.7 + hip 24947 & 2008 - 11 - 22 & asdi & 90 & 5335 & 32.2 + ab pic & 2009 - 03 - 07 & adi & 20 & 1208 & 8.9 + ab pic & 2009 - 03 - 07 & asdi & 45 & 2701 & 23.0 + hip 104308 & 2011 - 10 - 30 & adi & 40 & 2416 & 17.6 + hip 107345 & 2011 - 11 - 03 & adi & 33 & 1981 & 14.9 + hip 118121 & 2010 - 08 - 29 & adi & 20 & 1208 & 8.5 + + ab dor , 100 myr + hip 5191 & 2011 - 11 - 04 & adi & 43 & 2598 & 44.0 + hd 19668 & 2010 - 08 - 29 & adi & 20 & 1208 & 9.1 + hd 19668 & 2010 - 08 - 29 & asdi & 45 & 2701 & 30.6 + hip 17695 & 2009 - 01 - 18 & asdi & 45 & 2701 & 24.0 + hip 17695 & 2009 - 01 - 18 & adi & 20 & 1208 & 8.4 + hd 25457 & 2009 - 01 - 13 & asdi & 45 & 2633 & 23.9 + hd 25457 & 2009 - 01 - 13 & adi & 20 & 1208 & 8.5 + lp 776 - 25 & 2008 - 11 - 17 & adi & 45 & 2687 & 51.3 + lp 776 - 25 & 2008 - 11 - 17 & asdi & 45 & 2687 & 28.3 + gj 2036b & 2009 - 01 - 16 & asdi & 45 & 2701 & 16.9 + gj 2036a & 2008 - 12 - 16 & asdi & 43 & 2581 & 26.3 + gj 2036a & 2008 - 12 - 16 & adi & 20 & 1200 & 11.9 + hip 25283 & 2009 - 10 - 30 & adi & 20 & 1208 & 10.0 + hip 25283 & 2009 - 10 - 30 & asdi & 45 & 2719 & 53.7 + hip 25283 & 2010 - 12 - 26 & adi & 20 & 1208 & 5.3 + hip 25283 & 2011 - 10 - 21 & adi & 20 & 1208 & 5.8 + hip 26369 & 2008 - 11 - 21 & asdi & 47 & 2839 & 20.6 + uy pic & 2009 - 03 - 13 & asdi & 45 & 2719 & 24.3 + uy pic & 2009 - 03 - 13 & adi & 20 & 1208 & 7.4 + uy pic & 2009 - 12 - 04 & adi & 20 & 1208 & 10.8 + uy pic & 2009 - 12 - 04 & asdi & 45 & 2719 & 34.4 + uy pic & 2010 - 04 - 09 & asdi & 50 & 3021 & 15.1 + bd-13 1328 & 2009 - 01 - 15 & adi & 32 & 1933 & 29.2 + bd-13 1328 & 2010 - 01 - 05 & adi & 45 & 2718 & 2.8 + cd-35 2722 & 2009 - 01 - 16 & asdi & 45 & 2701 & 56.2 + cd-35 2722 & 2009 - 01 - 16 & adi & 20 & 1208 & 46.5 + cd-35 2722 & 2010 - 01 - 10 & asdi & 45 & 2701 & 49.2 + hd 45270 & 2009 - 02 - 12 & asdi & 45 & 2650 & 22.8 + hd 45270 & 2009 - 02 - 12 & adi & 20 & 1208 & 9.8 + gsc 8894 - 0426 & 2008 - 12 - 17 & adi & 50 & 3002 & 18.7 + gsc 8894 - 0426 & 2010 - 02 - 28 & adi & 50 & 3002 & 20.5 + bd+07 1919 a & 2009 - 03 - 12 & asdi & 45 & 2701 & 18.4 + bd+07 1919 a & 2009 - 03 - 12 & adi & 20 & 1208 & 7.7 + bd+07 1919 a & 2010 - 02 - 28 & adi & 23 & 1389 & 11.0 + hd 92945 & 2009 - 01 - 15 & asdi & 45 & 2770 & 51.3 + hd 92945 & 2009 - 01 - 15 & adi & 20 & 1208 & 30.2 + hd 92945 & 2010 - 04 - 08 & adi & 45 & 2718 & 153.1 + hd 92945 & 2010 - 04 - 10 & asdi & 45 & 2770 & 155.6 + hd 92945 & 2011 - 04 - 02 & adi & 48 & 2900 & 9.9 + hd 92945 & 2011 - 05 - 15 & adi & 20 & 1208 & 1.1 + hip 81084 & 2009 - 04 - 10 & asdi & 45 & 2701 & 14.8 + hip 81084 & 2009 - 04 - 10 & adi & 27 & 1631 & 4.8 + hip 82688 & 2010 - 04 - 09 & adi & 20 & 1208 & 9.1 + hip 82688 & 2010 - 04 - 09 & asdi & 45 & 2701 & 25.2 + hip 82688 & 2011 - 03 - 21 & adi & 26 & 1570 & 10.9 + hd 159911 & 2010 - 04 - 10 & asdi & 45 & 2701 & 32.7 + + her / lyr , 200 myr + hd 70573 & 2009 - 02 - 07 & asdi & 37 & 2225 & 20.1 + hd 70573 & 2009 - 02 - 07 & adi & 20 & 1208 & 9.3 + dx leo & 2009 - 03 - 13 & asdi & 45 & 2736 & 13.7 + dx leo & 2009 - 03 - 13 & adi & 20 & 1208 & 5.5 + gj 560 a & 2009 - 03 - 11 & asdi & 43 & 2451 & 19.1 + hd 139664 & 2009 - 02 - 06 & asdi & 37 & 2320 & 15.1 + hd 139664 & 2010 - 05 - 09 & asdi & 45 & 2821 & 22.7 + hd 139664 & 2010 - 05 - 09 & adi & 20 & 1208 & 14.3 + + possible ab dor or her / lyr , 200 myr + bd + 1 2447 & 2009 - 02 - 07 & adi & 20 & 1208 & 7.6 + bd + 1 2447 & 2009 - 02 - 07 & asdi & 45 & 2359 & 18.9 + [ tab : observations ] lrrrrrrrrrrrrrrrrrr + all stars + median contrast & 11.0 & 12.4 & 13.4 & 13.9 & 14.3 & 14.5 & 14.6 & 14.5 & 14.4 & - & - & - & - & - & - & - & - + + tw hya , 10 myr + twa 7 & 11.1 & 12.2 & 13.2 & 13.6 & 14.1 & 14.3 & 14.3 & 14.4 & 14.4 & 14.3 & 1.00 & 13.8 & 1.00 & 12.9 & 0.73 & 12.6 & 0.27 + tw hya & 10.5 & 11.6 & 12.5 & 12.8 & 13.4 & 13.5 & 13.7 & 13.7 & 13.7 & 13.2 & 1.00 & 11.8 & 1.00 & 10.6 & 0.69 & 11.9 & 0.20 + twa 13a & 10.7 & 12.1 & 12.7 & 13.0 & 13.4 & 13.4 & 13.3 & 13.4 & 13.1 & 13.1 & 0.95 & 13.1 & 0.65 & 12.7 & 0.32 & 12.1 & 0.07 + twa 13b & 10.0 & 11.2 & 12.2 & 12.6 & 12.8 & 12.9 & 12.9 & 12.8 & 12.7 & 12.7 & 1.00 & 12.2 & 0.85 & 11.2 & 0.33 & 11.0 & 0.04 + twa 8a & 11.1 & 12.2 & 13.0 & 13.4 & 13.8 & 13.9 & 14.0 & 14.1 & 14.0 & 14.0 & 1.00 & 13.4 & 1.00 & 12.7 & 0.72 & 12.4 & 0.19 + twa 25 & 11.3 & 12.6 & 13.5 & 13.8 & 14.0 & 14.2 & 14.0 & 14.0 & 14.0 & 13.8 & 0.97 & 13.4 & 0.72 & 12.2 & 0.27 & 12.6 & 0.04 + hr 4796 a & 11.6 & 12.6 & 13.6 & 14.0 & 14.8 & 15.0 & 15.0 & 15.1 & 14.8 & 15.0 & 1.00 & 15.0 & 1.00 & 15.0 & 1.00 & 15.0 & 1.00 + + beta pic , 12 myr + hr 9 & 11.0 & 12.6 & 13.6 & 14.1 & 14.8 & 15.0 & 15.1 & 15.0 & 15.1 & 14.8 & 0.98 & 14.5 & 0.74 & 13.7 & 0.35 & 12.9 & 0.07 + hip 10679 & 11.0 & 12.3 & 13.3 & 13.9 & 14.4 & 14.6 & 14.7 & 14.8 & 14.6 & 14.5 & 0.93 & 14.2 & 0.68 & 13.7 & 0.28 & 13.0 & 0.05 + hd 15115 & 11.4 & 13.0 & 14.1 & 14.5 & 15.0 & 15.1 & 15.2 & 15.0 & 15.0 & 14.9 & 0.98 & 14.6 & 0.71 & 13.8 & 0.29 & 13.1 & 0.04 + 51 eri & 9.1 & 10.7 & 12.6 & 13.2 & 13.6 & 13.8 & 13.8 & 13.8 & 13.6 & 13.7 & 0.89 & 13.4 & 0.66 & 12.9 & 0.28 & 12.1 & 0.05 + gj 182 & 11.4 & 12.9 & 13.8 & 14.2 & 14.8 & 14.9 & 14.9 & 14.9 & 14.7 & 14.7 & 0.98 & 14.2 & 0.74 & 13.1 & 0.30 & 13.5 & 0.05 + hip 23309 & 11.0 & 12.1 & 13.1 & 13.9 & 14.3 & 14.5 & 14.6 & 14.6 & 14.5 & 14.4 & 0.99 & 14.1 & 0.78 & 13.5 & 0.35 & 12.8 & 0.07 + bd-21 1074a & 10.6 & 11.7 & 12.6 & 12.8 & 13.2 & 13.9 & 13.9 & 13.8 & 13.8 & 13.3 & 1.00 & 12.8 & 0.89 & 11.9 & 0.52 & 11.9 & 0.07 + hip 25486 & 11.9 & 13.2 & 14.3 & 14.9 & 15.3 & 15.6 & 15.7 & 15.4 & 15.2 & 15.1 & 0.99 & 14.6 & 0.82 & 13.7 & 0.42 & 13.1 & 0.04 + ao men & 11.0 & 12.4 & 13.2 & 13.8 & 14.3 & 14.2 & 14.3 & 14.2 & 14.2 & 14.1 & 0.98 & 13.7 & 0.73 & 12.8 & 0.28 & 12.8 & 0.05 + v343 nor & 11.4 & 13.0 & 13.9 & 14.3 & 15.0 & 15.0 & 15.0 & 15.0 & 15.0 & 14.9 & 0.99 & 14.6 & 0.75 & 14.0 & 0.27 & 13.3 & 0.04 + hd 155555 c & 10.9 & 11.9 & 12.8 & 13.2 & 13.4 & 13.3 & 13.4 & 13.3 & 13.3 & 13.1 & 0.97 & 12.9 & 0.72 & 12.2 & 0.29 & 11.9 & 0.05 + cd-54 7336 & 11.1 & 12.1 & 12.9 & 13.6 & 13.7 & 13.9 & 13.8 & 13.8 & 13.9 & 13.6 & 0.98 & 13.4 & 0.73 & 12.7 & 0.29 & 12.5 & 0.05 + hd 164249b & 10.4 & 11.2 & 11.8 & 12.1 & 12.0 & 12.2 & 11.9 & 12.0 & 11.7 & 11.6 & 1.00 & 11.2 & 0.90 & 10.4 & 0.35 & 10.8 & 0.01 + hd 172555 a & 11.1 & 12.9 & 14.2 & 14.9 & 15.3 & 15.3 & 15.4 & 15.3 & 15.4 & 15.8 & 1.00 & 17.1 & 1.00 & 19.2 & 1.00 & 21.1 & 1.00 + pz tel & 10.8 & 12.5 & 13.5 & 14.0 & 14.4 & 14.5 & 14.8 & 14.7 & 14.6 & 14.5 & 0.99 & 14.2 & 0.74 & 13.5 & 0.27 & 13.0 & 0.04 + hr 7329 & 10.6 & 11.8 & 12.9 & 13.6 & 14.4 & 14.6 & 14.8 & 14.6 & 14.7 & 14.5 & 0.98 & 14.3 & 0.76 & 13.7 & 0.32 & 13.1 & 0.06 + hip 95270 & 12.1 & 13.3 & 14.1 & 14.6 & 14.9 & 14.9 & 15.1 & 15.0 & 14.9 & 14.7 & 0.99 & 14.2 & 0.80 & 12.6 & 0.37 & 12.5 & 0.07 + hd 191089 & 10.6 & 12.1 & 13.4 & 14.0 & 14.5 & 14.7 & 14.7 & 14.7 & 14.8 & 14.5 & 0.95 & 14.3 & 0.75 & 13.9 & 0.33 & 13.0 & 0.07 + gj 799a & 10.9 & 12.4 & 13.3 & 13.6 & 13.7 & 13.9 & 13.8 & 13.9 & 14.2 & 14.1 & 1.00 & 13.9 & 0.82 & 13.1 & 0.30 & 12.2 & 0.01 + gj 803 & 11.0 & 12.4 & 13.7 & 14.5 & 14.8 & 15.1 & 15.0 & 14.9 & 14.8 & 14.5 & 1.00 & 13.9 & 0.94 & 13.2 & 0.50 & 12.3 & 0.11 + cp-72 2713 & 6.3 & 8.3 & 10.3 & 11.0 & 12.0 & 12.0 & 12.3 & 12.3 & 12.3 & 12.0 & 0.98 & 11.9 & 0.65 & 11.5 & 0.18 & 11.0 & 0.01 + hip 112312a & 10.4 & 12.0 & 13.2 & 13.7 & 14.0 & 14.1 & 14.3 & 14.2 & 14.1 & 13.7 & 1.00 & 13.0 & 1.00 & 11.8 & 1.00 & 10.7 & 1.00 + bd-13 6424 & 10.7 & 12.2 & 13.4 & 14.0 & 14.2 & 14.2 & 14.4 & 14.3 & 14.2 & 14.0 & 0.97 & 13.4 & 0.77 & 12.4 & 0.37 & 11.9 & 0.07 + + tuc / hor , 30 myr + hip 1481 & 10.7 & 11.9 & 13.2 & 13.6 & 14.0 & 14.2 & 14.1 & 14.1 & 14.0 & 13.9 & 0.99 & 13.4 & 0.74 & 12.4 & 0.31 & - & - + hip 12394 & 11.5 & 12.9 & 14.2 & 15.1 & 15.3 & 15.5 & 15.5 & 15.4 & 15.2 & 15.2 & 0.98 & 15.0 & 0.74 & 14.3 & 0.32 & 13.4 & 0.03 + hip 24947 & 11.0 & 12.6 & 13.6 & 14.1 & 14.4 & 14.5 & 14.6 & 14.5 & 14.4 & 14.3 & 0.99 & 13.5 & 0.81 & 12.6 & 0.40 & 12.9 & 0.09 + ab pic & 11.7 & 12.8 & 13.8 & 14.1 & 14.4 & 14.6 & 14.4 & 14.4 & 14.2 & 14.2 & 1.00 & 13.7 & 0.74 & 12.7 & 0.35 & 12.2 & 0.06 + + ab dor , 100 myr + hip 17695 & 11.5 & 12.7 & 13.6 & 13.8 & 14.3 & 14.4 & 14.3 & 14.3 & 14.2 & 14.1 & 1.00 & 13.4 & 0.78 & 11.4 & 0.32 & 12.0 & 0.06 + hd 25457 & 11.7 & 13.1 & 14.3 & 15.1 & 15.4 & 15.6 & 15.7 & 15.3 & 15.5 & 15.4 & 0.99 & 14.9 & 0.78 & 14.1 & 0.35 & 14.0 & 0.07 + lp 776 - 25 & 8.6 & 10.3 & 11.5 & 12.4 & 13.1 & 13.4 & 13.2 & 13.1 & 13.0 & 12.7 & 0.98 & 12.2 & 0.82 & 11.6 & 0.52 & 11.1 & 0.15 + gj 2036b & 11.4 & 12.4 & 13.3 & 13.7 & 13.8 & 14.0 & 13.9 & 14.0 & 14.0 & 13.5 & 0.97 & 13.1 & 0.72 & 12.5 & 0.29 & 12.4 & 0.04 + gj 2036a & 10.6 & 11.9 & 12.8 & 13.4 & 13.9 & 14.0 & 14.2 & 14.2 & 14.1 & 13.8 & 0.99 & 13.2 & 0.78 & 12.6 & 0.36 & 12.3 & 0.03 + hip 25283 & 11.6 & 12.9 & 13.9 & 14.6 & 14.9 & 14.6 & 14.6 & 14.6 & 14.5 & 14.2 & 1.00 & 13.4 & 0.91 & 12.7 & 0.54 & 12.1 & 0.10 + hip 26369 & 10.0 & 11.8 & 13.1 & 13.5 & 13.9 & 14.0 & 14.2 & 14.1 & 14.1 & 13.9 & 0.99 & 13.6 & 0.76 & 13.0 & 0.31 & 12.6 & 0.06 + uy pic & 11.2 & 12.6 & 13.8 & 14.5 & 15.1 & 15.2 & 15.2 & 15.4 & 15.2 & 15.1 & 1.00 & 14.6 & 0.70 & 14.0 & 0.28 & 14.3 & 0.08 + cd-35 2722 & 10.8 & 12.0 & 12.7 & 13.3 & 13.4 & 13.5 & 13.4 & 13.3 & 13.3 & 12.8 & 1.00 & 11.5 & 0.92 & 10.8 & 0.55 & 10.9 & 0.07 + hd 45270 & 11.9 & 13.5 & 14.6 & 15.0 & 15.5 & 15.6 & 15.7 & 15.5 & 15.5 & 15.3 & 0.98 & 14.9 & 0.76 & 14.2 & 0.33 & 13.9 & 0.06 + bd+07 1919 a & 10.9 & 12.0 & 13.0 & 13.2 & 13.7 & 13.6 & 13.7 & 13.5 & 13.5 & 13.5 & 0.92 & 13.4 & 0.71 & 12.9 & 0.36 & 12.1 & 0.09 + hd 92945 & 11.6 & 13.1 & 14.0 & 14.6 & 15.3 & 15.3 & 15.5 & 15.6 & 15.6 & 15.5 & 1.00 & 15.0 & 0.90 & 14.4 & 0.50 & 14.1 & 0.08 + hip 81084 & 11.2 & 12.3 & 13.0 & 13.2 & 13.6 & 13.8 & 13.6 & 13.6 & 13.4 & 13.5 & 1.00 & 13.2 & 0.77 & 12.4 & 0.24 & 12.2 & 0.03 + hip 82688 & 11.1 & 12.5 & 13.4 & 13.9 & 14.3 & 14.4 & 14.4 & 14.4 & 14.3 & 14.1 & 0.96 & 13.9 & 0.76 & 13.4 & 0.34 & 12.5 & 0.08 + hd 159911 & 11.3 & 12.6 & 13.4 & 13.9 & 14.3 & 14.3 & 14.3 & 14.3 & 14.2 & 13.9 & 0.98 & 13.6 & 0.80 & 12.9 & 0.39 & 12.1 & 0.09 + + her / lyr , 200 myr + hd 70573 & 11.2 & 12.4 & 13.1 & 13.6 & 13.9 & 14.1 & 14.1 & 14.1 & 14.2 & 13.8 & 0.99 & 13.3 & 0.75 & 12.4 & 0.30 & 12.2 & 0.05 + dx leo & 10.6 & 12.2 & 13.4 & 14.1 & 14.7 & 14.9 & 14.9 & 15.0 & 15.0 & 14.9 & 1.00 & 14.6 & 0.74 & 13.6 & 0.24 & 12.8 & 0.02 + gj 560 a & 10.8 & 12.5 & 13.8 & 14.9 & 15.7 & 15.9 & 15.9 & 15.9 & 15.8 & 15.7 & 0.98 & 15.5 & 0.74 & 14.9 & 0.31 & 14.3 & 0.06 + hd 139664 & 11.6 & 12.9 & 14.5 & 15.2 & 15.8 & 15.7 & 15.7 & 15.8 & 15.6 & 15.6 & 1.00 & 15.3 & 0.78 & 14.7 & 0.33 & 14.2 & 0.03 + + possible ab dor or her / lyr , 200 myr + bd + 1 2447 & 11.6 & 13.0 & 13.9 & 14.3 & 14.8 & 15.3 & 15.3 & 15.2 & 15.1 & 15.0 & 0.98 & 14.8 & 0.74 & 14.2 & 0.30 & 13.7 & 0.05 + [ tab : contrastsasdi ] lrrrrrrrrrrrrrrrrrr + all stars + median contrast & 9.1 & 10.4 & 11.6 & 13.2 & 14.4 & 15.1 & 15.4 & 15.5 & 15.4 & - & - & - & - & - & - & - & - + + tw hya , 10 myr + twa 6 & 8.9 & 10.0 & 11.4 & 12.5 & 13.6 & 13.9 & 13.8 & 13.6 & 13.5 & 12.7 & 1.00 & 11.2 & 1.00 & 10.5 & 0.73 & 11.0 & 0.38 + twa 7 & 7.8 & 9.1 & 10.6 & 12.3 & 14.3 & 15.2 & 15.3 & 15.7 & 15.5 & 15.4 & 0.90 & 15.0 & 0.67 & 13.6 & 0.29 & 13.7 & 0.05 + tw hya & 7.4 & 9.3 & 11.7 & 13.0 & 14.4 & 14.9 & 14.9 & 15.0 & 14.9 & 15.1 & 0.88 & 14.7 & 0.63 & 14.0 & 0.26 & 13.4 & 0.04 + twa 14 & 9.5 & 11.0 & 12.6 & 13.7 & 14.3 & 14.2 & 14.3 & 14.3 & 14.2 & 14.1 & 0.88 & 13.8 & 0.65 & 13.0 & 0.29 & 11.6 & 0.05 + twa 13a & 5.9 & 7.8 & 10.1 & 11.6 & 13.5 & 13.9 & 14.0 & 13.9 & 13.9 & 13.9 & 0.86 & 13.9 & 0.61 & 13.5 & 0.23 & 13.0 & 0.02 + twa 8a & 7.1 & 9.0 & 11.2 & 12.9 & 14.4 & 14.7 & 14.9 & 14.7 & 14.6 & 14.7 & 0.87 & 14.7 & 0.60 & 14.4 & 0.21 & 14.5 & 0.02 + twa 9b & 9.9 & 11.2 & 12.6 & 13.3 & 13.7 & 13.8 & 13.7 & 13.6 & 13.6 & 13.2 & 0.90 & 13.1 & 0.68 & 12.1 & 0.31 & 12.0 & 0.06 + twa 9a & 9.2 & 10.1 & 11.8 & 12.8 & 14.4 & 14.8 & 14.7 & 14.8 & 14.7 & 14.4 & 0.94 & 14.2 & 0.75 & 13.3 & 0.37 & 13.3 & 0.09 + twa 25 & 7.6 & 9.6 & 11.6 & 13.2 & 14.5 & 15.0 & 15.0 & 14.9 & 14.9 & 14.9 & 0.87 & 14.7 & 0.62 & 14.4 & 0.24 & 13.6 & 0.03 + twa 20 & 9.2 & 10.8 & 12.1 & 13.3 & 14.0 & 14.2 & 14.2 & 14.3 & 14.3 & 14.1 & 0.89 & 13.7 & 0.67 & 12.5 & 0.31 & 11.9 & 0.06 + twa 10 & 10.2 & 11.2 & 12.8 & 13.8 & 14.7 & 14.7 & 14.5 & 14.5 & 14.2 & 14.0 & 0.95 & 13.4 & 0.80 & 12.8 & 0.27 & 13.0 & 0.03 + hr 4796 b & 8.6 & 9.9 & 11.4 & 12.4 & 13.5 & 13.5 & 13.5 & 13.4 & 13.1 & 12.5 & 0.94 & 12.1 & 0.78 & 11.8 & 0.39 & 12.0 & 0.11 + hr 4796 a & - & 10.8 & 12.3 & 13.4 & 15.2 & 15.9 & 16.2 & 16.2 & 16.1 & 17.6 & 0.01 & 21.6 & 0.01 & 27.7 & 0.01 & 33.4 & 0.01 + + beta pic , 12 myr + hr 9 & - & - & - & 1.4 & 14.1 & 15.3 & 16.5 & 16.6 & 17.0 & 17.0 & 0.85 & 17.0 & 0.58 & 16.9 & 0.24 & 16.5 & 0.03 + hd 15115 & 9.4 & 11.2 & 12.8 & 14.1 & 15.8 & 16.4 & 16.8 & 16.8 & 16.9 & 16.8 & 0.86 & 16.6 & 0.60 & 16.1 & 0.26 & 15.4 & 0.06 + gj 182 & - & - & 12.8 & 14.4 & 15.8 & 16.1 & 16.4 & 16.4 & 16.4 & 16.2 & 0.88 & 16.0 & 0.62 & 15.5 & 0.25 & 14.7 & 0.03 + hip 23309 & - & - & 11.3 & 12.6 & 14.4 & 15.2 & 15.7 & 15.7 & 15.8 & 15.6 & 0.88 & 15.4 & 0.64 & 14.9 & 0.26 & 14.6 & 0.04 + hip 25486 & - & - & - & 13.1 & 14.4 & 15.7 & 16.5 & 16.6 & 16.7 & 16.6 & 0.88 & 16.4 & 0.65 & 15.7 & 0.29 & 14.8 & 0.05 + ao men & - & 5.5 & 11.6 & 13.0 & 14.9 & 15.4 & 15.6 & 15.7 & 15.6 & 15.5 & 0.86 & 15.4 & 0.59 & 15.1 & 0.28 & 14.5 & 0.04 + hd 139084 b & 6.9 & 8.7 & 10.8 & 11.7 & 12.3 & 12.2 & 12.3 & 12.2 & 12.2 & 12.0 & 0.87 & 12.0 & 0.62 & 11.6 & 0.25 & 10.1 & 0.03 + hd 155555 c & 6.9 & 8.9 & 10.8 & 12.2 & 13.7 & 14.0 & 14.1 & 14.1 & 14.0 & 14.0 & 0.87 & 13.8 & 0.62 & 13.6 & 0.24 & 13.2 & 0.03 + cd-54 7336 & 8.0 & 9.9 & 12.0 & 13.5 & 14.8 & 15.1 & 15.0 & 15.2 & 15.1 & 15.0 & 0.87 & 14.9 & 0.62 & 14.5 & 0.25 & 14.3 & 0.03 + hd 164249b & 7.7 & 9.0 & 10.5 & 13.0 & 14.7 & 15.6 & 16.3 & 16.4 & 16.2 & 16.1 & 0.91 & 16.0 & 0.66 & 15.5 & 0.28 & 15.0 & 0.05 + hd 172555 a & - & - & 11.9 & 12.8 & 14.2 & 15.6 & 16.9 & 17.3 & 17.5 & 15.7 & 0.01 & 11.0 & 0.01 & 3.9 & 0.01 & - & - + tyc 9073 - 0762 - 1 & 9.1 & 10.4 & 12.1 & 13.3 & 14.6 & 14.7 & 14.9 & 14.8 & 14.7 & 16.6 & 0.01 & 21.8 & 0.01 & 29.7 & 0.01 & 37.0 & 0.01 + cd-31 16041 & 9.5 & 10.7 & 12.2 & 13.3 & 14.7 & 14.9 & 15.0 & 14.9 & 15.0 & 14.7 & 0.89 & 14.3 & 0.67 & 13.4 & 0.29 & 13.3 & 0.05 + pz tel & - & - & 11.8 & 13.2 & 15.0 & 15.5 & 16.1 & 16.0 & 16.1 & 16.0 & 0.87 & 15.8 & 0.62 & 15.5 & 0.25 & 14.9 & 0.03 + hr 7329 & - & - & - & 1.4 & 14.4 & 15.5 & 16.6 & 16.7 & 16.8 & 16.9 & 0.88 & 16.7 & 0.63 & 16.0 & 0.26 & 15.7 & 0.04 + hip 95270 & - & - & 12.6 & 14.0 & 15.7 & 16.3 & 16.7 & 16.7 & 16.9 & 16.6 & 0.87 & 16.4 & 0.63 & 16.1 & 0.26 & 15.6 & 0.04 + tyc 7443 - 1102 - 1b & 8.6 & 10.1 & 11.5 & 12.8 & 13.6 & 14.0 & 14.0 & 13.9 & 14.1 & 13.8 & 0.90 & 13.5 & 0.66 & 12.1 & 0.29 & 11.8 & 0.05 + tyc 7443 - 1102 - 1a & 8.6 & 10.2 & 12.2 & 13.4 & 14.2 & 14.4 & 14.6 & 14.6 & 14.6 & 14.5 & 0.90 & 14.2 & 0.65 & 13.1 & 0.26 & 12.6 & 0.04 + hd 191089 & - & - & 11.6 & 12.6 & 14.1 & 14.9 & 15.2 & 15.1 & 15.1 & 14.5 & 1.00 & 13.8 & 0.95 & 13.2 & 0.51 & 13.1 & 0.18 + gj 803 & 9.1 & 10.2 & 11.8 & 13.1 & 14.6 & 15.3 & 15.4 & 15.4 & 15.4 & 15.2 & 0.95 & 14.8 & 0.78 & 14.2 & 0.41 & 13.6 & 0.12 + cp-72 2713 & 10.0 & 11.5 & 13.0 & 14.2 & 15.4 & 15.6 & 15.7 & 15.7 & 15.7 & 15.7 & 0.86 & 15.5 & 0.63 & 15.3 & 0.33 & 14.6 & 0.07 + hip 112312a & 5.1 & 6.8 & 9.3 & 10.9 & 13.2 & 14.1 & 14.4 & 14.4 & 14.4 & 14.4 & 0.01 & 14.3 & 0.01 & 14.3 & 0.01 & 14.2 & 0.01 + tx psa & 9.2 & 10.5 & 12.1 & 13.2 & 14.3 & 14.5 & 14.5 & 14.5 & 14.4 & 14.3 & 1.00 & 13.8 & 0.96 & 13.4 & 0.53 & 13.3 & 0.23 + bd-13 6424 & 9.7 & 10.6 & 12.0 & 13.3 & 14.6 & 15.4 & 15.6 & 15.6 & 15.6 & 15.3 & 0.91 & 15.0 & 0.67 & 14.5 & 0.29 & 14.1 & 0.06 + + tuc / hor , 30 myr + hip 1481 & - & - & 11.5 & 12.8 & 14.4 & 15.2 & 15.5 & 15.6 & 15.4 & 15.4 & 0.89 & 15.0 & 0.65 & 13.6 & 0.31 & 13.6 & 0.06 + hip 2729 & 9.4 & 10.0 & 11.7 & 12.7 & 14.3 & 15.1 & 15.5 & 15.3 & 15.3 & 15.5 & 0.88 & 15.1 & 0.66 & 14.4 & 0.34 & 13.8 & 0.08 + hd 12894 & - & - & 11.2 & 12.2 & 14.0 & 14.8 & 15.4 & 15.5 & 15.6 & 15.3 & 0.89 & 14.8 & 0.70 & 14.1 & 0.36 & 13.5 & 0.09 + hip 12394 & - & - & - & - & 10.0 & 16.1 & 17.4 & 17.8 & 17.7 & 17.9 & 0.86 & 17.7 & 0.62 & 17.4 & 0.27 & 16.9 & 0.04 + cd-53 544 & 7.7 & 8.9 & 10.3 & 11.4 & 13.1 & 14.0 & 14.4 & 14.4 & 14.4 & 14.3 & 0.86 & 14.0 & 0.62 & 13.7 & 0.32 & 13.4 & 0.06 + af hor & 6.0 & 7.6 & 10.7 & 11.5 & 13.0 & 13.6 & 13.6 & 13.6 & 13.5 & 13.4 & 0.88 & 12.8 & 0.67 & 11.9 & 0.35 & 11.0 & 0.08 + cd-58 553 & 8.4 & 9.8 & 11.1 & 12.3 & 13.8 & 14.4 & 14.6 & 14.6 & 14.5 & 14.3 & 0.90 & 13.9 & 0.68 & 12.9 & 0.33 & 12.8 & 0.07 + ab pic & 9.1 & 10.5 & 12.4 & 14.0 & 15.1 & 15.4 & 15.6 & 15.4 & 15.5 & 15.2 & 1.00 & 14.8 & 0.93 & 14.3 & 0.48 & 14.0 & 0.08 + hip 104308 & - & - & 12.5 & 14.1 & 15.5 & 16.0 & 16.4 & 16.3 & 16.3 & 16.2 & 0.88 & 15.8 & 0.66 & 14.3 & 0.33 & 14.5 & 0.07 + hip 107345 & 9.5 & 10.7 & 12.3 & 13.6 & 14.3 & 14.7 & 14.5 & 14.4 & 14.5 & 14.3 & 0.87 & 14.0 & 0.64 & 13.5 & 0.32 & 12.2 & 0.07 + hip 118121 & - & - & - & 12.4 & 14.3 & 15.5 & 16.4 & 16.4 & 16.5 & 16.4 & 0.88 & 16.2 & 0.62 & 15.7 & 0.27 & 14.8 & 0.04 + + ab dor , 100 myr + hip 5191 & 9.8 & 10.9 & 12.4 & 13.3 & 14.7 & 15.0 & 15.0 & 14.8 & 14.7 & 14.4 & 0.95 & 13.7 & 0.80 & 13.2 & 0.45 & 13.1 & 0.15 + hd 19668 & 8.1 & 9.9 & 11.5 & 12.8 & 14.4 & 15.1 & 15.1 & 15.1 & 15.1 & 14.9 & 0.89 & 14.7 & 0.63 & 14.0 & 0.27 & 13.2 & 0.04 + hip 17695 & 9.1 & 10.8 & 12.6 & 13.5 & 14.8 & 15.1 & 15.3 & 15.5 & 15.3 & 15.2 & 0.87 & 14.9 & 0.62 & 14.4 & 0.27 & 13.5 & 0.04 + hd 25457 & 9.2 & 10.8 & 11.6 & 12.0 & 13.6 & 14.6 & 16.1 & 16.7 & 17.1 & 17.1 & 0.87 & 16.9 & 0.62 & 16.4 & 0.27 & 15.7 & 0.04 + lp 776 - 25 & 8.8 & 10.5 & 12.0 & 13.2 & 14.3 & 14.7 & 14.6 & 14.6 & 14.4 & 14.1 & 0.98 & 13.6 & 0.82 & 13.0 & 0.52 & 12.5 & 0.15 + gj 2036a & - & - & - & 12.5 & 14.2 & 15.0 & 15.4 & 15.5 & 15.4 & 15.2 & 0.88 & 14.8 & 0.64 & 14.5 & 0.29 & 14.2 & 0.05 + hip 25283 & 10.7 & 12.2 & 13.9 & 15.6 & 17.4 & 18.3 & 18.8 & 18.8 & 18.8 & 18.7 & 0.87 & 19.2 & 0.01 & 22.1 & 0.01 & 24.8 & 0.01 + uy pic & - & - & 12.3 & 13.3 & 15.0 & 15.7 & 16.8 & 16.8 & 16.9 & 16.6 & 0.96 & 16.1 & 0.81 & 15.8 & 0.36 & 17.3 & 0.09 + bd-13 1328 & 10.0 & 11.2 & 12.6 & 13.8 & 14.8 & 14.8 & 15.0 & 14.9 & 14.9 & 14.8 & 0.85 & 14.7 & 0.58 & 14.6 & 0.25 & 13.2 & 0.03 + cd-35 2722 & 9.3 & 10.0 & 11.4 & 12.6 & 14.0 & 14.0 & 14.0 & 13.9 & 13.7 & 13.2 & 0.97 & 12.5 & 0.84 & 11.6 & 0.44 & 11.6 & 0.15 + hd 45270 & - & - & 4.4 & 14.2 & 15.8 & 16.5 & 17.5 & 17.6 & 17.4 & 17.4 & 0.88 & 17.1 & 0.64 & 16.3 & 0.27 & 15.7 & 0.04 + gsc 8894 - 0426 & - & - & 12.6 & 14.1 & 15.4 & 15.7 & 15.8 & 16.0 & 15.8 & 15.7 & 0.90 & 15.2 & 0.68 & 13.9 & 0.33 & 14.9 & 0.06 + bd+07 1919 a & 9.7 & 11.0 & 12.9 & 14.1 & 15.6 & 16.2 & 16.1 & 16.0 & 15.3 & 16.0 & 0.87 & 15.8 & 0.63 & 15.2 & 0.26 & 14.6 & 0.04 + hd 92945 & 9.9 & 11.0 & 12.4 & 13.9 & 15.3 & 16.1 & 16.8 & 16.9 & 16.9 & 16.9 & 1.00 & 16.5 & 1.00 & 16.3 & 0.25 & 15.8 & 0.02 + hip 81084 & 6.9 & 8.9 & 11.1 & 12.6 & 14.2 & 14.6 & 14.7 & 14.7 & 14.8 & 14.6 & 0.87 & 14.4 & 0.62 & 14.0 & 0.24 & 13.4 & 0.03 + hip 82688 & 8.3 & 10.2 & 12.6 & 13.6 & 15.2 & 15.9 & 16.1 & 16.1 & 16.2 & 16.0 & 0.89 & 15.8 & 0.64 & 15.5 & 0.26 & 14.3 & 0.04 + + her / lyr , 200 myr + hd 70573 & 9.1 & 10.4 & 12.2 & 13.4 & 14.8 & 15.3 & 15.4 & 15.3 & 15.3 & 15.1 & 0.88 & 14.8 & 0.64 & 14.0 & 0.27 & 13.7 & 0.04 + dx leo & - & - & - & 12.7 & 14.4 & 15.5 & 16.3 & 16.6 & 16.6 & 16.5 & 0.89 & 16.3 & 0.64 & 15.6 & 0.26 & 14.3 & 0.03 + hd 139664 & - & - & - & - & 13.8 & 14.9 & 16.4 & 17.2 & 17.4 & 17.2 & 0.91 & 16.8 & 0.67 & 15.4 & 0.30 & 15.4 & 0.05 + + possible ab dor or her / lyr , 200 myr + bd + 1 2447 & - & - & 4.2 & 13.7 & 15.1 & 16.0 & 16.8 & 16.9 & 17.0 & 16.7 & 0.87 & 16.4 & 0.63 & 15.8 & 0.26 & 15.4 & 0.04 + [ tab : contrastsadi ] lrrrrrrrrr + all stars + median limiting magnitude & 16.4 & 17.7 & 18.9 & 20.1 & 21.6 & 22.0 & 22.5 & 22.5 & 22.5 + + tw hya , 10 myr + twa 6 & 17.1 & 18.2 & 19.6 & 20.7 & 21.8 & 22.1 & 22.0 & 21.8 & 21.7 + twa 7 & 15.5 & 16.9 & 18.3 & 20.1 & 22.0 & 22.9 & 23.1 & 23.4 & 23.3 + tw hya & 15.0 & 16.8 & 19.2 & 20.6 & 22.0 & 22.4 & 22.5 & 22.6 & 22.5 + twa 14 & 18.3 & 19.8 & 21.3 & 22.4 & 23.0 & 23.0 & 23.0 & 23.0 & 22.9 + twa 13a & 13.6 & 15.6 & 17.8 & 19.3 & 21.2 & 21.6 & 21.8 & 21.7 & 21.7 + twa 8a & 14.8 & 16.6 & 18.9 & 20.5 & 22.1 & 22.3 & 22.5 & 22.4 & 22.2 + twa 9b & 19.3 & 20.5 & 22.0 & 22.7 & 23.1 & 23.2 & 23.1 & 23.0 & 23.0 + twa 9a & 17.2 & 18.1 & 19.8 & 20.8 & 22.4 & 22.9 & 22.7 & 22.8 & 22.7 + twa 25 & 15.1 & 17.1 & 19.1 & 20.7 & 22.0 & 22.5 & 22.5 & 22.4 & 22.4 + twa 20 & 18.1 & 19.6 & 21.0 & 22.2 & 22.9 & 23.1 & 23.1 & 23.2 & 23.2 + twa 10 & 19.3 & 20.2 & 21.9 & 22.8 & 23.7 & 23.8 & 23.5 & 23.6 & 23.2 + hr 4796 b & 17.2 & 18.4 & 19.9 & 20.9 & 22.0 & 22.0 & 22.0 & 21.9 & 21.7 + hr 4796 a & - & 16.6 & 18.1 & 19.2 & 20.9 & 21.7 & 22.0 & 22.0 & 21.9 + + beta pic , 12 myr + hr 9 & - & - & - & 6.7 & 19.4 & 20.6 & 21.8 & 22.0 & 22.3 + hd 15115 & 15.2 & 17.0 & 18.6 & 19.9 & 21.6 & 22.2 & 22.6 & 22.6 & 22.8 + gj 182 & - & - & 19.2 & 20.8 & 22.2 & 22.6 & 22.9 & 22.8 & 22.9 + hip 23309 & - & - & 17.8 & 19.0 & 20.8 & 21.6 & 22.1 & 22.1 & 22.3 + hip 25486 & - & - & - & 18.2 & 19.5 & 20.8 & 21.6 & 21.7 & 21.8 + beta pic & - & - & - & - & 18.6 & 20.0 & 21.5 & 22.1 & 22.4 + ao men & - & 12.5 & 18.6 & 20.0 & 21.9 & 22.4 & 22.6 & 22.7 & 22.6 + hd 139084 b & 16.4 & 18.1 & 20.2 & 21.2 & 21.8 & 21.6 & 21.8 & 21.7 & 21.6 + hd 155555 c & 14.8 & 16.9 & 18.7 & 20.1 & 21.6 & 21.9 & 22.0 & 22.0 & 21.9 + cd-54 7336 & 15.5 & 17.5 & 19.6 & 21.1 & 22.3 & 22.7 & 22.5 & 22.8 & 22.6 + hd 164249b & 16.3 & 17.6 & 19.0 & 21.5 & 23.3 & 24.2 & 24.9 & 25.0 & 24.8 + hd 172555 a & - & - & 16.2 & 17.0 & 18.5 & 19.8 & 21.2 & 21.5 & 21.8 + tyc 9073 - 0762 - 1 & 17.5 & 18.8 & 20.4 & 21.7 & 23.0 & 23.1 & 23.2 & 23.2 & 23.1 + cd-31 16041 & 17.7 & 18.9 & 20.4 & 21.5 & 22.9 & 23.1 & 23.2 & 23.1 & 23.2 + pz tel & - & - & 18.3 & 19.6 & 21.5 & 22.0 & 22.5 & 22.5 & 22.6 + hr 7329 & - & - & - & 6.6 & 19.6 & 20.6 & 21.7 & 21.9 & 22.0 + hip 95270 & - & - & 18.6 & 20.0 & 21.7 & 22.3 & 22.7 & 22.7 & 22.8 + tyc 7443 - 1102 - 1b & 16.9 & 18.5 & 19.9 & 21.1 & 22.0 & 22.3 & 22.4 & 22.3 & 22.4 + tyc 7443 - 1102 - 1a & 16.7 & 18.2 & 20.2 & 21.4 & 22.2 & 22.4 & 22.6 & 22.6 & 22.6 + hd 191089 & - & - & 17.8 & 18.8 & 20.3 & 21.0 & 21.3 & 21.3 & 21.3 + gj 803 & 13.9 & 15.0 & 16.6 & 17.9 & 19.5 & 20.1 & 20.3 & 20.2 & 20.3 + cp-72 2713 & 17.6 & 19.1 & 20.6 & 21.8 & 23.0 & 23.2 & 23.3 & 23.3 & 23.3 + hip 112312a & 12.2 & 13.9 & 16.4 & 18.1 & 20.4 & 21.2 & 21.5 & 21.6 & 21.5 + tx psa & 17.2 & 18.5 & 20.2 & 21.3 & 22.3 & 22.6 & 22.5 & 22.6 & 22.5 + bd-13 6424 & 16.7 & 17.7 & 19.0 & 20.4 & 21.7 & 22.5 & 22.7 & 22.6 & 22.6 + + tuc / hor , 30 myr + hip 1481 & - & - & 17.8 & 19.1 & 20.6 & 21.4 & 21.7 & 21.8 & 21.6 + hip 2729 & 16.1 & 16.7 & 18.4 & 19.5 & 21.1 & 21.8 & 22.2 & 22.0 & 22.1 + hd 12894 & - & - & 16.7 & 17.7 & 19.5 & 20.3 & 20.9 & 21.0 & 21.1 + hip 12394 & - & - & - & - & 14.5 & 20.6 & 21.8 & 22.2 & 22.2 + cd-53 544 & 15.1 & 16.2 & 17.6 & 18.8 & 20.4 & 21.3 & 21.7 & 21.7 & 21.8 + af hor & 14.3 & 15.9 & 19.0 & 19.8 & 21.3 & 21.9 & 21.9 & 21.9 & 21.8 + cd-58 553 & 16.6 & 18.0 & 19.4 & 20.5 & 22.1 & 22.6 & 22.8 & 22.8 & 22.7 + ab pic & 16.1 & 17.6 & 19.5 & 21.1 & 22.2 & 22.5 & 22.7 & 22.5 & 22.6 + hip 104308 & - & - & 18.7 & 20.2 & 21.6 & 22.1 & 22.5 & 22.4 & 22.4 + hip 107345 & 17.6 & 18.8 & 20.3 & 21.7 & 22.4 & 22.8 & 22.6 & 22.5 & 22.6 + hip 118121 & - & - & - & 17.4 & 19.3 & 20.5 & 21.3 & 21.4 & 21.5 + + ab dor , 100 myr + hip 5191 & 17.2 & 18.3 & 19.8 & 20.7 & 22.2 & 22.4 & 22.4 & 22.3 & 22.1 + hd 19668 & 14.9 & 16.7 & 18.3 & 19.6 & 21.2 & 21.9 & 21.9 & 21.9 & 21.9 + hip 17695 & 16.3 & 17.9 & 19.8 & 20.7 & 22.0 & 22.3 & 22.5 & 22.6 & 22.5 + hd 25457 & 13.5 & 15.1 & 16.0 & 16.3 & 18.0 & 18.9 & 20.5 & 21.0 & 21.5 + lp 776 - 25 & 16.0 & 17.7 & 19.2 & 20.4 & 21.5 & 21.9 & 21.8 & 21.8 & 21.6 + gj 2036a & - & - & - & 19.3 & 20.9 & 21.7 & 22.1 & 22.3 & 22.1 + hip 25283 & 16.8 & 18.3 & 20.0 & 21.7 & 23.5 & 24.4 & 24.9 & 24.9 & 25.0 + uy pic & - & - & 18.2 & 19.2 & 21.0 & 21.7 & 22.7 & 22.8 & 22.9 + bd-13 1328 & 18.1 & 19.3 & 20.7 & 21.8 & 22.8 & 22.9 & 23.0 & 23.0 & 22.9 + cd-35 2722 & 16.5 & 17.3 & 18.7 & 19.9 & 21.3 & 21.3 & 21.2 & 21.2 & 21.0 + hd 45270 & - & - & 9.5 & 19.4 & 21.0 & 21.7 & 22.6 & 22.7 & 22.6 + gsc 8894 - 0426 & - & - & 20.9 & 22.4 & 23.7 & 24.0 & 24.1 & 24.3 & 24.1 + bd+07 1919 a & 17.0 & 18.3 & 20.2 & 21.5 & 22.9 & 23.5 & 23.4 & 23.3 & 22.6 + hd 92945 & 15.6 & 16.7 & 18.2 & 19.7 & 21.1 & 21.9 & 22.5 & 22.7 & 22.7 + hip 81084 & 14.6 & 16.6 & 18.9 & 20.4 & 22.0 & 22.4 & 22.5 & 22.5 & 22.6 + hip 82688 & 14.8 & 16.6 & 19.1 & 20.1 & 21.7 & 22.4 & 22.6 & 22.6 & 22.7 + + her / lyr , 200 myr + hd 70573 & 16.4 & 17.7 & 19.5 & 20.6 & 22.1 & 22.6 & 22.6 & 22.5 & 22.6 + dx leo & - & - & - & 17.9 & 19.6 & 20.7 & 21.5 & 21.8 & 21.8 + hd 139664 & - & - & - & - & 17.5 & 18.6 & 20.2 & 20.9 & 21.1 + + possible ab dor or her / lyr , 200 myr + bd + 1 2447 & - & - & 9.8 & 19.3 & 20.7 & 21.6 & 22.4 & 22.5 & 22.6 + [ tab : appmagadi ] lrrrrrrrrr + all stars + median detectable mass & 12.3 & 11.1 & 6.8 & 6.0 & 4.4 & 3.7 & 3.2 & 3.1 & 3.1 + + tw hya , 10 myr + twa 6 & 7.7 & 6.4 & 5.1 & 4.2 & 3.5 & 3.3 & 3.3 & 3.5 & 3.5 + twa 7 & 6.9 & 5.5 & 4.3 & 3.1 & 1.9 & 1.4 & 1.3 & 1.1 & 1.2 + tw hya & 10.9 & 7.1 & 4.7 & 3.7 & 2.8 & 2.6 & 2.5 & 2.4 & 2.5 + twa 14 & 6.8 & 5.3 & 4.0 & 3.4 & 3.0 & 3.0 & 3.0 & 3.0 & 3.0 + twa 13a & 15.3 & 9.8 & 6.0 & 4.7 & 3.3 & 3.1 & 3.0 & 3.1 & 3.1 + twa 8a & 10.7 & 6.9 & 4.8 & 3.6 & 2.6 & 2.4 & 2.3 & 2.4 & 2.5 + twa 9b & 4.4 & 3.6 & 2.6 & 2.2 & 1.9 & 1.9 & 1.9 & 2.0 & 2.0 + twa 9a & 6.3 & 5.4 & 4.0 & 3.4 & 2.4 & 2.1 & 2.2 & 2.1 & 2.2 + twa 25 & 10.4 & 6.6 & 4.7 & 3.6 & 2.8 & 2.4 & 2.4 & 2.5 & 2.5 + twa 20 & 6.5 & 5.1 & 4.0 & 3.2 & 2.8 & 2.6 & 2.6 & 2.6 & 2.6 + twa 10 & 4.6 & 3.9 & 2.9 & 2.2 & 1.7 & 1.7 & 1.8 & 1.8 & 2.0 + hr 4796 b & 7.4 & 6.0 & 4.7 & 3.9 & 3.3 & 3.2 & 3.2 & 3.3 & 3.5 + hr 4796 a & - & 8.6 & 6.4 & 5.2 & 3.9 & 3.5 & 3.3 & 3.3 & 3.3 + + beta pic , 12 myr + hr 9 & - & - & - & 56.2 & 4.6 & 3.7 & 2.9 & 2.8 & 2.5 + hd 15115 & 10.4 & 7.1 & 5.5 & 4.4 & 3.3 & 2.8 & 2.5 & 2.5 & 2.4 + gj 182 & - & - & 4.0 & 2.9 & 2.0 & 1.7 & 1.5 & 1.5 & 1.5 + hip 23309 & - & - & 5.3 & 4.2 & 3.0 & 2.4 & 2.1 & 2.0 & 2.0 + hip 25486 & - & - & - & 4.9 & 3.9 & 3.0 & 2.4 & 2.4 & 2.3 + beta pic & - & - & - & - & 4.0 & 3.1 & 2.0 & 1.6 & 1.3 + ao men & - & 19.3 & 5.3 & 4.1 & 2.8 & 2.4 & 2.3 & 2.2 & 2.3 + hd 139084 b & 7.4 & 5.7 & 4.0 & 3.3 & 2.9 & 2.9 & 2.9 & 2.9 & 3.0 + hd 155555 c & 9.0 & 6.4 & 4.9 & 3.7 & 2.7 & 2.5 & 2.4 & 2.4 & 2.4 + cd-54 7336 & 11.7 & 7.5 & 5.4 & 4.2 & 3.3 & 3.0 & 3.1 & 3.0 & 3.1 + hd 164249b & 8.1 & 6.6 & 5.3 & 3.3 & 2.1 & 1.5 & 1.2 & 1.2 & 1.2 + hd 172555 a & - & - & 6.8 & 6.0 & 4.8 & 3.8 & 2.9 & 2.6 & 2.4 + tyc 9073 - 0762 - 1 & 6.9 & 5.7 & 4.4 & 3.4 & 2.5 & 2.4 & 2.4 & 2.4 & 2.4 + cd-31 16041 & 6.6 & 5.5 & 4.3 & 3.5 & 2.5 & 2.4 & 2.3 & 2.4 & 2.3 + pz tel & - & - & 6.0 & 4.9 & 3.5 & 3.1 & 2.8 & 2.8 & 2.7 + hr 7329 & - & - & - & 56.2 & 4.9 & 4.0 & 3.2 & 3.1 & 3.1 + hip 95270 & - & - & 5.8 & 4.6 & 3.4 & 2.9 & 2.7 & 2.7 & 2.6 + tyc 7443 - 1102 - 1b & 7.8 & 6.1 & 4.9 & 4.0 & 3.3 & 3.1 & 3.1 & 3.1 & 3.1 + tyc 7443 - 1102 - 1a & 8.1 & 6.3 & 4.7 & 3.7 & 3.2 & 3.1 & 2.9 & 2.9 & 2.9 + hd 191089 & - & - & 6.6 & 5.7 & 4.4 & 3.8 & 3.6 & 3.7 & 3.7 + gj 803 & 6.8 & 5.8 & 4.4 & 3.5 & 2.4 & 2.0 & 1.9 & 1.9 & 1.9 + cp-72 2713 & 6.0 & 4.7 & 3.6 & 2.7 & 1.9 & 1.7 & 1.7 & 1.7 & 1.7 + hip 112312a & 15.4 & 10.0 & 6.2 & 4.8 & 3.1 & 2.5 & 2.3 & 2.2 & 2.2 + tx psa & 5.2 & 4.1 & 3.0 & 2.2 & 1.5 & 1.3 & 1.3 & 1.3 & 1.4 + bd-13 6424 & 6.2 & 5.4 & 4.4 & 3.3 & 2.4 & 1.9 & 1.7 & 1.7 & 1.8 + + tuc / hor , 30 myr + hip 1481 & - & - & 8.6 & 7.6 & 5.7 & 4.9 & 4.7 & 4.6 & 4.8 + hip 2729 & 10.4 & 9.6 & 8.1 & 7.4 & 5.4 & 4.7 & 4.3 & 4.5 & 4.5 + hd 12894 & - & - & 9.7 & 8.8 & 7.5 & 6.6 & 5.7 & 5.7 & 5.6 + hip 12394 & - & - & - & - & 17.5 & 5.9 & 4.8 & 4.5 & 4.5 + cd-53 544 & 12.2 & 10.0 & 8.8 & 7.8 & 5.9 & 5.0 & 4.7 & 4.7 & 4.7 + af hor & 17.5 & 10.5 & 7.7 & 6.9 & 5.0 & 4.6 & 4.6 & 4.6 & 4.7 + cd-58 553 & 10.0 & 8.7 & 7.7 & 6.4 & 4.8 & 4.2 & 4.0 & 4.0 & 4.1 + ab pic & 10.4 & 8.9 & 7.5 & 5.5 & 4.5 & 4.1 & 4.0 & 4.2 & 4.1 + hip 104308 & - & - & 8.8 & 7.6 & 5.8 & 5.4 & 5.0 & 5.1 & 5.1 + hip 107345 & 8.8 & 7.9 & 6.1 & 4.8 & 4.1 & 3.8 & 4.0 & 4.0 & 4.0 + hip 118121 & - & - & - & 9.1 & 7.7 & 6.1 & 5.3 & 5.3 & 5.2 + + ab dor , 100 myr + hip 5191 & 19.6 & 12.2 & 11.4 & 10.9 & 10.2 & 10.0 & 10.0 & 10.1 & 10.2 + hd 19668 & 28.8 & 19.6 & 11.9 & 11.3 & 10.4 & 10.1 & 10.0 & 10.0 & 10.0 + hip 17695 & 12.0 & 11.2 & 10.2 & 9.7 & 9.1 & 8.9 & 8.8 & 8.7 & 8.8 + hd 25457 & 27.8 & 20.1 & 12.3 & 12.2 & 11.3 & 10.8 & 10.0 & 9.7 & 9.5 + lp 776 - 25 & 12.1 & 11.3 & 10.5 & 9.9 & 9.3 & 9.1 & 9.2 & 9.2 & 9.3 + gj 2036a & - & - & - & 10.0 & 9.2 & 8.8 & 8.5 & 8.5 & 8.5 + hip 25283 & 11.9 & 11.1 & 10.2 & 9.3 & 8.4 & 7.9 & 7.7 & 7.7 & 7.7 + uy pic & - & - & 11.5 & 11.0 & 10.1 & 9.7 & 9.2 & 9.2 & 9.1 + bd-13 1328 & 12.1 & 11.4 & 10.8 & 10.1 & 9.6 & 9.6 & 9.5 & 9.5 & 9.6 + cd-35 2722 & 12.3 & 11.9 & 11.2 & 10.6 & 9.9 & 9.8 & 9.9 & 9.9 & 10.0 + hd 45270 & - & - & - & 10.9 & 10.0 & 9.7 & 9.2 & 9.1 & 9.2 + gsc 8894 - 0426 & - & - & 10.0 & 9.2 & 8.5 & 8.4 & 8.3 & 8.2 & 8.3 + bd+07 1919 a & 12.5 & 11.8 & 10.9 & 10.2 & 9.5 & 9.2 & 9.2 & 9.2 & 9.6 + hd 92945 & 19.1 & 12.1 & 11.3 & 10.6 & 9.8 & 9.4 & 9.1 & 9.0 & 9.0 + hip 81084 & 27.4 & 12.5 & 11.4 & 10.6 & 9.8 & 9.6 & 9.5 & 9.5 & 9.5 + hip 82688 & 31.2 & 22.5 & 11.7 & 11.2 & 10.4 & 10.0 & 9.9 & 9.9 & 9.9 + + her / lyr , 200 myr + hd 70573 & 30.2 & 24.7 & 16.9 & 11.9 & 5.7 & 4.7 & 4.6 & 4.7 & 4.7 + dx leo & - & - & - & 14.8 & 7.7 & 4.6 & 3.9 & 3.6 & 3.6 + hd 139664 & - & - & - & - & 16.0 & 11.5 & 5.3 & 4.4 & 4.1 + + possible ab dor or her / lyr , 200 myr + bd + 1 2447 & - & - & 42.5 & 9.5 & 8.8 & 8.3 & 7.9 & 7.9 & 7.8 + [ tab : minmassadi_dusty ] lrrrrrrrrr + all stars + median detectable mass & 10.8 & 6.1 & 3.6 & 2.0 & 1.4 & 1.1 & 1.1 & 1.1 & 1.1 + + tw hya , 10 myr + twa 6 & 6.1 & 4.2 & 2.6 & 1.8 & 1.3 & 1.1 & 1.2 & 1.3 & 1.3 + twa 7 & 4.9 & 3.0 & 1.9 & 1.0 & 1.0 & 1.0 & 1.0 & 1.0 & 1.0 + tw hya & 9.7 & 5.1 & 2.2 & 1.4 & 1.0 & 1.0 & 1.0 & 1.0 & 1.0 + twa 14 & 4.8 & 2.8 & 1.6 & 1.2 & 1.0 & 1.0 & 1.0 & 1.0 & 1.0 + twa 13a & 15.2 & 8.1 & 3.6 & 2.1 & 1.2 & 1.0 & 1.0 & 1.0 & 1.0 + twa 8a & 9.4 & 4.9 & 2.3 & 1.3 & 1.0 & 1.0 & 1.0 & 1.0 & 1.0 + twa 9b & 1.9 & 1.3 & 1.0 & 1.0 & 1.0 & 1.0 & 1.0 & 1.0 & 1.0 + twa 9a & 4.0 & 2.9 & 1.6 & 1.2 & 1.0 & 1.0 & 1.0 & 1.0 & 1.0 + twa 25 & 9.0 & 4.5 & 2.1 & 1.3 & 1.0 & 1.0 & 1.0 & 1.0 & 1.0 + twa 20 & 4.3 & 2.6 & 1.6 & 1.1 & 1.0 & 1.0 & 1.0 & 1.0 & 1.0 + twa 10 & 2.0 & 1.5 & 1.0 & 1.0 & 1.0 & 1.0 & 1.0 & 1.0 & 1.0 + hr 4796 b & 5.7 & 3.6 & 2.2 & 1.5 & 1.1 & 1.1 & 1.1 & 1.2 & 1.3 + hr 4796 a & - & 7.1 & 4.2 & 2.7 & 1.5 & 1.3 & 1.1 & 1.1 & 1.2 + + beta pic , 12 myr + hr 9 & - & - & - & 51.9 & 1.9 & 1.2 & 1.1 & 1.1 & 1.1 + hd 15115 & 8.6 & 4.7 & 2.7 & 1.8 & 1.1 & 1.1 & 1.1 & 1.1 & 1.1 + gj 182 & - & - & 1.5 & 1.1 & 1.1 & 1.1 & 1.1 & 1.1 & 1.1 + hip 23309 & - & - & 2.4 & 1.6 & 1.1 & 1.1 & 1.1 & 1.1 & 1.1 + hip 25486 & - & - & - & 2.1 & 1.4 & 1.1 & 1.1 & 1.1 & 1.1 + beta pic & - & - & - & - & 1.5 & 1.1 & 1.1 & 1.1 & 1.0 + ao men & - & 19.3 & 2.4 & 1.6 & 1.1 & 1.1 & 1.1 & 1.1 & 1.1 + hd 139084 b & 5.2 & 2.9 & 1.5 & 1.1 & 1.1 & 1.1 & 1.1 & 1.1 & 1.1 + hd 155555 c & 7.5 & 3.9 & 2.1 & 1.2 & 1.1 & 1.1 & 1.1 & 1.1 & 1.1 + cd-54 7336 & 10.2 & 5.4 & 2.6 & 1.6 & 1.1 & 1.1 & 1.1 & 1.1 & 1.1 + hd 164249b & 6.3 & 4.1 & 2.4 & 1.1 & 1.1 & 1.1 & 1.0 & 1.0 & 1.0 + hd 172555 a & - & - & 4.4 & 3.3 & 2.0 & 1.3 & 1.1 & 1.1 & 1.1 + tyc 9073 - 0762 - 1 & 4.6 & 2.9 & 1.7 & 1.1 & 1.1 & 1.1 & 1.1 & 1.1 & 1.1 + cd-31 16041 & 4.1 & 2.7 & 1.7 & 1.1 & 1.1 & 1.1 & 1.1 & 1.1 & 1.1 + pz tel & - & - & 3.3 & 2.1 & 1.1 & 1.1 & 1.1 & 1.1 & 1.1 + hr 7329 & - & - & - & 51.9 & 2.1 & 1.5 & 1.1 & 1.1 & 1.1 + hip 95270 & - & - & 3.0 & 1.9 & 1.1 & 1.1 & 1.1 & 1.1 & 1.1 + tyc 7443 - 1102 - 1b & 5.8 & 3.4 & 2.1 & 1.5 & 1.1 & 1.1 & 1.1 & 1.1 & 1.1 + tyc 7443 - 1102 - 1a & 6.3 & 3.7 & 2.0 & 1.2 & 1.1 & 1.1 & 1.1 & 1.1 & 1.1 + hd 191089 & - & - & 4.1 & 2.9 & 1.8 & 1.3 & 1.2 & 1.2 & 1.2 + gj 803 & 4.4 & 3.0 & 1.8 & 1.1 & 1.1 & 1.1 & 1.1 & 1.1 & 1.1 + cp-72 2713 & 3.2 & 2.0 & 1.2 & 1.1 & 1.1 & 1.1 & 1.1 & 1.1 & 1.1 + hip 112312a & 15.1 & 8.3 & 3.6 & 2.0 & 1.1 & 1.1 & 1.1 & 1.1 & 1.1 + tx psa & 2.4 & 1.6 & 1.1 & 1.1 & 1.0 & 1.0 & 1.0 & 1.0 & 1.0 + bd-13 6424 & 3.6 & 2.6 & 1.7 & 1.1 & 1.1 & 1.1 & 1.1 & 1.1 & 1.1 + + tuc / hor , 30 myr + hip 1481 & - & - & 5.5 & 3.6 & 2.1 & 1.7 & 1.5 & 1.5 & 1.6 + hip 2729 & 8.7 & 7.6 & 4.5 & 3.2 & 1.9 & 1.5 & 1.4 & 1.5 & 1.5 + hd 12894 & - & - & 7.9 & 6.0 & 3.5 & 2.6 & 2.1 & 2.1 & 2.0 + hip 12394 & - & - & - & - & 13.1 & 2.3 & 1.6 & 1.5 & 1.5 + cd-53 544 & 10.7 & 8.2 & 5.8 & 4.0 & 2.3 & 1.7 & 1.5 & 1.5 & 1.5 + af hor & 13.1 & 8.9 & 3.7 & 2.8 & 1.7 & 1.5 & 1.5 & 1.5 & 1.5 + cd-58 553 & 8.2 & 5.6 & 3.7 & 2.6 & 1.6 & 1.4 & 1.4 & 1.4 & 1.4 + ab pic & 8.7 & 6.2 & 3.3 & 1.9 & 1.5 & 1.4 & 1.4 & 1.4 & 1.4 + hip 104308 & - & - & 5.8 & 3.6 & 2.2 & 1.9 & 1.7 & 1.7 & 1.7 + hip 107345 & 6.0 & 4.1 & 2.4 & 1.6 & 1.4 & 1.3 & 1.4 & 1.4 & 1.4 + hip 118121 & - & - & - & 6.6 & 3.7 & 2.4 & 1.8 & 1.8 & 1.8 + + ab dor , 100 myr + hip 5191 & 11.7 & 10.4 & 6.5 & 4.8 & 3.0 & 2.8 & 2.8 & 2.9 & 3.0 + hd 19668 & 25.0 & 11.7 & 9.0 & 6.1 & 3.6 & 2.9 & 2.8 & 2.8 & 2.8 + hip 17695 & 9.5 & 5.7 & 3.0 & 2.2 & 1.6 & 1.5 & 1.4 & 1.4 & 1.4 + hd 25457 & 23.2 & 11.8 & 10.8 & 10.4 & 6.3 & 4.5 & 2.7 & 2.2 & 2.0 + lp 776 - 25 & 10.1 & 6.1 & 3.7 & 2.5 & 1.8 & 1.7 & 1.7 & 1.7 & 1.8 + gj 2036a & - & - & - & 2.8 & 1.8 & 1.4 & 1.2 & 1.2 & 1.2 + hip 25283 & 9.0 & 5.3 & 3.1 & 1.8 & 1.1 & 1.0 & 1.0 & 1.0 & 1.0 + uy pic & - & - & 6.9 & 5.0 & 2.9 & 2.2 & 1.8 & 1.7 & 1.7 + bd-13 1328 & 9.8 & 6.7 & 4.4 & 2.9 & 2.1 & 2.0 & 2.0 & 2.0 & 2.0 + cd-35 2722 & 10.8 & 9.0 & 5.9 & 4.0 & 2.5 & 2.4 & 2.5 & 2.5 & 2.7 + hd 45270 & - & - & - & 4.7 & 2.7 & 2.1 & 1.7 & 1.7 & 1.8 + gsc 8894 - 0426 & - & - & 2.7 & 1.8 & 1.2 & 1.1 & 1.1 & 1.0 & 1.1 + bd+07 1919 a & 11.2 & 8.8 & 4.7 & 3.0 & 2.0 & 1.7 & 1.7 & 1.8 & 2.0 + hd 92945 & 11.5 & 10.1 & 6.3 & 3.8 & 2.4 & 1.9 & 1.7 & 1.6 & 1.6 + hip 81084 & 22.2 & 11.2 & 6.5 & 4.0 & 2.3 & 2.0 & 2.0 & 2.0 & 2.0 + hip 82688 & 29.6 & 12.3 & 8.2 & 5.9 & 3.5 & 2.8 & 2.6 & 2.6 & 2.5 + her / lyr , 200 myr + hd 70573 & 22.6 & 15.0 & 9.6 & 6.5 & 4.2 & 3.7 & 3.6 & 3.7 & 3.7 + dx leo & - & - & - & 7.8 & 4.9 & 3.5 & 2.7 & 2.6 & 2.6 + hd 139664 & - & - & - & - & 8.5 & 6.3 & 4.0 & 3.3 & 3.0 + + possible ab dor or her / lyr , 200 myr + bd + 1 2447 & - & - & 42.3 & 2.0 & 1.4 & 1.1 & 1.0 & 1.0 & 1.0 + [ tab : minmassadi_cond ] ccccccccccc hip 2729 & 1 & 9.986 & 30.9 & 10.2 & 8.33 & 3 & 6.61 & 31.72 & 4 & bg + hd 12894 & 1 & 0.322 & 242.9 & 3.0 & 3.22 & 3 & 322.08 & 110.70 & 4 & cpm + cd -58 553 & 1 & 6.924 & 85.3 & 10.8 & 4.01 & 4 & 0.06 & 83.96 & 6 & bg + hd 19668 & 1 & 5.470 & 151.0 & 9.8 & 9.93 & 8 & 0.49 & 30.57 & 14 & bg + lp 776 - 25 & 1 & 2.570 & 87.8 & 7.5 & 3.13 & 2 & 0.16 & 67.91 & 2 & bg + lp 776 - 25 & 2 & 6.340 & 264.7 & 12.9 & 3.13 & 2 & 0.17 & 14.56 & 2 & bg + lp 776 - 25 & 3 & 7.010 & 270.1 & 12.8 & 3.13 & 2 & 1.28 & 14.48 & 2 & bg + gj 182 & 1 & 4.870 & 224.7 & 12.5 & 4.25 & 2 & 2.21 & 249.10 & 2 & bg + gj 182 & 2 & 7.210 & 236.9 & 10.5 & 4.25 & 2 & 3.98 & 150.44 & 2 & bg + hip 25283 & 1 & 8.162 & 83.7 & 15.0 & 1.98 & 3 & 1.60 & 8.96 & 4 & bg + uy pic & 1 & 0.757 & 138.7 & 11.2 & 1.07 & 3 & 4.98 & 86.99 & 4 & bg + bd -13 1328 & 1 & 2.057 & 314.3 & 12.2 & 0.98 & 2 & 0.07 & 33.11 & 2 & bg + cd -35 2722 & 1 & 3.148 & 243.8 & 4.7 & 2.76 & 4 & 12.08 & 5.59 & 6 & cpm + cd -35 2722 & 2 & 9.202 & 105.5 & 10.6 & 2.76 & 4 & 0.24 & 17.96 & 6 & bg + cd -35 2722 & 3 & 5.289 & 275.4 & 13.3 & 0.98 & 2 & 0.12 & 3.84 & 2 & bg + ao men & 1 & 1.501 & 353.4 & 13.3 & 0.92 & 2 & 0.26 & 13.60 & 2 & bg + ao men & 2 & 2.790 & 206.3 & 14.4 & 0.92 & 2 & 0.42 & 15.62 & 2 & bg + ao men & 3 & 4.086 & 307.3 & 13.3 & 0.92 & 2 & 0.25 & 9.31 & 2 & bg + ao men & 4 & 4.600 & 275.3 & 15.4 & 0.92 & 2 & 7.49 & 34.05 & 2 & bg + ao men & 5 & 5.233 & 230.0 & 14.2 & 0.92 & 2 & 0.63 & 7.95 & 2 & bg + ao men & 6 & 5.552 & 248.5 & 15.2 & 0.92 & 2 & 0.39 & 7.23 & 2 & bg + ao men & 7 & 5.768 & 335.8 & 14.9 & 0.92 & 2 & 0.17 & 10.24 & 2 & bg + ao men & 8 & 6.722 & 141.9 & 11.5 & 0.92 & 2 & 0.17 & 7.84 & 2 & bg + ao men & 9 & 8.333 & 151.4 & 13.0 & 0.92 & 2 & 0.87 & 16.91 & 2 & bg + ao men & 10 & 9.763 & 267.7 & 13.9 & 0.92 & 2 & 1.42 & 3.84 & 2 & bg + ab pic & 1 & 5.406 & 175.4 & 6.7 & & & & & & cpm + gsc 8894 -0426 & 1 & 5.232 & 159.8 & 12.5 & 1.20 & 2 & 0.75 & 106.77 & 2 & bg + bd + 07 1919 b & 1 & 0.398 & 108.2 & 2.45 & 0.94 & 2 & 48.10 & 2.07 & 2 & cpm + bd + 07 1919 a & 1 & 4.667 & 86.0 & 16.0 & 0.96 & 2 & 0.06 & 20.36 & 2 & bg + bd + 07 1919 a & 2 & 7.702 & 151.8 & 13.6 & 0.96 & 2 & 0.05 & 52.51 & 2 & bg + hd 70573 & 1 & 3.490 & 134.0 & 15.5 & 10.27 & 2 & 0.17 & 17.00 & 2 & bg + hd 70573 & 2 & 6.890 & 305.5 & 16.6 & 10.27 & 2 & 0.39 & 23.65 & 2 & bg + dx leo & 1 & 4.190 & 183.6 & 12.0 & 4.22 & 2 & 0.25 & 2186.44 & 2 & bg + twa 6 & 1 & 2.090 & 285.8 & 11.1 & 6.61 & 2 & 0.38 & 2.50 & 2 & bg + twa 7 & 1 & 3.203 & 121.3 & 8.6 & 1.05 & 2 & 0.14 & 59.09 & 2 & bg + twa 7 & 2 & 3.931 & 92.4 & 14.5 & 1.05 & 2 & 0.04 & 62.50 & 2 & bg + twa 7 & 3 & 4.948 & 172.8 & 13.6 & 1.05 & 2 & 0.45 & 21.66 & 2 & bg + twa 14 & 1 & 2.386 & 261.3 & 11.8 & 3.18 & 3 & 0.40 & 30.76 & 4 & bg + hd 92945 & 1 & 8.961 & 234.7 & 10.6 & 3.22 & 4 & 1.39 & 751.35 & 6 & bg + twa 8a & 1 & 10.550 & 37.4 & 7.4 & 10.58 & 2 & 0.01 & 40.63 & 2 & bg + twa 9a & 1 & 7.286 & 123.1 & 12.2 & 1.80 & 2 & 2.14 & 23.36 & 2 & bg + twa 9a & 2 & 8.260 & 92.2 & 10.7 & 1.80 & 2 & 1.47 & 36.68 & 2 & bg + twa 25 & 1 & 3.499 & 116.9 & 14.8 & 2.13 & 2 & 0.28 & 93.57 & 2 & bg + twa 25 & 2 & 8.615 & 10.0 & 11.5 & 2.13 & 2 & 0.17 & 23.40 & 2 & bg + twa 25 & 3 & 8.638 & 60.9 & 13.1 & 2.13 & 2 & 0.22 & 114.03 & 2 & bg + twa 25 & 4 & 8.943 & 344.5 & 10.3 & 2.13 & 2 & 0.31 & 7.96 & 2 & bg + twa 20 & 1 & 2.903 & 190.5 & 14.1 & 2.27 & 2 & 0.10 & 55.04 & 2 & bg + twa 20 & 2 & 6.694 & 10.0 & 11.1 & 2.27 & 2 & 1.44 & 25.93 & 2 & bg + twa 10 & 1 & 9.168 & 313.5 & 12.3 & 3.23 & 2 & 2.27 & 19.11 & 2 & bg + hr 4796 a & 1 & 4.472 & 322.7 & 8.6 & 18.07 & 3 & 0.69 & 57.48 & 4 & bg + hd 139664 & 1 & 1.940 & 48.0 & 13.9 & 1.25 & 2 & 0.89 & 534.56 & 2 & bg + hd 139664 & 2 & 4.550 & 303.9 & 15.0 & 1.25 & 2 & 0.40 & 166.53 & 2 & bg + hd 139664 & 3 & 5.620 & 95.0 & 14.5 & 1.25 & 2 & 0.04 & 265.72 & 2 & bg + hd 139664 & 4 & 5.640 & 267.8 & 14.5 & 1.25 & 2 & 3.24 & 236.41 & 2 & bg + hd 139664 & 5 & 6.240 & 115.8 & 13.1 & 1.25 & 2 & 0.27 & 117.47 & 2 & bg + hd 139664 & 6 & 6.770 & 2.4 & 14.6 & 1.25 & 2 & 0.04 & 443.94 & 2 & bg + hip 81084 & 1 & 6.190 & 238.6 & 12.7 & 3.97 & 2 & 0.69 & 790.36 & 2 & bg + hip 82688 & 1 & 3.819 & 56.9 & 12.9 & 8.55 & 5 & 655.97 & 11.84 & 8 & cpm + hip 82688 & 2 & 2.386 & 316.5 & 14.3 & 0.95 & 2 & 3.85 & 15.74 & 2 & bg + hip 82688 & 3 & 7.811 & 283.2 & 7.2 & 1.87 & 2 & 3.31 & 20.03 & 2 & bg + hd 155555 c & 1 & 2.668 & 164.3 & 10.9 & 1.01 & 2 & 1.09 & 48.30 & 2 & bg + hd 155555 c & 2 & 4.735 & 336.3 & 12.8 & 1.01 & 2 & 0.05 & 52.36 & 2 & bg + hd 155555 c & 3 & 5.634 & 37.3 & 8.0 & 1.01 & 2 & 0.06 & 46.33 & 2 & bg + hd 155555 c & 4 & 6.371 & 292.4 & 11.9 & 1.01 & 2 & 0.33 & 18.38 & 2 & bg + hd 155555 c & 5 & 7.019 & 323.2 & 12.0 & 1.01 & 2 & 0.08 & 38.87 & 2 & bg + hd 155555 c & 6 & 8.049 & 105.7 & 9.6 & 1.01 & 2 & 1.10 & 6.61 & 2 & bg + hd 172555 a & 1 & 7.730 & 318.9 & 15.5 & 3.92 & 2 & 0.01 & 12.11 & 2 & bg + tyc 9073 - 762 - 1 & 1 & 1.339 & 105.8 & 13.4 & 1.08 & 2 & 0.20 & 62.61 & 2 & bg + tyc 9073 - 762 - 1 & 2 & 4.142 & 329.0 & 7.1 & 3.96 & 3 & 0.31 & 31.40 & 4 & bg + tyc 9073 - 762 - 1 & 3 & 7.295 & 69.1 & 12.1 & 1.08 & 2 & 1.04 & 3.51 & 2 & bg + pz tel & 1 & 0.330 & 58.7 & 5.4 & 4.82 & 8 & 318.93 & 22.58 & 14 & cpm + pz tel & 2 & 3.870 & 167.6 & 10.0 & 2.99 & 4 & 1.43 & 118.26 & 6 & bg + pz tel & 3 & 5.302 & 37.6 & 14.8 & 3.03 & 2 & 0.06 & 10.23 & 2 & bg + pz tel & 4 & 6.244 & 309.0 & 15.9 & 3.03 & 2 & 0.01 & 5.67 & 2 & bg + pz tel & 5 & 8.446 & 335.2 & 10.2 & 3.03 & 2 & 0.11 & 2.56 & 2 & bg + pz tel & 6 & 8.912 & 160.9 & 11.9 & 3.03 & 2 & 1.75 & 3.63 & 2 & bg + hr 7329 & 1 & 4.175 & 167.6 & 6.6 & 10.78 & 2 & 133.91 & 3.79 & 2 & cpm + hip 95270 & 1 & 4.920 & 254.6 & 13.1 & 3.95 & 2 & 0.08 & 5.08 & 2 & bg + hip 95270 & 2 & 6.040 & 276.6 & 11.8 & 3.95 & 2 & 0.04 & 4.36 & 2 & bg + hip 104308 & 1 & 1.681 & 281.9 & 12.4 & 0.75 & 2 & 0.68 & 27.31 & 2 & bg + hip 104308 & 2 & 2.895 & 264.0 & 15.5 & 0.75 & 2 & 0.05 & 14.08 & 2 & bg + hip 107345 & 1 & 13.393 & 109.5 & 7.1 & 36.24 & 3 & 0.06 & 5.50 & 4 & bg + tx psa & 1 & 5.235 & 54.0 & 8.86 & 1.05 & 2 & 0.50 & 37.83 & 2 & bg + tx psa & 2 & 7.209 & 283.2 & 12.1 & 1.05 & 2 & 0.13 & 137.66 & 2 & bg + cccccccccccccc hip 2729 & 1 & 2011.89 & 9.986 & 0.009 & 30.9 & 0.2 & & & & & & n & bg + & & 2003.56 & 10.174 & 0.024 & 35.3 & 0.5 & & 9.990 & 0.012 & 35.671 & 0.191 & v & bg + & & 2005.78 & 10.044 & 0.288 & 33.6 & 1.0 & & 10.001 & 0.011 & 34.415 & 0.191 & s & bg + hd 12894 & 1 & 2011.84 & 0.322 & 0.009 & 242.9 & 0.2 & & & & & & n & cpm + & & 2008.62 & 0.339 & 0.020 & 259.0 & 1.0 & & 0.251 & 0.006 & 194.735 & 1.455 & v & cpm + & & 2008.85 & 0.331 & 0.020 & 257.2 & 1.0 & & 0.229 & 0.006 & 194.856 & 1.596 & v & cpm + cd -58 553 & 1 & 2011.70 & 6.924 & 0.009 & 85.3 & 0.2 & & & & & & n & bg + & & 2007.69 & 7.265 & 0.040 & 85.2 & 1.0 & & 7.266 & 0.016 & 85.734 & 0.194 & v & bg + & & 2008.85 & 7.186 & 0.040 & 85.2 & 1.0 & & 7.184 & 0.017 & 85.554 & 0.204 & v & bg + & & 2009.04 & 7.176 & 0.009 & 85.7 & 0.2 & & 7.184 & 0.023 & 85.643 & 0.192 & n & bg + hd 19668 & 1 & 2010.66 & 5.470 & 0.009 & 151.0 & 0.2 & & & & & & n & bg + & & 2001.93 & 6.781 & 0.144 & 149.1 & 1.3 & & 6.719 & 0.012 & 148.750 & 0.190 & s & bg + & & 2002.97 & 6.662 & 0.288 & 150.3 & 2.6 & & 6.574 & 0.011 & 148.894 & 0.186 & s & bg + & & 2009.74 & 5.566 & 0.013 & 150.0 & 0.5 & & 5.597 & 0.009 & 150.616 & 0.179 & v & bg + & & 2010.82 & 5.429 & 0.013 & 150.0 & 0.5 & & 5.444 & 0.009 & 150.758 & 0.183 & v & bg + & & 2010.98 & 5.415 & 0.009 & 150.8 & 0.2 & & 5.431 & 0.009 & 150.564 & 0.183 & n & bg + & & 2011.70 & 5.320 & 0.009 & 151.4 & 0.2 & & 5.320 & 0.009 & 151.189 & 0.188 & n & bg + & & 2011.86 & 5.301 & 0.013 & 151.2 & 0.3 & & 5.298 & 0.009 & 150.972 & 0.189 & l & bg + lp 776 - 25 & 1 & 2008.87 & 2.570 & 0.009 & 87.8 & 0.2 & & & & & & n & bg + & & 2005.74 & 2.972 & 0.044 & 102.8 & 2.0 & & 2.964 & 0.018 & 101.486 & 0.359 & g & bg + lp 776 - 25 & 2 & 2008.87 & 6.340 & 0.009 & 264.7 & 0.2 & & & & & & n & bg + & & 2005.74 & 6.146 & 0.044 & 258.2 & 2.0 & & 6.111 & 0.018 & 257.961 & 0.267 & g & bg + lp 776 - 25 & 3 & 2008.87 & 7.010 & 0.009 & 270.1 & 0.2 & & & & & & n & bg + & & 2005.74 & 6.806 & 0.044 & 264.3 & 2.0 & & 6.708 & 0.018 & 264.210 & 0.270 & g & bg + gj 182 & 1 & 2009.10 & 4.870 & 0.009 & 224.7 & 0.2 & & & & & & n & bg + & & 2004.85 & 5.150 & 0.015 & 220.3 & 0.2 & & 5.101 & 0.013 & 220.783 & 0.235 & g & bg + gj 182 & 2 & 2009.10 & 7.210 & 0.009 & 236.9 & 0.2 & & & & & & n & bg + & & 2004.85 & 7.440 & 0.015 & 233.7 & 0.2 & & 7.360 & 0.014 & 233.887 & 0.219 & g & bg + hip 25283 & 1 & 2009.83 & 8.162 & 0.009 & 83.7 & 0.2 & & & & & & n & bg + & & 2010.98 & 8.205 & 0.009 & 83.4 & 0.2 & & 8.171 & 0.009 & 83.157 & 0.191 & n & bg + & & 2011.80 & 8.104 & 0.009 & 83.3 & 0.2 & & 8.082 & 0.009 & 82.880 & 0.193 & n & bg
+ uy pic & 1 & 2009.19 & 0.757 & 0.009 & 138.7 & 0.2 & & & & & & n & bg + & & 2009.92 & 0.698 & 0.009 & 141.9 & 0.2 & & 0.686 & 0.009 & 140.898 & 0.240 & n & bg + & & 2010.27 & 0.767 & 0.009 & 142.7 & 0.2 & & 0.748 & 0.009 & 141.158 & 0.236 & n & bg + bd -13 1328 & 1 & 2009.04 & 2.057 & 0.009 & 314.3 & 0.2 & & & & & & n & bg + & & 2010.01 & 2.123 & 0.009 & 316.0 & 0.2 & & 2.117 & 0.010 & 316.120 & 0.204 & n & bg + cd -35 2722 & 1 & 2009.18 & 3.148 & 0.009 & 243.8 & 0.2 & & & & & & n & cpm + & & 2009.04 & 3.154 & 0.009 & 244.2 & 0.2 & & 3.163 & 0.009 & 244.260 & 0.227 & n & cpm + & & 2010.02 & 3.121 & 0.009 & 243.9 & 0.2 & & 3.137 & 0.010 & 245.183 & 0.238 & n & cpm + & & 2011.80 & 3.090 & 0.009 & 243.1 & 0.2 & & 3.139 & 0.020 & 246.848 & 0.276 & n & cpm + cd -35 2722 & 2 & 2009.18 & 9.202 & 0.009 & 105.5 & 0.2 & & & & & & n & bg + & & 2009.04 & 9.190 & 0.009 & 105.4 & 0.2 & & 9.174 & 0.014 & 105.462 & 0.209 & n & bg + & & 2010.02 & 9.175 & 0.009 & 105.2 & 0.2 & & 9.161 & 0.015 & 105.114 & 0.209 & n & bg + & & 2011.80 & 9.083 & 0.009 & 104.5 & 0.2 & & 9.102 & 0.025 & 104.677 & 0.213 & n & bg + cd -35 2722 & 3 & 2009.04 & 5.289 & 0.009 & 275.4 & 0.2 & & & & & & n & bg + & & 2010.02 & 5.299 & 0.009 & 276.2 & 0.2 & & 5.293 & 0.009 & 276.035 & 0.201 & & bg + ao men & 1 & 2009.10 & 1.501 & 0.009 & 353.4 & 0.2 & & & & & & n & bg + & & 2010.02 & 1.443 & 0.009 & 352.7 & 0.2 & & 1.442 & 0.009 & 353.031 & 0.230 & n & bg + ao men & 2 & 2009.10 & 2.790 & 0.009 & 206.3 & 0.2 & & & & & & n & bg + & & 2010.02 & 2.851 & 0.009 & 205.5 & 0.2 & & 2.845 & 0.008 & 205.802 & 0.199 & n & bg + ao men & 3 & 2009.10 & 4.086 & 0.009 & 307.3 & 0.2 & & & & & & n & bg + & & 2010.02 & 4.053 & 0.009 & 306.3 & 0.2 & & 4.053 & 0.009 & 306.592 & 0.183 & n & bg + ao men & 4 & 2009.10 & 4.600 & 0.009 & 275.3 & 0.2 & & & & & & n & bg + & & 2010.02 & 4.568 & 0.009 & 273.1 & 0.2 & & 4.598 & 0.010 & 274.551 & 0.216 & n & bg + ao men & 5 & 2009.10 & 5.233 & 0.009 & 230.0 & 0.2 & & & & & & n & bg + & & 2010.02 & 5.266 & 0.009 & 229.2 & 0.2 & & 5.274 & 0.011 & 229.542 & 0.167 & n & bg + ao men & 6 & 2009.10 & 5.552 & 0.009 & 248.5 & 0.2 & & & & & & n & bg + & & 2010.02 & 5.578 & 0.009 & 247.5 & 0.2 & & 5.578 & 0.008 & 247.892 & 0.195 & n & bg + ao men & 7 & 2009.10 & 5.768 & 0.009 & 335.8 & 0.2 & & & & & & n & bg + & & 2010.02 & 5.715 & 0.009 & 335.3 & 0.2 & & 5.715 & 0.010 & 335.517 & 0.185 & n & bg + ao men & 8 & 2009.10 & 6.722 & 0.009 & 141.9 & 0.2 & & & & & & n & bg + & & 2010.02 & 6.772 & 0.009 & 142.0 & 0.2 & & 6.767 & 0.010 & 142.238 & 0.175 & n & bg + ao men & 9 & 2009.10 & 8.333 & 0.009 & 151.4 & 0.2 & & & & & & n & bg + & & 2010.02 & 8.407 & 0.009 & 151.4 & 0.2 & & 8.384 & 0.010 & 151.576 & 0.208 & n & bg + ao men & 10 & 2009.10 & 9.763 & 0.009 & 267.7 & 0.2 & & & & & & n & bg + & & 2010.02 & 9.739 & 0.009 & 267.2 & 0.2 & & 9.769 & 0.010 & 267.389 & 0.191 & n & bg + gsc 8894 -0426 & 1 & 2008.96 & 5.232 & 0.009 & 159.8 & 0.2 & & & & & & n & bg + & & 2010.16 & 5.415 & 0.009 & 159.0 & 0.2 & & 5.387 & 0.024 & 159.417 & 0.243 & n & bg + bd + 07 1919 b & 1 & 2009.19 & 0.398 & 0.009 & 108.2 & 0.2 & & & & & & n & cpm + & & 2008.25 & 0.358 & 0.027 & 109.7 & 1.0 & & 0.463 & 0.009 & 123.833 & 0.508 & v & cpm + bd + 07 1919 a & 1 & 2009.19 & 4.667 & 0.009 & 86.0 & 0.2 & & & & & & n & bg + & & 2010.16 & 4.672 & 0.009 & 84.2 & 0.2 & & 4.672 & 0.010 & 84.332 & 0.197 & n & bg + bd + 07 1919 a & 2 & 2009.19 & 7.702 & 0.009 & 151.8 & 0.2 & & & & & & n & bg + & & 2010.16 & 7.573 & 0.009 & 151.3 & 0.2 & & 7.579 & 0.011 & 151.327 & 0.209 & n & bg + hd 70573 & 1 & 2009.10 & 3.490 & 0.009 & 134.0 & 0.2 & & & & & & n & bg + & & 1998.83 & 3.450 & 0.120 & 145.7 & 2.0 & & 3.526 & 0.017 & 146.119 & 0.365 & h & bg + hd 70573 & 2 & 2009.10 & 6.890 & 0.009 & 305.5 & 0.2 & & & & & & n & bg + & & 1998.83 & 7.170 & 0.120 & 298.9 & 1.0 & & 7.079 & 0.019 & 299.614 & 0.202 & h & bg + dx leo & 1 & 2009.20 & 4.190 & 0.009 & 183.6 & 0.2 & & & & & & n & bg + & & 2004.98 & 5.270 & 0.015 & 190.3 & 0.2 & & 5.286 & 0.010 & 190.411 & 0.162 & g & bg + twa 6 & 1 & 2009.04 & 2.090 & 0.009 & 285.8 & 0.2 & & & & & & n & bg + & & 2002.43 & 2.300 & 0.120 & 281.6 & 3.0 & & 2.405 & 0.024 & 279.972 & 0.365 & h & bg
+ twa 7 & 1 & 2009.11 & 3.203 & 0.009 & 121.3 & 0.2 & & & & & & n & bg + & & 2010.16 & 3.316 & 0.009 & 119.5 & 0.2 & & 3.307 & 0.012 & 119.658 & 0.196 & n & bg + twa 7 & 2 & 2009.11 & 3.931 & 0.009 & 92.4 & 0.2 & & & & & & n & bg + & & 2010.16 & 4.072 & 0.009 & 92.0 & 0.2 & & 4.067 & 0.010 & 91.918 & 0.198 & n & bg + twa 7 & 3 & 2009.11 & 4.948 & 0.009 & 172.8 & 0.2 & & & & & & n & bg + & & 2010.16 & 4.953 & 0.009 & 170.9 & 0.2 & & 4.939 & 0.009 & 171.173 & 0.197 & n & bg + twa 14 & 1 & 2009.04 & 2.386 & 0.009 & 261.3 & 0.2 & & & & & & n & bg + & & 2010.02 & 2.346 & 0.009 & 261.1 & 0.2 & & 2.343 & 0.010 & 261.304 & 0.228 & n & bg + & & 2012.21 & 2.253 & 0.009 & 260.8 & 0.2 & & 2.237 & 0.016 & 261.245 & 0.304 & n & bg + hd 92945 & 1 & 2009.04 & 8.961 & 0.009 & 234.7 & 0.2 & & & & & & n & bg + & & 2010.27 & 8.691 & 0.009 & 233.9 & 0.2 & & 8.673 & 0.008 & 233.751 & 0.181 & n & bg + & & 2011.25 & 8.512 & 0.009 & 233.2 & 0.2 & & 8.476 & 0.008 & 233.217 & 0.185 & n & bg + & & 2012.26 & 8.298 & 0.009 & 232.5 & 0.2 & & 8.271 & 0.008 & 232.569 & 0.189 & n & bg + twa 8a & 1 & 2009.10 & 10.550 & 0.009 & 37.4 & 0.2 & & & & & & n & bg + & & 1998.52 & 9.660 & 0.100 & 34.1 & 2.0 & & 9.666 & 0.055 & 33.866 & 0.387 & h & bg
+ twa 9 a & 1 & 2010.98 & 7.286 & 0.009 & 123.1 & 0.2 & & & & & & n & bg + & & 2009.18 & 7.203 & 0.009 & 123.7 & 0.2 & & 7.240 & 0.009 & 123.653 & 0.205 & n & bg + twa 9 a & 2 & 2010.98 & 8.260 & 0.009 & 92.2 & 0.2 & & & & & & n & bg + & & 2009.18 & 8.151 & 0.009 & 92.3 & 0.2 & & 8.183 & 0.011 & 92.504 & 0.188 & n & bg + twa 25 & 1 & 2009.18 & 3.499 & 0.009 & 116.9 & 0.2 & & & & & & n & bg + & & 2011.31 & 3.640 & 0.009 & 114.6 & 0.2 & & 3.636 & 0.013 & 114.907 & 0.219 & n & bg + twa 25 & 2 & 2009.18 & 8.615 & 0.009 & 10.0 & 0.2 & & & & & & n & bg + & & 2011.31 & 8.693 & 0.009 & 10.8 & 0.2 & & 8.697 & 0.011 & 11.044 & 0.217 & n & bg + twa 25 & 3 & 2009.18 & 8.638 & 0.009 & 60.9 & 0.2 & & & & & & n & bg + & & 2011.31 & 8.830 & 0.009 & 61.1 & 0.2 & & 8.817 & 0.012 & 61.199 & 0.197 & n & bg + twa 25 & 4 & 2009.18 & 8.943 & 0.009 & 344.5 & 0.2 & & & & & & n & bg + & & 2011.31 & 8.930 & 0.009 & 345.6 & 0.2 & & 8.946 & 0.012 & 345.654 & 0.208 & n & bg + twa 20 & 1 & 2009.10 & 2.903 & 0.009 & 190.5 & 0.2 & & & & & & n & bg + & & 2011.36 & 2.840 & 0.009 & 187.9 & 0.2 & & 2.850 & 0.015 & 187.921 & 0.329 & n & bg + twa 20 & 2 & 2009.10 & 6.694 & 0.009 & 10.0 & 0.2 & & & & & & n & bg + & & 2011.36 & 6.781 & 0.009 & 10.7 & 0.2 & & 6.750 & 0.014 & 11.128 & 0.235 & n & bg +
twa 10 & 1 & 2009.04 & 9.168 & 0.009 & 313.5 & 0.2 & & & & & & n & bg + & & 2012.27 & 9.102 & 0.009 & 314.5 & 0.2 & & 9.061 & 0.017 & 315.137 & 0.251 & n & bg + hr 4796 a & 1 & 2009.04 & 4.472 & 0.009 & 322.7 & 0.2 & & & & & & n & bg + & & 1994.20 & 4.700 & 0.050 & 311.0 & 1.0 & & 4.767 & 0.011 & 311.974 & 0.217 & e & bg + & & 2012.27 & 4.422 & 0.009 & 324.9 & 0.2 & & 4.423 & 0.011 & 325.341 & 0.232 & n & bg + hd 139664 & 1 & 2009.10 & 1.940 & 0.009 & 48.0 & 0.2 & & & & & & n & bg + & & 2010.35 & 2.346 & 0.009 & 46.0 & 0.2 & & 2.368 & 0.009 & 45.818 & 0.162 & n & bg + hd 139664 & 2 & 2009.10 & 4.550 & 0.009 & 303.9 & 0.2 & & & & & & n & bg + & & 2010.35 & 4.560 & 0.009 & 309.1 & 0.2 & & 4.554 & 0.010 & 309.390 & 0.194 & n & bg + hd 139664 & 3 & 2009.10 & 5.620 & 0.009 & 95.0 & 0.2 & & & & & & n & bg + & & 2010.35 & 5.860 & 0.009 & 91.2 & 0.2 & & 5.857 & 0.009 & 91.343 & 0.188 & n & bg + hd 139664 & 4 & 2009.10 & 5.640 & 0.009 & 267.8 & 0.2 & & & & & & n & bg + & & 2010.35 & 5.426 & 0.009 & 271.7 & 0.2 & & 5.381 & 0.009 & 271.448 & 0.198 & n & bg + hd 139664 & 5 & 2009.10 & 6.240 & 0.009 & 115.8 & 0.2 & & & & & & n & bg + & & 2010.35 & 6.319 & 0.009 & 111.8 & 0.2 & & 6.332 & 0.010 & 111.915 & 0.192 & n & bg + hd 139664 & 6 & 2009.10 & 6.770 & 0.009 & 2.4 & 0.2 & & & & & & n & bg + & & 2010.35 & 7.140 & 0.009 & 4.3 & 0.2 & & 7.137 & 0.010 & 4.341 & 0.185 & n & bg + hip 81084 & 1 & 2009.27 & 6.190 & 0.009 & 238.6 & 0.2 & & & & & & n & bg + & & 2005.30 & 6.840 & 0.015 & 234.6 & 0.2 & & 6.190 & 0.009 & 238.600 & 0.189 & g & bg + hip 82688 & 1 & 2002.66 & 3.819 & 0.008 & 56.9 & 0.2 & & & & & & p & cpm + & & 2005.56 & 3.830 & 0.050 & 57.5 & 1.0 & & 4.089 & 0.008 & 53.803 & 0.144 & h & cpm + & & 2008.40 & 3.819 & 0.027 & 57.2 & 0.3 & & 4.355 & 0.009 & 51.033 & 0.143 & v & cpm + & & 2010.27 & 3.791 & 0.009 & 58.8 & 0.2 & & 4.538 & 0.010 & 49.339 & 0.147 & n & cpm + & & 2011.21 & 3.786 & 0.009 & 57.9 & 0.2 & & 4.634 & 0.011 & 48.566 & 0.149 & n & cpm + hip 82688 & 2 & 2010.27 & 2.386 & 0.009 & 316.5 & 0.2 & & & & & & n & bg + & & 2011.21 & 2.423 & 0.009 & 317.9 & 0.2 & & 2.446 & 0.009 & 318.859 & 0.193 & & bg + hip 82688 & 3 & 2010.27 & 7.811 & 0.009 & 283.2 & 0.2 & & & & & & n & bg + & & 2008.40 & 7.860 & 0.030 & 279.9 & 0.5 & & 7.819 & 0.009 & 281.564 & 0.204 & v & bg + hd 155555 c & 1 & 2009.27 & 2.668 & 0.009 & 164.3 & 0.2 & & & & & & n & bg + & & 2010.27 & 2.547 & 0.009 & 163.6 & 0.2 & & 2.542 & 0.010 & 163.015 & 0.229 & n & bg + hd 155555 c & 2 & 2009.27 & 4.735 & 0.009 & 336.3 & 0.2 & & & & & & n & bg + & & 2010.27 & 4.859 & 0.009 & 337.2 & 0.2 & & 4.853 & 0.009 & 337.196 & 0.212 & n & bg + hd 155555 c & 3 & 2009.27 & 5.634 & 0.009 & 37.3 & 0.2 & & & & & & n & bg + & & 2010.27 & 5.754 & 0.009 & 36.7 & 0.2 & & 5.758 & 0.008 & 36.639 & 0.193 & n & bg + hd 155555 c & 4 & 2009.27 & 6.371 & 0.009 & 292.4 & 0.2 & & & & & & n & bg + & & 2010.27 & 6.418 & 0.009 & 293.7 & 0.2 & & 6.404 & 0.010 & 293.605 & 0.207 & n & bg + hd 155555 c & 5 & 2009.27 & 7.019 & 0.009 & 323.2 & 0.2 & & & & & & n & bg + & & 2010.27 & 7.124 & 0.009 & 324.1 & 0.2 & & 7.117 & 0.010 & 324.036 & 0.209 & n & bg + hd 155555 c & 6 & 2009.27 & 8.049 & 0.009 & 105.7 & 0.2 & & & & & & n & bg + & & 2010.27 & 8.013 & 0.009 & 105.0 & 0.2 & & 8.034 & 0.009 & 104.667 & 0.192 & n & bg + hd 172555 a & 1 & 2009.27 & 7.730 & 0.009 & 318.9 & 0.2 & & & & & & n & bg + & & 2005.35 & 7.210 & 0.120 & 316.5 & 1.0 & & 7.215 & 0.009 & 316.704 & 0.205 & h & bg + tyc 9073 - 762 - 1 & 1 & 2010.35 & 1.339 & 0.009 & 105.8 & 0.2 & & & & & & n & bg + & & 2009.27 & 1.378 & 0.009 & 108.8 & 0.2 & & 1.381 & 0.010 & 109.138 & 0.309 & n & bg + tyc
9073 - 762 - 1 & 2 & 2010.35 & 4.142 & 0.009 & 329.0 & 0.2 & & & & & & n & bg + & & 2006.39 & 3.883 & 0.027 & 327.6 & 0.4 & & 3.846 & 0.017 & 327.630 & 0.308 & v & bg + & & 2009.27 & 4.070 & 0.009 & 328.7 & 0.2 & & 4.058 & 0.009 & 328.538 & 0.210 & n & bg + tyc
9073 - 762 - 1 & 3 & 2010.35 & 7.295 & 0.009 & 69.1 & 0.2 & & & & & & n & bg + & & 2009.27 & 7.308 & 0.009 & 69.8 & 0.2 & & 7.279 & 0.011 & 69.763 & 0.240 & n & bg + pz tel & 1 & 2009.27 & 0.330 & 0.009 & 58.7 & 0.2 & & & & & & n & cpm + & & 2007.45 & 0.255 & 0.002 & 61.7 & 0.6 & & 0.328 & 0.009 & 85.328 & 0.822 & v & cpm + & & 2009.74 & 0.337 & 0.002 & 60.5 & 0.2 & & 0.378 & 0.010 & 55.649 & 0.243 & v & cpm + & & 2010.34 & 0.356 & 0.001 & 60.5 & 0.2 & & 0.375 & 0.010 & 45.177 & 0.407 & v & cpm + & & 2010.35 & 0.355 & 0.001 & 60.4 & 0.2 & & 0.375 & 0.010 & 45.122 & 0.408 & v & cpm + & & 2010.36 & 0.360 & 0.009 & 59.9 & 0.2 & & 0.377 & 0.010 & 45.010 & 0.409 & n & cpm + & & 2011.31 & 0.373 & 0.009 & 58.7 & 0.2 & & 0.423 & 0.009 & 35.718 & 0.584 & n & cpm + & & 2012.26 & 0.397 & 0.009 & 60.4 & 0.2 & & 0.479 & 0.009 & 28.577 & 0.684 & n &
cpm + pz tel & 2 & 2009.27 & 3.870 & 0.009 & 167.6 & 0.2 & & & & & & n & bg + & & 2010.36 & 3.780 & 0.009 & 168.1 & 0.2 & & 3.774 & 0.009 & 167.516 & 0.199 & n & bg + & & 2011.31 & 3.680 & 0.009 & 166.7 & 0.2 & & 3.694 & 0.010 & 167.555 & 0.206 & n & bg + & & 2012.26 & 3.609 & 0.009 & 167.1 & 0.2 & & 3.615 & 0.010 & 167.570 & 0.215 & n & bg + pz tel & 3 & 2009.27 & 5.302 & 0.009 & 37.6 & 0.2 & & & & & & n & bg + & & 2006.24 & 5.170 & 0.100 & 39.9 & 0.5 & & 5.136 & 0.008 & 39.777 & 0.242 & & bg +
pz tel & 4 & 2009.27 & 6.244 & 0.009 & 309.0 & 0.2 & & & & & & n & bg + & & 2006.24 & 6.040 & 0.100 & 307.6 & 0.5 & & 6.044 & 0.010 & 307.487 & 0.217 & & bg +
pz tel & 5 & 2009.27 & 8.446 & 0.009 & 335.2 & 0.2 & & & & & & n & bg + & & 2006.24 & 8.220 & 0.100 & 335.1 & 0.5 & & 8.192 & 0.009 & 334.825 & 0.199 & a & bg + pz tel & 6 & 2009.27 & 8.912 & 0.009 & 160.9 & 0.2 & & & & & & n & bg + & & 2006.24 & 9.100 & 0.100 & 159.9 & 0.5 & & 9.171 & 0.009 & 161.149 & 0.212 & a & bg + hr 7329 & 1 & 2009.27 & 4.175 & 0.009 & 167.6 & 0.2 & & & & & & n & cpm + & & 1998.49 & 4.170 & 0.050 & 166.8 & 0.2 & & 5.128 & 0.009 & 166.582 & 0.139 & h & cpm + hip 95270 & 1 & 2009.28 & 4.920 & 0.009 & 254.6 & 0.2 & & & & & & n & bg + & & 2005.33 & 4.940 & 0.120 & 250.1 & 1.4 & & 4.923 & 0.010 & 250.715 & 0.196 & h & bg + hip 95270 & 2 & 2009.28 & 6.040 & 0.009 & 276.6 & 0.2 & & & & & & n & bg + & & 2005.33 & 5.900 & 0.120 & 273.3 & 1.2 & & 5.916 & 0.009 & 273.632 & 0.173 & h & bg + hip 104308 & 1 & 2011.82 & 1.681 & 0.009 & 281.9 & 0.2 & & & & & & n & bg + & & 2012.57 & 1.745 & 0.009 & 283.4 & 0.2 & & 1.729 & 0.009 & 283.708 & 0.203 & n & bg + hip 104308 & 2 & 2011.82 & 2.895 & 0.009 & 264.0 & 0.2 & & & & & & n & bg + & & 2012.57 & 2.929 & 0.009 & 265.3 & 0.2 & & 2.924 & 0.008 & 265.279 & 0.212 & n & bg + hip 107345 & 1 & 2011.84 & 13.393 & 0.009 & 109.5 & 0.2 & & & & & & n & bg + & & 1975.60 & 16.172 & 1.700 & 120.6 & 3.0 & & 16.064 & 0.078 & 119.020 & 0.275 & d & bg + & & 2008.61 & 13.606 & 0.144 & 110.5 & 0.5 & & 13.592 & 0.013 & 110.533 & 0.221 & s & bg + tx psa & 1 & 2010.65 & 5.235 & 0.009 & 54.0 & 0.2 & & & & & & n & bg + & & 2011.70 & 5.166 & 0.009 & 52.1 & 0.2 & & 5.170 & 0.011 & 51.708 & 0.229 & n & bg + tx psa & 2 & 2010.65 & 7.209 & 0.009 & 283.2 & 0.2 & & & & & & n & bg + & & 2011.70 & 7.419 & 0.009 & 283.7 & 0.2 & & 7.415 & 0.009 & 283.845 & 0.174 & n & bg + lrccccc bd+07 1919 & 2008 - 04 - 02 & vlt naco & @xmath39 & direct imaging & 450 & + & 2008 - 04 - 02 & vlt naco & @xmath2 & direct imaging & 450 & + & 2008 - 04 - 02 & vlt naco & @xmath13s & direct imaging & 150 & + & 2009 - 03 - 12 & gemini nici & @xmath91 & asdi & 2701 & b saturated + & & & & & & except in short exps + & 2009 - 03 - 12 & gemini nici & @xmath2 & adi & 1208 & b saturated + & & & & & & in all exposures + & 2010 - 02 - 28 & gemini nici & @xmath2 & adi & 1389 & b saturated + & & & & & & in all exposures + hd 12894 & 2008 - 08 - 14 & vlt naco & @xmath15 & direct imaging & 1230 & + & 2008 - 11 - 07 & vlt naco & @xmath13s & direct imaging & 480 & + & 2011 - 11 - 03 & gemini nici & @xmath2 & adi & 2372 & b saturated + & & & & & & except in short exps + & & & & & & a under mask + [ tab : binarydata ] lcc parallax ( mas ) & + age ( myr ) & + proper motion ( @xmath92 , @xmath93 ) ( mas / yr ) & + separation : 2008.62 , vlt naco & + position angle : 2008.62 , vlt naco & + separation : 2008.85 , vlt naco & + position angle : 2008.85 , vlt naco & + separation : 2011.84 , gemini nici & + position angle : 2011.84 , gemini nici & + @xmath94(mag ) & & [email protected] + @xmath95(mag ) & & [email protected] + @xmath96 ( mag ) & [email protected] & [email protected] + @xmath97 ( mag ) & [email protected] & [email protected] + spectral type & f2 & + estimated mass ( @xmath50 ) & [email protected] & [email protected] + estimated semi - major axis ( au ) & + estimated period ( yr ) & + lccc parallax ( mas ) & + age ( myr ) & + proper motion ( @xmath92 , @xmath93 ) ( mas / yr ) & + bc separation : 2008.25 , vlt naco & + bc position angle : 2008.25 , vlt naco & + bc separation : 2009.19 , gemini nici & + bc position angle : 2009.19 , gemini nici & + @xmath98(mag ) & & & [email protected] + @xmath99(mag ) & & & [email protected] + @xmath94(mag ) & & & [email protected] + @xmath39 ( mag ) & [email protected] & [email protected] & [email protected] + @xmath2 ( mag ) & [email protected] & [email protected] & [email protected] + @xmath96 ( mag ) & [email protected] & [email protected] & [email protected] + @xmath100 ( mag ) & [email protected] & [email protected] & [email protected] + @xmath101 ( mag ) & [email protected] & [email protected] & [email protected] + @xmath97 ( mag ) & [email protected] & [email protected] & [email protected] + estimated mass & [email protected] & [email protected] & [email protected] m@xmath7 + estimated semi - major axis ( bc ) ( au ) & + estimated period ( bc ) ( yr ) & + lcc @xmath98(mag ) & & [email protected] + @xmath94(mag ) & & [email protected] + @xmath39 ( mag ) & [email protected] & [email protected] + @xmath96 ( mag ) & [email protected] & [email protected] + @xmath39 - @xmath96 ( mag ) & [email protected] & [email protected] + @xmath100 ( mag ) & [email protected] & [email protected] + @xmath97 ( mag ) & [email protected] & [email protected] + lccc bd + 7 1919a & 211 & k5 & @xcite + bd + 7 1919b & 14 & @xmath100.2 m@xmath7 & discovered in this work + hd 12894 & 15.5 & @xmath100.5 m@xmath7 & discovered in this work + pz tel & 16.4 & @xmath1040 m@xmath6 & @xcite + cd-35 2722 & 67 & @xmath1030 m@xmath6 & @xcite + hr 7329 & 200 & m7-m8 & @xcite
+ ab pic & 250 & l0.5 & @xcite + tyc 7443 - 1102 - 1 & 1520 & m0 & @xcite + cd-53 544 & 936 & k6 & companion is af hor , @xcite + af hor & 936 & m2 & companion is cd -53 544 , @xcite + gj 2036 a & 129 & m3 & companion is gj 2036 b , @xcite + gj 2036 b & 129 & m3 & companion is gj 2036 a , @xcite + hip 26369 & 621 & k6 & companion is uy pic , @xcite + uy pic & 621 & k0 & companion is hip 26369 , @xcite + hd 45270 & 376 & & @xcite + dx leo & 1150 & m4.5 & @xcite + hip 82688 & 181 & 0.31 m@xmath7 & @xcite + hr 4796 a & 560 , 2900 & m2.5 , & @xcite + hr 4796 b & 560 , 2900 & a0 , & @xcite + gj 560 a & 260 & k5 & @xcite + twa 8a & 611 & m5 & @xcite + twa 9a & 271 & m1 & comp . is twa 9a , @xcite + twa 9b & 271 & k5 & comp . is twa 9b , @xcite + hip 10679 & 375 & f5 & @xcite + bd-21 1074a & 398 & m3 & @xcite + v343 nor & 411 & m4 & @xcite + hd 164249b & 309 & f6 & @xcite + hd 172555a & 2052 & k5 & @xcite + gj 799a & 39 & m4 & @xcite + hip 112312a & 839 & m5 & companion to tx psa , @xcite + tx psa & 839 & m4 & companion to hip 112312a , @xcite + [ tab : binaries ] ccccc 1 & & & 67 290 au & 24 610 au + 2 & & 49 290 au & 22 550 au & 12 880 au + 4 & & 25 380 au & 14 640 au & 9.3 1100 au + 10 & & 19 390 au & 12 660 au & 7.4 1100 au + ccc + dusty & @xmath3018@xmath9 & @xmath102 + cond & @xmath306@xmath9 & @xmath103 + + dusty & @xmath3021@xmath9 & @xmath104 + cond & @xmath105 & @xmath106 + ) of stars in our sample have ages less than 100 myr and distances less than 60 pc .
we observed 14 stars from the tw hya association , 30 stars from the @xmath0 pic moving group , 12 stars from the tucana - horologium association , 19 stars from the ab dor moving group .
4 stars from the hercules - lyra association , and 1 star ( bd + 1 2447 ) which is either a hercules - lyra or ab dor moving group member .
[ fig : mg_sample ] , width=576 ] of our sample have spectral types of k or later ; thus , our sample is comprised largely of lower mass stars .
[ fig : mg_hists],title="fig:",width=307 ] of our sample have spectral types of k or later ; thus , our sample is comprised largely of lower mass stars .
[ fig : mg_hists],title="fig:",width=307 ] .
top : asdi contrasts .
bottom : adi contrasts.,title="fig:",width=576 ] .
top : asdi contrasts .
bottom : adi contrasts.,title="fig:",width=576 ] pic moving group stars ( 1st half ) , with plot range from 0.3 ( coronagraph inner working angle ) to 6 .
top : asdi contrasts .
bottom : adi contrasts.,title="fig:",width=576 ] pic moving group stars ( 1st half ) , with plot range from 0.3 ( coronagraph inner working angle ) to 6 .
top : asdi contrasts .
bottom : adi contrasts.,title="fig:",width=576 ] moving group ( 2nd half ) , with plot range from 0.3 ( coronagraph inner working angle ) to 6 .
top : asdi contrasts .
bottom : adi contrasts.,title="fig:",width=576 ] moving group ( 2nd half ) , with plot range from 0.3 ( coronagraph inner working angle ) to 6 .
top : asdi contrasts .
bottom : adi contrasts.,title="fig:",width=576 ] , 1@xmath22 , 2@xmath22 , and 4@xmath22 , using the dusty models of @xcite to convert from contrasts to masses . at 0.5@xmath22
we are sensitive to companions of @xmath3013 m@xmath6 for all but one star . at 2@xmath22
we are sensitive to companions with masses @xmath810 m@xmath6 for all stars .
the minimum detectable mass varies by star ( according to spectral type , magnitude , distance , etc . )
but we are generally sensitive to @xmath185 m@xmath6 companions at 2@xmath22 . around all sample stars .
[ fig : minmass_plots_dusty],title="fig:",width=307 ] , 1@xmath22 , 2@xmath22 , and 4@xmath22 , using the dusty models of @xcite to convert from contrasts to masses . at 0.5@xmath22
we are sensitive to companions of @xmath3013 m@xmath6 for all but one star . at 2@xmath22
we are sensitive to companions with masses @xmath810 m@xmath6 for all stars .
the minimum detectable mass varies by star ( according to spectral type , magnitude , distance , etc . )
but we are generally sensitive to @xmath185 m@xmath6 companions at 2@xmath22 . around all sample stars .
[ fig : minmass_plots_dusty],title="fig:",width=307 ] , 1@xmath22 , 2@xmath22 , and 4@xmath22 , using the dusty models of @xcite to convert from contrasts to masses . at 0.5@xmath22
we are sensitive to companions of @xmath3013 m@xmath6 for all but one star . at 2@xmath22
we are sensitive to companions with masses @xmath810 m@xmath6 for all stars .
the minimum detectable mass varies by star ( according to spectral type , magnitude , distance , etc . )
but we are generally sensitive to @xmath185 m@xmath6 companions at 2@xmath22 . around all sample stars .
[ fig : minmass_plots_dusty],title="fig:",width=307 ] , 1@xmath22 , 2@xmath22 , and 4@xmath22 , using the dusty models of @xcite to convert from contrasts to masses . at 0.5@xmath22
we are sensitive to companions of @xmath3013 m@xmath6 for all but one star . at 2@xmath22
we are sensitive to companions with masses @xmath810 m@xmath6 for all stars .
the minimum detectable mass varies by star ( according to spectral type , magnitude , distance , etc . )
but we are generally sensitive to @xmath185 m@xmath6 companions at 2@xmath22 . around all sample stars .
[ fig : minmass_plots_dusty],title="fig:",width=307 ] , 1@xmath22 , 2@xmath22 , and 4@xmath22 , using the cond models of @xcite to convert from contrasts to masses .
at 0.5@xmath22 we are sensitive to companions of @xmath3010 m@xmath6 for all but one star . at 2@xmath22
we are sensitive to companions with masses @xmath85 m@xmath6 for all stars .
the minimum detectable mass varies by star ( according to spectral type , magnitude , distance , etc . )
but we are generally sensitive to @xmath185 m@xmath6 companions at 2@xmath22 around all sample stars .
[ fig : minmass_plots_cond],title="fig:",width=307 ] , 1@xmath22 , 2@xmath22 , and 4@xmath22 , using the cond models of @xcite to convert from contrasts to masses .
at 0.5@xmath22 we are sensitive to companions of @xmath3010 m@xmath6 for all but one star . at 2@xmath22
we are sensitive to companions with masses @xmath85 m@xmath6 for all stars .
the minimum detectable mass varies by star ( according to spectral type , magnitude , distance , etc . )
but we are generally sensitive to @xmath185 m@xmath6 companions at 2@xmath22 around all sample stars .
[ fig : minmass_plots_cond],title="fig:",width=307 ] , 1@xmath22 , 2@xmath22 , and 4@xmath22 , using the cond models of @xcite to convert from contrasts to masses .
at 0.5@xmath22 we are sensitive to companions of @xmath3010 m@xmath6 for all but one star . at 2@xmath22
we are sensitive to companions with masses @xmath85 m@xmath6 for all stars .
the minimum detectable mass varies by star ( according to spectral type , magnitude , distance , etc . )
but we are generally sensitive to @xmath185 m@xmath6 companions at 2@xmath22 around all sample stars .
[ fig : minmass_plots_cond],title="fig:",width=307 ] , 1@xmath22 , 2@xmath22 , and 4@xmath22 , using the cond models of @xcite to convert from contrasts to masses .
at 0.5@xmath22 we are sensitive to companions of @xmath3010 m@xmath6 for all but one star . at 2@xmath22
we are sensitive to companions with masses @xmath85 m@xmath6 for all stars .
the minimum detectable mass varies by star ( according to spectral type , magnitude , distance , etc . )
but we are generally sensitive to @xmath185 m@xmath6 companions at 2@xmath22 around all sample stars .
[ fig : minmass_plots_cond],title="fig:",width=307 ] , title="fig:",width=192 ] -0.3 in , title="fig:",width=192 ] -0.3 in , title="fig:",width=192 ] -0.3 in , title="fig:",width=192 ] -0.2 in , title="fig:",width=192 ] -0.3 in , title="fig:",width=192 ] -0.3 in , title="fig:",width=192 ] -0.3 in , title="fig:",width=192 ] -0.2 in , title="fig:",width=192 ] -0.3 in , title="fig:",width=192 ] -0.3 in , title="fig:",width=192 ] -0.3 in , title="fig:",width=192 ] -0.2 in , title="fig:",width=192 ] -0.3 in , title="fig:",width=192 ] -0.3 in , title="fig:",width=192 ] -0.3 in , title="fig:",width=192 ] pic mg objects .
[ fig : betapic_skyplots1],title="fig:",width=192 ] -0.3 in pic mg objects .
[ fig : betapic_skyplots1],title="fig:",width=192 ] -0.3 in pic mg objects .
[ fig : betapic_skyplots1],title="fig:",width=192 ] -0.3 in pic mg objects .
[ fig : betapic_skyplots1],title="fig:",width=192 ] -0.2 in pic mg objects .
[ fig : betapic_skyplots1],title="fig:",width=192 ] -0.3 in pic mg objects .
[ fig : betapic_skyplots1],title="fig:",width=192 ] -0.3 in pic mg objects .
[ fig : betapic_skyplots1],title="fig:",width=192 ] -0.3 in pic mg objects .
[ fig : betapic_skyplots1],title="fig:",width=192 ] -0.2 in pic mg objects .
[ fig : betapic_skyplots1],title="fig:",width=192 ] -0.3 in pic mg objects .
[ fig : betapic_skyplots1],title="fig:",width=192 ] -0.3 in pic mg objects .
[ fig : betapic_skyplots1],title="fig:",width=192 ] -0.3 in pic mg objects .
[ fig : betapic_skyplots1],title="fig:",width=192 ] -0.2 in pic mg objects .
[ fig : betapic_skyplots1],title="fig:",width=192 ] -0.3 in pic mg objects .
[ fig : betapic_skyplots1],title="fig:",width=192 ] -0.3 in pic mg objects .
[ fig : betapic_skyplots1],title="fig:",width=192 ] -0.3 in pic mg objects .
[ fig : betapic_skyplots1],title="fig:",width=192 ] -0.2 in on - sky plots for @xmath0 pic mg objects ( continued).,title="fig:",width=192 ] -0.3 in on - sky plots for @xmath0 pic mg objects ( continued).,title="fig:",width=192 ] -0.3 in on - sky plots for @xmath0 pic mg objects ( continued).,title="fig:",width=192 ] -0.3 in on - sky plots for @xmath0 pic mg objects ( continued).,title="fig:",width=192 ] -0.2 in on - sky plots for @xmath0 pic mg objects ( continued).,title="fig:",width=192 ] -0.3 in on - sky plots for @xmath0 pic mg objects ( continued).,title="fig:",width=192 ] -0.3 in on - sky plots for @xmath0 pic mg objects ( continued).,title="fig:",width=192 ] -0.3 in on - sky plots for @xmath0 pic mg objects ( continued).,title="fig:",width=192 ] -0.2 in on - sky plots for @xmath0 pic mg objects ( continued).,title="fig:",width=192 ] -0.3 in on - sky plots for @xmath0 pic mg objects ( continued).,title="fig:",width=192 ] -0.3 in on - sky plots for @xmath0 pic mg objects ( continued).,title="fig:",width=192 ] -0.3 in on - sky plots for @xmath0 pic mg objects ( continued).,title="fig:",width=192 ] -0.2 in on - sky plots for @xmath0 pic mg objects ( continued).,title="fig:",width=192 ] -0.3 in on - sky plots for @xmath0 pic mg objects ( continued).,title="fig:",width=192 ] -0.3 in on - sky plots for @xmath0 pic mg objects ( continued).,title="fig:",width=192 ] -0.3 in on - sky plots for @xmath0 pic mg objects ( continued).,title="fig:",width=192 ] -0.2 in on - sky plots for @xmath0 pic mg objects ( continued).,title="fig:",width=192 ] -0.3 in on - sky plots for tucana - horologium and hercules - lyra association objects .
, title="fig:",width=192 ] -0.3 in on - sky plots for tucana - horologium and hercules - lyra association objects .
, title="fig:",width=192 ] -0.3 in on - sky plots for tucana - horologium and hercules - lyra association objects .
, title="fig:",width=192 ] -0.3 in on - sky plots for tucana - horologium and hercules - lyra association objects . ,
title="fig:",width=192 ] -0.2 in on - sky plots for tucana - horologium and hercules - lyra association objects .
, title="fig:",width=192 ] -0.3 in on - sky plots for tucana - horologium and hercules - lyra association objects .
, title="fig:",width=192 ] -0.3 in on - sky plots for tucana - horologium and hercules - lyra association objects .
, title="fig:",width=192 ] -0.3 in on - sky plots for tucana - horologium and hercules - lyra association objects . ,
title="fig:",width=192 ] -0.2 in on - sky plots for tucana - horologium and hercules - lyra association objects .
, title="fig:",width=192 ] -0.3 in on - sky plots for tucana - horologium and hercules - lyra association objects .
, title="fig:",width=192 ] -0.3 in on - sky plots for tucana - horologium and hercules - lyra association objects .
, title="fig:",width=192 ] -0.3 in on - sky plots for tucana - horologium and hercules - lyra association objects . ,
title="fig:",width=192 ] -0.2 in on - sky plots for tucana - horologium and hercules - lyra association objects .
, title="fig:",width=192 ] -0.3 in on - sky plots for tucana - horologium and hercules - lyra association objects .
, title="fig:",width=192 ] -0.3 in on - sky plots for tucana - horologium and hercules - lyra association objects .
, title="fig:",width=192 ] -0.3 in -0.2 in , title="fig:",width=192 ] -0.3 in , title="fig:",width=192 ] -0.3 in , title="fig:",width=192 ] -0.3 in , title="fig:",width=192 ] -0.2 in , title="fig:",width=192 ] -0.3 in , title="fig:",width=192 ] -0.3 in , title="fig:",width=192 ] -0.3 in , title="fig:",width=192 ] -0.2 in , title="fig:",width=192 ] -0.3 in , title="fig:",width=192 ] -0.3 in , title="fig:",width=192 ] -0.3 in , title="fig:",width=192 ] -0.2 in , title="fig:",width=192 ] -0.3 in , title="fig:",width=192 ] -0.3 in , title="fig:",width=192 ] -0.3 in , title="fig:",width=192 ] -0.2 in , title="fig:",width=192 ] -0.3 in , title="fig:",width=192 ] -0.3 in , title="fig:",width=192 ] -0.3 in -0.2 in 0.08 m@xmath7 ) and bd + 07 1919c ( [email protected] m@xmath7 ) .
north is up and east is left in all images .
[ fig : binaryimages],title="fig:",width=192 ] 0.08 m@xmath7 ) and bd + 07 1919c ( [email protected] m@xmath7 ) .
north is up and east is left in all images .
[ fig : binaryimages],title="fig:",width=192 ] 0.08 m@xmath7 ) and bd + 07 1919c ( [email protected] m@xmath7 ) .
north is up and east is left in all images .
[ fig : binaryimages],title="fig:",width=192 ] 0.08 m@xmath7 ) and bd + 07 1919c ( [email protected] m@xmath7 ) .
north is up and east is left in all images .
[ fig : binaryimages],title="fig:",width=192 ] 0.2 in 0.08 m@xmath7 ) and bd + 07 1919c ( [email protected] m@xmath7 ) .
north is up and east is left in all images .
[ fig : binaryimages],title="fig:",width=192 ] 0.08 m@xmath7 ) and bd + 07 1919c ( [email protected] m@xmath7 ) .
north is up and east is left in all images .
[ fig : binaryimages],title="fig:",width=192 ] 0.08 m@xmath7 ) and bd + 07 1919c ( [email protected] m@xmath7 ) .
north is up and east is left in all images .
[ fig : binaryimages],title="fig:",width=192 ] , title="fig:",width=288 ] , title="fig:",width=288 ] of stars to possess a 2 m@xmath6 planet at separations of 49 to 290 au , at a 95@xmath9 confidence level . note that this analysis does not assume a particular distribution of planets as a function of mass and semi - major axis .
[ fig : mg_ct ] , width=576 ] [ cols="^,^ " , ] -0.2 in pic b detection , * ( center ) * only the ab pic b detection , and * ( right ) * both detections .
posterior probability is plotted in logarithmic units . for ab pic b , in the case of small ( @xmath850 au ) cutoffs , ln(pdf ) trended to negative infinity and is not plotted here .
[ fig:1d_det ] , title="fig:",width=240 ] -0.2 in pic b detection , * ( center ) * only the ab pic b detection , and * ( right ) * both detections .
posterior probability is plotted in logarithmic units . for ab pic b , in the case of small ( @xmath850 au ) cutoffs , ln(pdf ) trended to negative infinity and is not plotted here .
[ fig:1d_det ] , title="fig:",width=240 ] -0.2 in pic b detection , * ( center ) * only the ab pic b detection , and * ( right ) * both detections .
posterior probability is plotted in logarithmic units . for ab pic b , in the case of small ( @xmath850 au ) cutoffs , ln(pdf ) trended to negative infinity and is not plotted here .
[ fig:1d_det ] , title="fig:",width=240 ] -0.2 in pic b detection , * ( center ) * only the ab pic b detection , and * ( right ) * both detections .
posterior probability is plotted in logarithmic units . for ab pic b , in the case of small ( @xmath850 au ) cutoffs , ln(pdf ) trended to negative infinity and is not plotted here .
[ fig:1d_det ] , title="fig:",width=240 ] -0.2 in pic b detection , * ( center ) * only the ab pic b detection , and * ( right ) * both detections .
posterior probability is plotted in logarithmic units . for ab pic b , in the case of small ( @xmath850 au ) cutoffs , ln(pdf ) trended to negative infinity and is not plotted here .
[ fig:1d_det ] , title="fig:",width=240 ] -0.2 in pic b detection , * ( center ) * only the ab pic b detection , and * ( right ) * both detections .
posterior probability is plotted in logarithmic units . for ab pic b , in the case of small ( @xmath850 au ) cutoffs , ln(pdf ) trended to negative infinity and is not plotted here .
[ fig:1d_det ] , title="fig:",width=240 ] -0.2 in pic b detection , * ( center ) * only the ab pic b detection , and * ( right ) * both detections .
posterior probability is plotted in logarithmic units . for ab pic b , in the case of small ( @xmath850 au ) cutoffs , ln(pdf ) trended to negative infinity and is not plotted here .
[ fig:1d_det ] , title="fig:",width=240 ] -0.2 in pic b detection , * ( center ) * only the ab pic b detection , and * ( right ) * both detections .
posterior probability is plotted in logarithmic units . for ab pic b , in the case of small ( @xmath850 au ) cutoffs , ln(pdf ) trended to negative infinity and is not plotted here .
[ fig:1d_det ] , title="fig:",width=240 ] -0.2 in pic b detection , * ( center ) * only the ab pic b detection , and * ( right ) * both detections .
posterior probability is plotted in logarithmic units . for ab pic b , in the case of small ( @xmath850 au ) cutoffs , ln(pdf ) trended to negative infinity and is not plotted here .
[ fig:1d_det ] , title="fig:",width=240 ] -0.2 in pic b detection , * ( center ) * only the ab pic b detection , and * ( right ) * both detections .
posterior probability is plotted in logarithmic units . for ab pic b , in the case of small ( @xmath850 au ) cutoffs , ln(pdf ) trended to negative infinity and is not plotted here .
[ fig:1d_det ] , title="fig:",width=240 ] -0.2 in pic b detection , * ( center ) * only the ab pic b detection , and * ( right ) * both detections .
posterior probability is plotted in logarithmic units . for ab pic b , in the case of small ( @xmath850 au ) cutoffs , ln(pdf ) trended to negative infinity and is not plotted here .
[ fig:1d_det ] , title="fig:",width=240 ] -0.2 in pic b detection , * ( center ) * only the ab pic b detection , and * ( right ) * both detections .
posterior probability is plotted in logarithmic units . for ab pic b , in the case of small ( @xmath850 au ) cutoffs , ln(pdf ) trended to negative infinity and is not plotted here .
[ fig:1d_det ] , title="fig:",width=240 ]
by the end of nici campaign observations , we were unable to obtain followup observations for a number of candidate companions at projected separations @xmath4400 au or in dense stellar fields ( @xmath420 objects in the field ) located in the galactic bulge or disk .
here we list target stars with candidate companions for which we only have a single epoch of data .
since we can not classify these as either cpm or background , we provide the properties of these candidates here and note the changes we make to the contrast curves in the action column .
contrast curves are edited either by reverting to a less sensitive contrast curve or restricting the contrast curve to within a given separation . for cases where we have only a single epoch and there are candidates inside the 100% coverage region for position angle
, we drop the star from our analysis .
cccccccc ao men & 1 & 4.74 & 183 & 108.1 & 15.6 & 2009.1041 & revert to 2009.10 asdi + & 2 & 6.87 & 264 & 32.1 & 15.6 & 2009.1041 & + & 3 & 6.92 & 266 & 102.4 & 14.7 & 2009.1041 & + & 4 & 7.86 & 303 & 65.4 & 15.2 & 2009.1041 & + cd-54 7336 & 1 & 2.92 & 193 & 252.5 & 8.1 & 2009.2684 & drop + & 2 & 4.26 & 281 & 228.6 & 15.0 & 2009.2684 & + & 3 & 4.32 & 285 & 27.7 & 10.6 & 2009.2684 & + & 4 & 4.74 & 313 & 207.1 & 13.2 & 2009.2684 & + & 5 & 5.39 & 356 & 264.5 & 14.1 & 2009.2684 & + & 6 & 6.28 & 415 & 132.0 & 11.6 & 2009.2684 & + & 7 & 6.36 & 420 & 100.6 & 5.7 & 2009.2684 & + & 8 & 6.55 & 433 & 332.6 & 14.6 & 2009.2684 & + & 9 & 6.95 & 459 & 296.6 & 13.3 & 2009.2684 & + & 10 & 7.04 & 465 & 167.6 & 11.5 & 2009.2684 & + & 11 & 7.15 & 472 & 91.6 & 11.7 & 2009.2684 & + & 12 & 7.68 & 507 & 105.5 & 11.1 & 2009.2684 & + & 13 & 8.17 & 539 & 284.1 & 13.4 & 2009.2684 & + & 14 & 8.24 & 544 & 346.0 & 13.2 & 2009.2684 & + & 15 & 8.67 & 572 & 100.1 & 14.7 & 2009.2684 & + cd-31 16041 & 1 & 1.94 & 99 & 65.3 & 11.7 & 2009.2657 & drop + & 2 & 2.59 & 132 & 87.2 & 12.9 & 2009.2657 & + & 3 & 3.12 & 159 & 123.5 & 14.2 & 2009.2657 & + & 4 & 3.47 & 177 & 176.1 & 11.1 & 2009.2657 & + & 5 & 3.60 & 184 & 156.5 & 11.8 & 2009.2657 & + & 6 & 3.62 & 185 & 12.0 & 14.8 & 2009.2657 & + & 7 & 3.63 & 185 & 174.3 & 14.6 & 2009.2657 & + & 8 & 3.69 & 188 & 160.7 & 15.1 & 2009.2657 & + & 9 & 3.69 & 188 & 16.7 & 10.9 & 2009.2657 & + & 10 & 3.89 & 199 & 111.4 & 14.4 & 2009.2657 & + & 11 & 4.45 & 227 & 188.1 & 12.1 & 2009.2657 & + & 12 & 4.50 & 229 & 71.1 & 14.0 & 2009.2657 & + & 13 & 4.85 & 247 & 115.4 & 14.8 & 2009.2657 & + & 14 & 5.01 & 256 & 65.0 & 10.7 & 2009.2657 & + & 15 & 5.09 & 260 & 77.6 & 12.8 & 2009.2657 & + & 16 & 5.13 & 262 & 69.1 & 14.9 & 2009.2657 & + & 17 & 5.61 & 286 & 273.7 & 13.6 & 2009.2657 & + & 18 & 5.62 & 287 & 201.2 & 12.2 & 2009.2657 & + & 19 & 5.77 & 294 & 77.6 & 14.7 & 2009.2657 & + & 20 & 5.85 & 298 & 345.2 & 14.2 & 2009.2657 & + & 21 & 6.01 & 307 & 282.4 & 14.0 & 2009.2657 & + & 22 & 6.50 & 332 & 16.0 & 13.1 & 2009.2657 & + & 23 & 6.79 & 346 & 235.9 & 15.0 & 2009.2657 & + & 24 & 6.87 & 350 & 177.0 & 12.1 & 2009.2657 & + & 25 & 7.12 & 363 & 249.0 & 13.1 & 2009.2657 & + & 26 & 7.92 & 404 & 90.4 & 14.3 & 2009.2657 & + & 27 & 8.09 & 413 & 326.3 & 13.7 & 2009.2657 & + & 28 & 8.35 & 426 & 72.7 & 14.6 & 2009.2657 & + & 29 & 8.42 & 430 & 3.8 & 14.5 & 2009.2657 & + & 30 & 8.98 & 458 & 340.4 & 13.2 & 2009.2657 & + & 31 & 9.12 & 465 & 76.2 & 11.4 & 2009.2657 & + & 32 & 9.17 & 468 & 37.4 & 14.8 & 2009.2657 & + & 33 & 9.32 & 475 & 74.1 & 11.2 & 2009.2657 & + & 34 & 10.05 & 512 & 328.5 & 13.2 & 2009.2657 & + hd 159911 & 1 & 3.79 & 171 & 118.3 & 13.8 & 2010.2712 & drop + & 2 & 3.99 & 179 & 79.1 & 10.7 & 2010.2712 & + & 3 & 4.86 & 219 & 288.3 & 14.0 & 2010.2712 & + & 4 & 4.95 & 223 & 315.9 & 13.3 & 2010.2712 & + & 5 & 5.24 & 236 & 159.3 & 13.1 & 2010.2712 & + & 6 & 5.35 & 241 & 315.1 & 11.2 & 2010.2712 & + & 7 & 5.74 & 258 & 343.1 & 10.5 & 2010.2712 & + & 8 & 5.75 & 259 & 357.5 & 12.5 & 2010.2712 & + & 9 & 5.79 & 261 & 259.3 & 13.2 & 2010.2712 & + & 10 & 6.24 & 281 & 165.8 & 12.9 & 2010.2712 & + & 11 & 6.27 & 282 & 248.5 & 14.8 & 2010.2712 & + & 12 & 7.14 & 321 & 287.0 & 11.3 & 2010.2712 & + & 13 & 7.23 & 326 & 248.6 & 12.8 & 2010.2712 & + & 14 & 7.67 & 345 & 18.6 & 8.0 & 2010.2712 & + tyc
7443 - 1102 - 1a & 1 & 7.94 & 458 & 155.4 & 12.3 & 2010.3507 & sep @xmath8 7.94 `` + & 2 & 8.89 & 513 & 355.0 & 13.6 & 2010.3507 & + tyc 7443 - 1102 - 1b & 1 & 8.85 & 511 & 12.7 & 12.8 & 2010.3534 & sep @xmath8 8.85 '' + gj 560 a & 1 & 5.71 & 94 & 230.7 & 15.9 & 2009.1891 & sep @xmath8 5.88 `` + & 2 & 5.88 & 96 & 185.6 & 15.8 & 2009.1891 & + & 3 & 6.15 & 101 & 27.5 & 16.3 & 2009.1891 & + & 4 & 6.30 & 103 & 52.6 & 16.1 & 2009.1891 & + & 5 & 6.86 & 113 & 269.0 & 14.6 & 2009.1891 & + & 6 & 7.34 & 120 & 284.4 & 15.0 & 2009.1891 & + hd 139664 & 1 & 4.49 & 79 & 241.6 & 16.2 & 2010.3507 & revert to 2009.10 asdi + & 2 & 5.19 & 91 & 311.6 & 17.2 & 2010.3507 & + & 3 & 6.08 & 106 & 335.6 & 15.8 & 2010.3507 & + & 4 & 6.45 & 113 & 310.0 & 16.8 & 2010.3507 & + & 5 & 7.04 & 123 & 16.6 & 15.8 & 2010.3507 & + & 6 & 8.48 & 148 & 68.1 & 15.7 & 2010.3507 & + & 7 & 9.13 & 160 & 301.2 & 15.2 & 2010.3507 & + & 8 & 10.55 & 185 & 37.7 & 16.7 & 2010.3507 & + hd 92945 & 1 & 8.43 & 182 & 253.4 & 15.9 & 2009.0383 & sep @xmath8 8.43 '' + tyc 9073 - 762 - 1 & 1 & 8.01 & 433 & 27.9 & 16.7 & 2010.3507 & no change + & 2 & 9.31 & 503 & -78.2 & 15.4 & 2010.3507 & + gsc
8894 - 0426 & 1 & 9.93 & 228 & -118.2 & 9.6 & 2010.1589 & sep @xmath8 9.93 `` + hd 164249b & 1 & 4.60 & 216 & 136.3 & 12.6 & 2009.2712 & sep @xmath8 6.27 '' , revert to 2009.27 asdi + & 2 & 6.27 & 295 & 80.9 & 7.0 & 2009.2712 & + & 3 & 7.72 & 363 & 58.1 & 12.9 & 2009.2712 & + & 4 & 8.27 & 389 & 36.5 & 10.2 & 2009.2712 & + hr
4796 b & 1 & 8.86 & 595 & 103.4 & 11.8 & 2009.1151 & sep @xmath8 8.86 " + v343 nor & 1 & 1.32 & 53 & 317.2 & 14.6 & 2009.1781 & drop + & 2 & 1.77 & 70 & 170.6 & 12.8 & 2009.1781 & + & 3 & 1.81 & 72 & 15.6 & 13.9 & 2009.1781 & + & 4 & 1.91 & 76 & 330.1 & 14.0 & 2009.1781 & + & 5 & 2.08 & 83 & 251.9 & 14.8 & 2009.1781 & + & 6 & 2.16 & 86 & 108.9 & 15.5 & 2009.1781 & + & 7 & 2.30 & 91 & 198.7 & 9.1 & 2009.1781 & + & 8 & 2.41 & 96 & 188.1 & 12.3 & 2009.1781 & + & 9 & 2.49 & 99 & 195.0 & 13.8 & 2009.1781 & + & 10 & 2.61 & 104 & 251.9 & 16.1 & 2009.1781 & + & 11 & 2.67 & 106 & 71.2 & 13.7 & 2009.1781 & + & 12 & 2.69 & 107 & 324.0 & 15.9 & 2009.1781 & + & 13 & 2.93 & 117 & 131.0 & 13.5 & 2009.1781 & + & 14 & 2.93 & 117 & 45.2 & 13.8 & 2009.1781 & + & 15 & 3.04 & 121 & 86.6 & 14.0 & 2009.1781 & + & 16 & 3.04 & 121 & 350.1 & 8.8 & 2009.1781 & + & 17 & 3.07 & 122 & 169.2 & 15.8 & 2009.1781 & + & 18 & 3.09 & 123 & 91.7 & 15.7 & 2009.1781 & + & 19 & 3.27 & 130 & 51.2 & 15.0 & 2009.1781 & + & 20 & 3.33 & 132 & 27.0 & 15.8 & 2009.1781 & + & 21 & 3.34 & 133 & 206.2 & 12.5 & 2009.1781 & + & 22 & 3.38 & 135 & 86.7 & 15.8 & 2009.1781 & + & 23 & 3.67 & 146 & 11.6 & 13.6 & 2009.1781 & + & 24 & 3.76 & 150 & 33.3 & 13.9 & 2009.1781 & + & 25 & 3.77 & 150 & 257.4 & 10.2 & 2009.1781 & + & 26 & 3.78 & 150 & 331.6 & 14.1 & 2009.1781 & + & 27 & 3.83 & 152 & 18.6 & 13.7 & 2009.1781 & + & 28 & 3.91 & 156 & 293.9 & 11.8 & 2009.1781 & + & 29 & 3.96 & 157 & 54.8 & 13.2 & 2009.1781 & + & 30 & 4.14 & 165 & 110.3 & 12.0 & 2009.1781 & + & 31 & 4.22 & 168 & 11.5 & 13.9 & 2009.1781 & + & 32 & 4.25 & 169 & 80.0 & 7.8 & 2009.1781 & + & 33 & 4.28 & 170 & 328.0 & 14.6 & 2009.1781 & + & 34 & 4.30 & 171 & 93.4 & 15.1 & 2009.1781 & + & 35 & 4.41 & 175 & 95.8 & 14.9 & 2009.1781 & + & 36 & 4.45 & 177 & 148.9 & 12.3 & 2009.1781 & + & 37 & 4.48 & 178 & 237.3 & 14.3 & 2009.1781 & + & 38 & 4.52 & 180 & 356.1 & 15.3 & 2009.1781 & + & 39 & 4.58 & 182 & 205.1 & 11.2 & 2009.1781 & + & 40 & 4.62 & 184 & 133.4 & 15.5 & 2009.1781 & + & 41 & 4.67 & 186 & 334.7 & 15.2 & 2009.1781 & + & 42 & 4.69 & 187 & 27.4 & 14.6 & 2009.1781 & + & 43 & 4.73 & 188 & 196.3 & 12.3 & 2009.1781 & + & 44 & 4.74 & 189 & 212.0 & 10.7 & 2009.1781 & + & 45 & 4.77 & 190 & 102.6 & 11.5 & 2009.1781 & + & 46 & 4.80 & 191 & 136.5 & 13.1 & 2009.1781 & + & 47 & 4.82 & 192 & 181.7 & 14.8 & 2009.1781 & + & 48 & 4.86 & 193 & 26.1 & 15.2 & 2009.1781 & + & 49 & 4.88 & 194 & 309.0 & 14.7 & 2009.1781 & + & 50 & 4.91 & 195 & 319.6 & 14.3 & 2009.1781 & + & 51 & 4.93 & 196 & 272.5 & 12.4 & 2009.1781 & + & 52 & 4.93 & 196 & 314.7 & 12.8 & 2009.1781 & + & 53 & 4.94 & 196 & 230.5 & 14.3 & 2009.1781 & + & 54 & 5.01 & 199 & 359.8 & 13.9 & 2009.1781 & + & 55 & 5.18 & 206 & 237.7 & 14.9 & 2009.1781 & + & 56 & 5.23 & 208 & 291.5 & 13.7 & 2009.1781 & + & 57 & 5.24 & 208 & 308.9 & 15.1 & 2009.1781 & + & 58 & 5.39 & 214 & 155.6 & 15.3 & 2009.1781 & + & 59 & 5.39 & 215 & 295.5 & 11.3 & 2009.1781 & + & 60 & 5.42 & 216 & 210.9 & 15.0 & 2009.1781 & + & 61 & 5.45 & 217 & 229.3 & 15.6 & 2009.1781 & + & 62 & 5.46 & 217 & 239.6 & 15.4 & 2009.1781 & + & 63 & 5.48 & 218 & 127.5 & 14.2 & 2009.1781 & + & 64 & 5.53 & 220 & 335.4 & 8.8 & 2009.1781 & + & 65 & 5.55 & 221 & 196.6 & 15.3 & 2009.1781 & + & 66 & 5.60 & 223 & 313.8 & 12.6 & 2009.1781 & + & 67 & 5.63 & 224 & 354.5 & 14.2 & 2009.1781 & + & 68 & 5.65 & 225 & 164.5 & 14.7 & 2009.1781 & + & 69 & 5.74 & 228 & 165.9 & 15.0 & 2009.1781 & + & 70 & 5.76 & 229 & 331.1 & 14.0 & 2009.1781 & + & 71 & 5.83 & 232 & 201.5 & 11.9 & 2009.1781 & + & 72 & 5.84 & 233 & 232.7 & 15.7 & 2009.1781 & + & 73 & 5.87 & 234 & 81.4 & 13.8 & 2009.1781 & + & 74 & 5.94 & 236 & 94.0 & 13.9 & 2009.1781 & + & 75 & 5.95 & 237 & 191.1 & 13.5 & 2009.1781 & + & 76 & 5.96 & 237 & 306.7 & 12.9 & 2009.1781 & + & 77 & 6.07 & 241 & 149.2 & 12.1 & 2009.1781 & + & 78 & 6.08 & 242 & 312.6 & 14.6 & 2009.1781 & + & 79 & 6.09 & 242 & 336.5 & 15.4 & 2009.1781 & + & 80 & 6.10 & 243 & 60.6 & 12.0 & 2009.1781 & + & 81 & 6.17 & 246 & 102.9 & 15.7 & 2009.1781 & + & 82 & 6.21 & 247 & 107.5 & 13.5 & 2009.1781 & + & 83 & 6.23 & 248 & 330.8 & 14.3 & 2009.1781 & + & 84 & 6.25 & 249 & 209.4 & 14.1 & 2009.1781 & + & 85 & 6.26 & 249 & 288.2 & 13.7 & 2009.1781 & + & 86 & 6.36 & 253 & 116.5 & 15.7 & 2009.1781 & + & 87 & 6.42 & 255 & 59.6 & 10.5 & 2009.1781 & + & 88 & 6.47 & 258 & 202.9 & 14.1 & 2009.1781 & + & 89 & 6.53 & 260 & 202.1 & 14.6 & 2009.1781 & + & 90 & 6.53 & 260 & 3.9 & 15.3 & 2009.1781 & + & 91 & 6.55 & 261 & 226.6 & 15.1 & 2009.1781 & + & 92 & 6.57 & 261 & 325.7 & 15.2 & 2009.1781 & + & 93 & 6.63 & 264 & 105.2 & 15.4 & 2009.1781 & + & 94 & 6.67 & 265 & 248.3 & 13.5 & 2009.1781 & + & 95 & 6.67 & 266 & 40.6 & 13.6 & 2009.1781 & + & 96 & 6.68 & 266 & 121.5 & 15.3 & 2009.1781 & + & 97 & 6.71 & 267 & 70.8 & 9.0 & 2009.1781 & + & 98 & 6.71 & 267 & 9.2 & 14.5 & 2009.1781 & + & 99 & 6.75 & 269 & 230.7 & 13.9 & 2009.1781 & + & 100 & 6.77 & 269 & 185.6 & 10.5 & 2009.1781 & + & 101 & 6.80 & 271 & 85.0 & 15.7 & 2009.1781 & + & 102 & 6.85 & 273 & 353.4 & 13.8 & 2009.1781 & + & 103 & 6.86 & 273 & 292.1 & 15.3 & 2009.1781 & + & 104 & 6.89 & 274 & 279.8 & 13.7 & 2009.1781 & + & 105 & 6.98 & 278 & 84.8 & 15.3 & 2009.1781 & + & 106 & 7.14 & 284 & 182.5 & 14.4 & 2009.1781 & + & 107 & 7.17 & 285 & 148.1 & 13.5 & 2009.1781 & + & 108 & 7.25 & 289 & 37.6 & 12.9 & 2009.1781 & + & 109 & 7.26 & 289 & 269.4 & 14.1 & 2009.1781 & + & 110 & 7.28 & 290 & 300.6 & 14.2 & 2009.1781 & + & 111 & 7.35 & 292 & 89.6 & 12.8 & 2009.1781 & + & 112 & 7.41 & 295 & 92.8 & 13.4 & 2009.1781 & + & 113 & 7.46 & 297 & 257.8 & 13.9 & 2009.1781 & + & 114 & 7.48 & 298 & 80.9 & 12.5 & 2009.1781 & + & 115 & 7.63 & 304 & 297.5 & 9.1 & 2009.1781 & + & 116 & 7.74 & 308 & 201.9 & 13.6 & 2009.1781 & + & 117 & 7.77 & 309 & 112.1 & 12.2 & 2009.1781 & + & 118 & 7.85 & 312 & 204.0 & 12.3 & 2009.1781 & + & 119 & 7.96 & 317 & 24.3 & 13.8 & 2009.1781 & + hd 139084 b & 1 & 0.92 & 37 & 206.6 & 8.1 & 2009.1123 & drop + & 2 & 1.19 & 48 & 347.8 & 9.5 & 2009.1123 & + & 3 & 1.38 & 55 & 119.6 & 11.2 & 2009.1123 & + & 4 & 1.51 & 60 & 32.4 & 12.3 & 2009.1123 & + & 5 & 1.78 & 71 & 351.3 & 12.1 & 2009.1123 & + & 6 & 1.84 & 73 & 241.4 & 11.1 & 2009.1123 & + & 7 & 1.96 & 78 & 355.3 & 11.6 & 2009.1123 & + & 8 & 2.11 & 84 & 258.7 & 9.4 & 2009.1123 & + & 9 & 2.22 & 88 & 14.0 & 11.3 & 2009.1123 & + & 10 & 2.23 & 89 & 267.2 & 11.4 & 2009.1123 & + & 11 & 2.26 & 90 & 226.3 & 7.3 & 2009.1123 & + & 12 & 2.39 & 95 & 57.3 & 9.3 & 2009.1123 & + & 13 & 2.43 & 97 & 241.7 & 11.7 & 2009.1123 & + & 14 & 2.45 & 97 & 302.4 & 9.8 & 2009.1123 & + & 15 & 2.73 & 109 & 290.8 & 11.9 & 2009.1123 & + & 16 & 2.79 & 111 & 326.0 & 10.0 & 2009.1123 & + & 17 & 2.82 & 112 & 39.6 & 6.2 & 2009.1123 & + & 18 & 2.85 & 113 & 26.6 & 11.7 & 2009.1123 & + & 19 & 2.86 & 114 & 129.4 & 12.0 & 2009.1123 & + & 20 & 2.88 & 115 & 303.3 & 11.9 & 2009.1123 & + & 21 & 3.02 & 120 & 183.7 & 9.8 & 2009.1123 & + & 22 & 3.45 & 137 & 335.7 & 11.8 & 2009.1123 & + & 23 & 3.60 & 143 & 188.6 & 9.7 & 2009.1123 & + & 24 & 3.62 & 144 & 83.4 & 11.4 & 2009.1123 & + & 25 & 3.69 & 147 & 198.4 & 10.7 & 2009.1123 & + & 26 & 3.72 & 148 & 192.6 & 12.2 & 2009.1123 & + & 27 & 3.77 & 150 & 139.6 & 11.6 & 2009.1123 & + & 28 & 3.87 & 154 & 23.2 & 10.3 & 2009.1123 & + & 29 & 3.98 & 158 & 8.8 & 11.0 & 2009.1123 & + & 30 & 4.06 & 162 & 210.9 & 11.6 & 2009.1123 & + & 31 & 4.17 & 166 & 182.7 & 12.3 & 2009.1123 & + & 32 & 4.22 & 168 & 132.6 & 10.6 & 2009.1123 & + & 33 & 4.30 & 171 & 60.4 & 9.6 & 2009.1123 & + & 34 & 4.46 & 178 & 6.0 & 10.6 & 2009.1123 & + & 35 & 4.48 & 178 & 158.4 & 11.1 & 2009.1123 & + & 36 & 4.56 & 182 & 181.4 & 11.0 & 2009.1123 & + & 37 & 4.65 & 185 & 134.4 & 10.5 & 2009.1123 & + & 38 & 4.66 & 186 & 194.7 & 11.5 & 2009.1123 & + & 39 & 4.80 & 191 & 268.3 & 9.2 & 2009.1123 & + & 40 & 4.85 & 193 & 310.9 & 11.2 & 2009.1123 & + & 41 & 4.88 & 194 & 154.9 & 9.0 & 2009.1123 & + & 42 & 4.91 & 195 & 164.1 & 9.3 & 2009.1123 & + & 43 & 5.03 & 200 & 203.5 & 10.5 & 2009.1123 & + & 44 & 5.04 & 201 & 130.7 & 5.0 & 2009.1123 & + & 45 & 5.08 & 202 & 248.9 & 9.8 & 2009.1123 & + & 46 & 5.13 & 204 & 289.6 & 9.3 & 2009.1123 & + & 47 & 5.16 & 205 & 310.6 & 11.2 & 2009.1123 & + & 48 & 5.27 & 210 & 210.1 & 11.7 & 2009.1123 & + & 49 & 5.36 & 213 & 22.4 & 10.1 & 2009.1123 & + & 50 & 5.38 & 214 & 326.3 & 10.4 & 2009.1123 & + & 51 & 5.38 & 214 & 25.1 & 9.1 & 2009.1123 & + & 52 & 5.40 & 215 & 158.1 & 11.9 & 2009.1123 & + & 53 & 5.42 & 216 & 147.3 & 10.8 & 2009.1123 & + & 54 & 5.47 & 218 & 82.3 & 10.5 & 2009.1123 & + & 55 & 5.50 & 219 & 151.9 & 9.3 & 2009.1123 & + & 56 & 5.56 & 221 & 105.7 & 9.8 & 2009.1123 & + & 57 & 5.58 & 222 & 345.7 & 9.9 & 2009.1123 & + & 58 & 5.58 & 222 & 228.0 & 5.9 & 2009.1123 & + & 59 & 5.65 & 225 & 333.8 & 8.8 & 2009.1123 & + & 60 & 5.66 & 225 & 27.4 & 10.4 & 2009.1123 & + & 61 & 5.67 & 226 & 260.1 & 8.5 & 2009.1123 & + & 62 & 5.70 & 227 & 182.5 & 12.0 & 2009.1123 & + & 63 & 5.72 & 228 & 340.9 & 11.6 & 2009.1123 & + & 64 & 5.74 & 228 & 277.4 & 12.0 & 2009.1123 & + & 65 & 5.76 & 229 & 237.6 & 11.4 & 2009.1123 & + & 66 & 5.83 & 232 & 43.4 & 10.8 & 2009.1123 & + & 67 & 5.90 & 235 & 77.5 & 9.7 & 2009.1123 & + & 68 & 6.00 & 239 & 269.7 & 9.8 & 2009.1123 & + & 69 & 6.02 & 240 & 60.5 & 8.3 & 2009.1123 & + & 70 & 6.05 & 241 & 140.8 & 11.2 & 2009.1123 & + & 71 & 6.10 & 243 & 168.5 & 7.8 & 2009.1123 & + & 72 & 6.13 & 244 & 328.2 & 11.1 & 2009.1123 & +
|
we report results of a direct imaging survey for giant planets around 80 members of the @xmath0 pic , tw hya , tucana - horologium , ab dor , and hercules - lyra moving groups , observed as part of the gemini nici planet - finding campaign . for this sample
, we obtained median contrasts of @xmath1@xmath2=13.9 mag at 1 in combined ch@xmath3 narrowband adi+sdi mode and median contrasts of @xmath1@xmath2=15.1 mag at 2 in @xmath2-band adi mode .
we found numerous ( @xmath470 ) candidate companions in our survey images .
some of these candidates were rejected as common - proper motion companions using archival data ; we reobserved with nici all other candidates that lay within 400 au of the star and were not in dense stellar fields .
the vast majority of candidate companions were confirmed as background objects from archival observations and/or dedicated nici campaign followup .
four co - moving companions of brown dwarf or stellar mass were discovered in this moving group sample : pz tel b ( 36@xmath56 m@xmath6 , [email protected] au , biller et al .
2010 ) , cd -35 2722b ( 31@xmath58 m@xmath6 , 67@xmath54 au , wahhaj et al .
2011 ) , hd 12894b ( [email protected] m@xmath7 , [email protected] au ) , and bd+07 1919c ( [email protected] m@xmath7 , [email protected] au ) . from a bayesian analysis of the achieved h band adi and asdi contrasts , using power - law models of planet distributions and hot - start evolutionary models , we restrict the frequency of 120 m@xmath6 companions at semi - major axes from 10150 au to @xmath818% at a 95.4@xmath9 confidence level using dusty models and to @xmath86% at a 95.4@xmath9 using cond models .
our results strongly constrain the frequency of planets within semi - major axes of 50 au as well .
we restrict the frequency of 120 m@xmath6 companions at semi - major axes from 1050 au to @xmath821% at a 95.4@xmath9 confidence level using dusty models and to @xmath87% at a 95.4@xmath9 using cond models .
this survey is the deepest search to date for giant planets around young moving group stars .
|
the advantage of ultracold fermi gases is the existence of highly tunable physical parameters@xcite .
a tunable pairing interaction associated with a feshbach resonance@xcite enables us to study fermi superfluids from the weak - coupling bcs regime to the strong - coupling bec limit in a unified manner@xcite .
the intermediate coupling regime ( which is also referred to as the bcs - bec crossover region in the literature ) is useful for the study of strong - coupling physics .
since correlation effects are important in high-@xmath3 cuprates@xcite , the bcs - bec crossover physics in ultracold fermi gases would be also useful for the study of this interacting electron system .
in addition to the tunable interaction , one can also adjust the atomic motion by using an optical lattice@xcite .
for example , when a fermi gas is loaded on a one - dimensional optical lattice , the particle motion in the the lattice direction is strongly suppressed for a very high lattice potential , which effectively realizes a two - dimensional gas .
thus , an optical lattice can be used to change the system dimension@xcite . using this optical lattice technique , feld and co - workers@xcite
have recently done the photoemission - type experiment on a two - dimensional @xmath4k fermi gas . in the bcs - bec crossover region
, they found that the photoemission spectrum exhibits a double peak structure ( although the system is in the normal state ) .
although a similar anomaly has also been observed in a three - dimensional @xmath4k fermi gas@xcite , the double peak structure observed in the former low - dimensional system is more remarkable than the latter , which indicates that the low - dimensionality enhances this anomaly .
indeed , evolution of a double - peak structure in the rf ( radio frequency)-spectrum has also been observed in a @xmath5li fermi gas@xcite , when the system dimension is changed from the three- to two - dimension by using an optical lattice . in the case of a three - dimensional fermi gas , it has been theoretically shown that the anomaly seen in the photoemission spectrum can be explained as a pseudogap phenomenon originating from strong pairing fluctuations@xcite . in this regard
, we note that pairing fluctuations are stronger in a lower dimensional system .
thus , although the existence of the pseudogap is still controversial in ultracold fermi gases@xcite , if the pseudogap scenario proposed in the three - dimensional case is correct , the anomalous photoemission spectrum observed in a two - dimensional @xmath4k fermi gas is also expected to originate from pairing fluctuations that are enhanced by the low - dimensionality of the system . in this paper
, we investigate single - particle properties of a two - dimensional fermi gas . including two - dimensional pairing fluctuations within a strong - coupling @xmath0-matrix approximation@xcite , as well as effects of a harmonic trap within the local density approximation ( lda)@xcite
, we calculate the local density of states in the normal state .
we also deal with the photoemission spectrum , to see if the observed double peak structure can be understood as a two - dimensional pseudogap phenomenon .
we briefly note that the possibility of the berezinskii - kosterlitz - thouless ( bkt ) transition@xcite in a two - dimensional fermi gas have recently been discussed by many researchers@xcite .
pseudogap physics in this system has also been studied in refs.@xcite .
this paper is organized as follows . in sec .
ii , we explain our formulation . in sec .
iii , we examine the local density of states . here , we show how two - dimensional pairing fluctuations affect this single - particle quantity , leading to the pseudogap phenomenon . in sec .
iv , we discuss the photoemission spectrum .
we compare our results with the recent experiment on a @xmath4k fermi gas done by feld and coworkers@xcite . throughout this paper , we set @xmath6 , and the system area @xmath7 is taken to be unity , for simplicity .
we consider a two - dimensional fermi gas with two atomic hyperfine states , described by the bcs hamiltonian , @xmath8 here , @xmath9 is a creation operator of a fermi atom with pseudospin @xmath10 , describing two atomic hyperfine states .
@xmath11 is the kinetic energy , measured from the chemical potential @xmath12 ( where @xmath13 is an atomic mass ) .
since we are considering a two - dimensional system , @xmath14 is a two - dimensional momentum .
@xmath15 is a pairing interaction , which is assumed to be tunable . within lda , effects of a harmonic trap
is conveniently incorporated into the theory by replacing the chemical potential @xmath12 with the lda expression , @xmath16 here , @xmath17 is a two - dimensional trap potential , where the position @xmath18 is measured from the trap center , and @xmath19 is a trap frequency . in lda ,
effects of spatial inhomogeneity enter into the theory through the lda chemical potential @xmath20 in eq .
( [ eq.mu ] ) .
the lda single - particle thermal green s function is given by @xmath21 where @xmath22 is the fermion matsubara frequency .
@xmath23^{-1}$ ] is the thermal green s function for a free fermi gas , where @xmath24 .
the lda self - energy @xmath25 in eq .
( [ eq.g ] ) describes fluctuation corrections to single - particle excitations . in the ordinary @xmath0-matrix approximation , it is given by@xcite , @xmath26 where @xmath27 is the boson matsubara frequency .
the particle - particle scattering matrix @xmath28 in eq .
( [ eq.s ] ) describes fluctuations in the cooper channel , having the form@xcite , @xmath29 here , @xmath30 is the lowest - order pair - propagator in lda@xcite . in the case of a _ uniform _ two - dimensional fermi gas ,
it is well known@xcite that the ordinary bcs - type superfluid phase transition is completely suppressed by strong pairing fluctuations . instead , the superfluid instability is dominated by the bkt transition@xcite . on the other hand , in the presence of a harmonic trap
, we will show that the bcs - type superfluid phase transition revives ( at least within lda ) , which is just the same as the case of bec in a trapped two - dimensional bose gas@xcite . in this case
, the lda superfluid phase transition temperature @xmath3 is simply determined from the bcs - type @xmath3-equation in the trap center@xcite , @xmath31 together with the equation for the total number @xmath32 of fermi atoms , @xmath33 in eq .
( [ numeq ] ) , @xmath34 is the lda particle density , given by @xmath35 where @xmath36 is an infinitesimally small positive number . as usual
, one needs to regularize the @xmath3-equation ( [ eq.gap ] ) to eliminate the ultraviolet divergence .
this can be achieved by introducing the two - dimensional @xmath37-wave scattering length @xmath38 , which is related to the pairing interaction @xmath39 as@xcite @xmath40 where @xmath41 , and @xmath42 is an infinitesimally small energy . in this regularization scheme ,
the interaction strength is conveniently measured in terms of the binding energy @xmath43 of a two - body bound state@xcite , @xmath44 in this scale , the increase of @xmath43 corresponds to the increase of the interaction strength .
we briefly note that @xmath43 always exists in the two - dimensional case , irrespective of the interaction strength @xmath39@xcite .
this is quite different from the three - dimensional case , where a threshold coupling strength exists to obtain a two - body bound state .
( a ) , and @xmath45 ( b ) , as functions of the interaction strength , measured in terms of @xmath46 ( where @xmath43 is the molecular binding energy in eq .
( [ bind ] ) , and @xmath47 is the lda fermi energy ) . in panel ( a ) , @xmath48 is the mean field result . ( c ) lda atomic density profile @xmath34 , when @xmath49 .
@xmath50 is the thomas - fermi radius , which gives the size of a free fermi gas at @xmath51 in lda .
( d ) calculated fermi chemical potential @xmath12 above @xmath3.,scaledwidth=90.0% ] figures [ fig1](a ) and ( b ) show the self - consistent solutions of the coupled eqs.([eq.gap ] ) and ( [ numeq ] ) . as in the three - dimensional case@xcite , @xmath3 and @xmath12 exhibit
typical bcs - bec crossover behaviors in the present two - dimensional case . in panel
( a ) , @xmath3 gradually deviates from the mean - field result ( @xmath48 ) with increasing the interaction strength , to approach a constant value in the strong - coupling bec regime ( @xmath52 , where @xmath47 is the lda fermi energy ) . in the bec limit , the system is well described by a trapped ideal bose gas of @xmath53 molecules , so that @xmath3 is determined from the lda molecular number equation@xcite , @xmath54/t_{\rm c}}-1},\ ] ] which gives @xmath55 here , @xmath56 is the fermi temperature .
figure [ fig1](a ) indicates that the crossover from the bcs regime to the bec regime occurs in the region @xmath57 . in fig.[fig1](b )
, the fermi chemical potential @xmath12 gradually deviates from the fermi energy @xmath58 with increasing the interaction strength , to be negative in the strong - coupling bec regime . in the bec regime ,
@xmath59 approaches the molecular binding energy @xmath43 , reflecting that the system becomes a gas of tightly bound molecules having this binding energy .
we note that , although our lda calculation gives a finite @xmath3 in the presence of a trap , one also sees a sign of the vanishing bcs superfluid phase transition in the uniform case . as shown in fig.[fig1](c ) , @xmath34 has a peak structure at @xmath3 , which is known as an artifact of lda@xcite . while the peak value is finite in the three - dimensional case , @xmath60 diverges in the two - dimensional case
. indeed , in the strong - coupling bec limit , one finds , @xmath61 which behaves as @xmath62 around @xmath63 .
however , since the total number equation ( [ numeq ] ) converges after taking the radial integration ( @xmath64 ) of this singular particle density @xmath34 , one obtains self - consistent solutions of the coupling equations ( [ eq.gap ] ) and ( [ numeq ] ) . in the uniform case , on the other hand ,
the divergence of the ( uniform ) particle density occurs everywhere , when the @xmath3-equation ( [ eq.gap ] ) is satisfied . as a result ,
the equation for the total number @xmath32 of fermi atoms also diverges . because of this singularity , the coupled equations ( [ eq.gap ] ) and ( [ numeq ] ) do not give any self - consistent solution in the uniform case , leading to the vanishing bcs phase transition . according to the thouless criterion@xcite , the @xmath3-equation ( [ eq.gap ] ) is directly related to the pole of the particle - particle scattering matrix @xmath65 in eq .
( [ gamma ] ) at @xmath66 .
since this quantity physically describes pairing fluctuations , the vanishing @xmath3 is due to strong pairing fluctuations in a two - dimensional fermi gas@xcite .
since all the current experiments on two - dimensional fermi gases have been done in the normal state@xcite , we also consider the region above @xmath3 in this paper . in the normal state
, one may only solve the number equation ( [ numeq ] ) to obtain @xmath12 shown in fig.[fig1](d ) .
then , the lda single - particle spectral weight @xmath67 , as well as the local density of states @xmath68 , are conveniently calculated from the analytic continued green s function as , @xmath69 , \label{lsw}\ ] ] @xmath70 the photoemission spectrum is also related to the spectral weight @xmath67 in eq .
( [ lsw ] ) . in this experiment ,
atoms in one of the two hyperfine states ( @xmath71 ) is transfered to another hyperfine state @xmath72 by rf - pulse , and the tunneling current between these states is measured .
since the final state interaction can be safely ignored in a @xmath4k fermi gas@xcite , one may safely treat @xmath72 as a non - interacting state . in this case
, the lda local photoemission current @xmath73 is given by@xcite @xmath74 where @xmath75 is the fermi distribution function .
@xmath76 is a transfer matrix element between @xmath77 and @xmath72 .
@xmath78 is the energy difference between the incident photon energy @xmath79 and the energy @xmath80 of the final state @xmath72 .
since the current photoemission - type experiments@xcite do not have enough spatial resolution , one needs to take the spatial average of eq .
( [ eq.100 ] ) over the gas cloud , which gives @xmath81 where @xmath50 is the thomas - fermi radius .
the observed photoemission spectrum@xcite ( @xmath82 ) is then given by@xcite @xmath83 in later sections , we numerically evaluate @xmath68 in eq .
( [ ldos ] ) , as well as @xmath84 in eq .
( [ eq.102 ] ) , to examine strong - coupling effects on single - particle properties of a two - dimensional fermi gas . before ending this section
, we briefly explain our idea about how to compare our theoretical results with experimental data .
although the prediction for the value of @xmath3 is usually difficult in a fermi superfluid , it is known in the field of superconductivity that various superconducting phenomena can be well explained theoretically , when one uses the _ scaled _
temperature @xmath85 ( even if the theory can not reproduce the observed value of @xmath3 ) .
however , the superfluid phase transition has not been realized yet in a two - dimensional fermi gas .
thus , apart from quantitative discussions , it seems difficult to quantitatively compare the calculated temperature dependence of a physical quantity with experiment data in the current stage of research .
however , we point out that the ( pseudo)gap size @xmath86 in the photoemission spectrum at @xmath1 observed in a two - dimensional @xmath4k fermi gas is almost @xmath0-independent in the wide temperature region , @xmath87@xcite . thus , as an alternative to using the scaled temperature @xmath85
, @xmath86 would be useful for the comparison of theory with experiment in the current stage of research . in sec .
iv , we will take this strategy to assess the pseudogap scenario . at @xmath88 .
( a ) @xmath89 .
( b ) @xmath90 .
( c ) @xmath49 . in these panels , as well as in fig.[fig3
] , we have offset the results by 0.2 .
( d ) pseudogap temperature @xmath91 , determined from the temperature dependence of the local density of states @xmath68 .
fine structures seen in panel ( c ) are due to a computational problem in calculating the analytic continued green s function . ,
scaledwidth=80.0% ] at @xmath92 .
we set @xmath90 .
, scaledwidth=50.0% ]
figures [ fig2](a)-(c ) show the local density of states @xmath68 at @xmath88 in the bcs - bec crossover regime of a two - dimensional fermi gas@xcite .
we find that @xmath68 has a gap - like ( pseudogap ) structure near @xmath3 .
in particular , when @xmath93 ( panel ( c ) ) , the pseudogap structure at @xmath3 is very similar to the bcs - type superfluid density of states with sharp coherence peaks at the gas edges .
since the superfluid order parameter vanishes in these figures , the pseudogap in @xmath68 originates from two - dimensional pairing fluctuations . indeed , noting that the particle - particle scattering matrix @xmath94 ( which physically describes pairing fluctuations ) diverges at @xmath3 , we may approximate eq .
( [ eq.s ] ) to @xmath95 near @xmath3 , where @xmath96 is sometimes referred to as the pseudogap parameter in the literature@xcite . substituting this into eq.([eq.g ] ) , we obtain @xmath97 which is the same form as the mean - field green s function in the bcs state , @xmath98 thus , pairing fluctuations ( that are described by @xmath99 in eq .
( [ eq.bcs ] ) ) are found to work like the bcs gap parameter @xmath100 , leading to the pseudogap in the single - particle density of states , as shown in fig.[fig2](a)-(c ) .
we also find from the first line in eq .
( [ eq.bcs ] ) that pairing fluctuations ( @xmath99 ) couple the particle branch @xmath101 with the hole branch @xmath102 , leading to the level repulsion between the two around @xmath103 ( when @xmath104 ) . in this sense
, the pseudogap may be viewed as a particle - hole coupling phenomenon , induced by strong pairing fluctuations .
when @xmath105 , fig.[fig1](a ) shows that @xmath3 is almost equal to the mean - field value ( @xmath48 ) .
in addition , the fermi chemical potential @xmath45 is also close to the fermi energy @xmath58 there , as shown in fig.[fig1](b ) .
however , even in this weak - coupling case , @xmath68 exhibits a pseudogap around the trap center , as shown in fig.[fig2](a ) .
since such a clear pseudogap does not appear in the weak - coupling regime of a three - dimensional fermi gas@xcite , the two - dimensionality is found to enhance the pseudogap phenomenon , as expected . in figs .
[ fig2](a)-(c ) , we find that the pseudogap gradually disappears , as one increases the temperature above @xmath3 .
when we conveniently define the pseudogap temperature @xmath91 as the temperature at which the dip structure in @xmath68 completely disappears in the trap center , fig.[fig2](d ) shows that @xmath91 is relatively high even in the weak - coupling regime ( @xmath106 ) .
we briefly note that , in the three - dimensional case@xcite , the pseudogap temperature is only slightly higher than @xmath3 in this regime .
thus , the present result also indicates the enhancement of the pseudogap phenomenon by the two - dimensionality of the system . at @xmath92 .
we take @xmath107=0 $ ] .
the intensity is normalized by @xmath108 . ] at @xmath88 .
we take @xmath109=-1 $ ] .
the intensity is normalized by @xmath108 . ]
figure [ fig3 ] shows the spatial variation of the local density of states . as in the three - dimensional case@xcite ,
the pseudogap remarkably depends on the spatial position .
that is , the dip structure gradually becomes obscure , as one goes away from the trap center . around the edge of the gas cloud ( @xmath110 ) , @xmath68 is almost equal to the density of states @xmath111 in a two - dimensional free fermi gas , @xmath112 where @xmath113 is the step function .
thus , the pseudogapped density of states and the free - particle - like density of states coexist in a trapped fermi gas ( although there is no clear boundary between them ) .
we briefly note that one can also see the pseudogap phenomenon in the spectral weight @xmath67 . in fig.[fig4 ] , while @xmath67 exhibits a double peak ( pseudogap ) structure near the trap center ( panel ( a ) ) , the lower peak in the spectral weight gradually becomes obscure , as one approaches the edge of the gas cloud ( panels ( b ) and ( c ) . ) at @xmath114 ( panel ( c ) ) , @xmath67 is dominated by a single peak line , which is close to the free particle dispersion ( @xmath115 ) .
this spatial dependence of @xmath67 is consistent with that of the local density of states @xmath68 shown in fig.[fig3 ] .
in addition , as expected from the temperature dependence of the pseudogap structure shown in figs.[fig2](a)-(c ) , fig.[fig5 ] shows that the double peak structure in the spectral weight @xmath116 gradually disappears , as one increases the temperature from @xmath3 . since the spectral weight is deeply related to the photoemission spectrum ( see eq .
( [ eq.101 ] ) . ) , the above results make us expect that the latter is a useful quantity to observe the pseudogap phenomenon in a two - dimensional fermi gas , which we will confirm in the next section . in the normal state .
we take @xmath49 .
the spectral intensity is normalized by @xmath117 . in each panel ,
the solid line shows the free particle dispersion , @xmath118 , which is close to the upper peak line of the spectrum .
the dashed line shows the lower peak line of the spectrum.,scaledwidth=50.0% ] at @xmath92 .
( a ) @xmath89 .
( b ) @xmath90 .
( c ) @xmath119 .
panel ( d ) shows the peak - to - peak energy @xmath86 of the calculated photoemission spectrum at @xmath1 .
solid circles are experimental data for a @xmath4k fermi gas@xcite . the dashed line shows the lower peak line of the spectrum . ]
figure [ fig6 ] shows the intensity of the photoemission spectrum @xmath84 in a two - dimensional fermi gas . in the intermediate coupling case ( @xmath93 )
, one sees a clear double peak structure , as observed in a @xmath4k fermi gas@xcite . in the case of a free fermi gas
, a single peak line along the free - particle dispersion @xmath118 only appears in the spectrum .
in addition , the superfluid order parameter vanishes in fig.[fig6 ] .
thus , the gap - like ( double peak ) structure in fig.[fig6 ] is considered to originate from two - dimensional pairing fluctuations . in fig.[fig6 ] , the upper peak line is well described by the free particle dispersion , @xmath118 .
this mean that this branch is dominated by single - particle excitations in the outer region of the gas cloud ( @xmath110 ) where a free - particle - like local density of states shown in fig.[fig3 ] is obtained .
actually , since the inner region of the gas cloud also contributes to this branch to some extent , it has a finite spectral width . on
the other hand , the lower peak line in fig.[fig6 ] ( which does not appear in a free fermi gas ) comes from the pseudogapped single - particle excitations around the trap center . since the pseudogap phenomenon is spatially inhomogeneous in the presence of a trap ( see figs.[fig3 ] and [ fig4 ] . )
, the spatial average of the local photoemission spectrum makes this branch broad , as shown in fig.[fig6 ] .
we note that the lower branch in fig.[fig6 ] exhibits the so - called back - bending behavior .
with increasing the temperature , the momentum region where this branch shows an upward behavior becomes wide .
in addition , the energy difference ( @xmath120 ) between the two peak lines at @xmath1 is found to be insensitive to the temperature , at least up to @xmath121 .
these behaviors are consistent with the recent experiment on a @xmath4k fermi gas@xcite . as mentioned in sec .
ii , the latter property is useful for the comparison of our theory with experimental data , which we will use soon later .
figure [ fig7](a)-(c ) show that , although the pseudogap structure ( double peak structure ) is clearly seen in the photoemission spectrum in the bec side ( panel ( c ) ) , it gradually becomes obscure , as one approaches the weak - coupling bcs regime .
the double peak structure still remains when @xmath122 ( fig.[fig7](b ) ) , a broad spectral peak only exists around @xmath1 when @xmath105 ( fig.[fig7](a ) ) . in the latter case
, the local density of states @xmath123 has a clear pseudogap structure near @xmath3 , as shown in fig.[fig2](a ) .
thus , the disappearance of the double peak structure in the low momentum regime of the photoemission spectrum is a result of the spatial average of the inhomogeneous pseudogap .
figure [ fig7](d ) shows the peak - to - peak energy at @xmath1 ( @xmath86 ) in the photoemission spectrum @xmath84 . as expected from fig.[fig6 ] , @xmath86 is almost @xmath0-independent up to @xmath121 in the whole bcs - bec crossover region .
in addition , our result agrees well with the recent experiment on a @xmath4k fermi gas@xcite ( solid circles in fig.[fig7](d ) ) . here , we emphasize that no fitting parameter is introduced in obtaining this result . since the double peak structure in our theoretical results originates from the pseudogap phenomenon associated with two - dimensional pairing fluctuations , this agreement supports the pseudogap scenario as the mechanism of the double peak structure observed in the recent photoemission - type experiment on a two - dimensional @xmath4k fermi gas@xcite .
to summarize , we have discussed single - particle properties and strong - coupling effects in the bcs - bec crossover regime of a two - dimensional fermi gas . within the framework of a combined @xmath0-matrix theory with lda
, we have calculated the local density of states @xmath68 in the normal state above @xmath3 .
we showed that @xmath68 has a pseudogap near @xmath3 , originating from strong two - dimensional pairing fluctuations . in the intermediate coupling regime ( @xmath93 ) , the pseudogap structure at @xmath3 is very similar to the bcs superfluid density of states , being accompanied by sharp coherence peaks at gap edges ( although the superfluid order parameter vanishes at @xmath3 ) . even in the weak - coupling bcs regime ( @xmath105 ) , where both @xmath3 and @xmath12 are well described by the simple mean - field theory , a dip structure is still obtained .
we also showed that the pseudogap remains to a high temperature , in comparison with the three - dimensional case .
these results indicate that the pseudogap phenomenon is enhanced by two - dimensional pairing fluctuations . in the presence of a trap potential ,
the pseudogap becomes spatially inhomogeneous .
that is , while the pseudogap dominates single - particle excitations in the trap center , the local density of states near the edge of the gas cloud is still similar to that for a free fermi gas .
including this inhomogeneous pseudogap effect , we calculated the photoemission spectrum .
although the spatial average of the spectrum smears the pseudogap structure to some extent , the calculated spectrum still exhibits a double peak structure , when @xmath124 .
the peak - to - peak energy @xmath86 at @xmath1 is almost @xmath0-independent in the wide temperature region ( @xmath125 ) , which is consistent with the recent experiment on a @xmath4k fermi gas@xcite .
in addition , the value of @xmath86 agrees well with this experiment over the entire bcs - bec crossover region .
this agreement indicates that the anomalous double peak structure observed in the photoemission spectrum@xcite may be understood as a pseudogap phenomenon , originating from two - dimensional pairing fluctuations .
since the pseudogap is a fundamental phenomenon in a strongly interacting fermi system , our results would be useful for the study how this many - body phenomenon is affected by the dimensionality of a system . in the current stage of research for a two - dimensional fermi gas ,
the achievement of the superfluid phase transition is one of the most important challenges .
while the combined @xmath0-matrix theory with lda used in this paper gives a finite bcs phase transition temperature , the bkt transition is predicted in a uniform fermi gas@xcite .
thus , it would be interesting to clarify whether the superfluid phase transition in a trapped two - dimensional fermi is close to the bcs - type or the bkt - type . for this purpose , an extension of our work to include
the bkt transition would be an important future problem .
in addition , since pairing fluctuations become strong with decreasing the temperature near @xmath3 , a phenomenon which is sensitive this temperature dependence would be helpful to see to what extend the current experiment approaches the two - dimensional superfluid instability . for this purpose ,
the peak - to - peak energy @xmath86 discussed in this paper is _ not _ useful , because it is insensitive to the temperature ( although it is useful for the comparison of theory with experiment in the current stage of research ) .
thus , the search for a two - dimensional pseudogap phenomenon which has a remarkable temperature dependence near @xmath3 would be also another important future challenge .
we would like to thank m. khl for fruitful discussions .
we also thank r. hanai for reading this manuscript and giving useful comments .
r.w . was supported by the japan society for the promotion of science .
y.o . was supported by grant - in - aid for scientific research from mext in japan ( no.23500056 , no.25400418 , no.25105511 ) .
99 s. giorgini , j. p. pitaevskii , and s. stringari , rev .
phys . * 80 * , 1215 ( 2008 ) .
i. bloch , j. dalibard , and w. zwerger , rev .
* 80 * , 885 ( 2008 ) .
w. ketterle and m. w. zwierlein , proceedings of the international school of physics `` enrico fermi '' , course clxiv , edited by m. inguscio , w. ketterle , and c. salomon ( ios press , amsterdam 2007 ) p.95 .
e. timmermans , k. furuya , p. w. milonni , and a. k. kerman , phys .
a * 285 * , 228 ( 2001 ) .
m. holland , s. j. j. m. f. kokkelmans , m. l. chiofalo , and r. walser , phys .
lett . * 87 * , 120406 ( 2001 ) . c. chin , r. grimm , p. julienne , and e. tiesinga , rev .
phys . * 82 * , 1225 ( 2010 ) .
d. m. eagles , phys . rev . *
186 * , 456 ( 1969 ) .
a. j. leggett , _ modern trends in the theory of condensed matter _
( springer , berlin , 1960 ) .
p. nozires and s. schmitt - rink , j. low .
phys . * 59 * , 195 ( 1985 ) . c. a. r. s de melo , m. randeria , and j. r. engelbrecht , phys .
lett , * 71 * , 3202 ( 1993 ) .
m. randeria , in _ bose - einstein condensation _ , edited by a. griffin , d. w. snoke , and s. stringari ( cambridge university press , new york 1995 ) , p. 355 .
y. ohashi and a. griffin , phys .
lett . * 89 * , 130402 ( 2002 ) . c. a. regal , m. greiner , and d. s. jin , phys .
. lett . * 92 * , 040403 ( 2004 ) .
m. w. zwierlein , c. a. stan , c. h. schunck , s. m. f. raupach , a. j. kerman , and w. ketterle , phys .
lett . * 92 * , 120403 ( 2004 ) .
j. kinast , s. l. hemmer , m. e. gehm , a. turlapov , and j. e. thomas , phys .
* 92 * , 150402 ( 2004 ) .
m. bartenstein , a. altmeyer , s. riedl , s. jochim , c. chin , j. hecker denschlag , and r. grimm , phys .
lett . * 92 * , 203201 ( 2004 ) .
a. damascelli , z. hussain , and z. shen , rev .
mod . phys . * 75 * , 473 ( 2003 ) .
y. yanase , t. jujo , t. nomura , h. ikeda , t. hotta , and k. yamada , phys .
rep . * 387 * , 1 ( 2003 ) .
p. lee , n. nagaosa , and x. wen , rev .
mod . phys . * 78 * , 17 ( 2006 ) .
o. fischer , m. kugler , i. maggio - aprile , c. berthod , and c. renner , rev .
phys . * 79 * , 353 ( 2007 ) .
m. feld , b. frhlich , e. vogt , m. koschorreck , and m. khl , _ nature _ * 480 * , 75 ( 2011 ) .
b. frhlich , m. feld , e. vogt , m. koschorreck , w. zwerger , and m. khl , phys .
* 106 * , 105301 ( 2011 ) .
a. t. sommer , l. w. cheuk , m. j. h. ku , w. s. bakr , and m. w. zwierlein , phys .
lett . * 108 * , 045302 ( 2012 ) .
j. t. stewart , c. a. regal , and d. s. jin , nature ( london ) * 454 * , 744 ( 2008 ) .
j. p. gaeber , j. t. stewart , t. e. drake , d. s. jin , a. perali , p. pieri , and g. c. strinati , nat .
* 6 * , 569 ( 2010 ) .
q. chen , and k. levin , phys .
* 102 * , 190402 ( 2009 ) .
s. tsuchiya , r. watanabe , and y. ohashi , phys .
a * 80 * , 033613 ( 2009 ) .
s. tsuchiya , r. watanabe , and y. ohashi , phys .
a * 82 * , 033629 ( 2010 ) .
s. tsuchiya , r. watanabe , and y. ohashi , phys .
a * 84 * , 043647 ( 2011 ) .
r. watanabe , s. tsuchiya , and y. ohashi , phys .
a * 82 * , 043630 ( 2010 ) ; * 85 * , 039908(e ) ( 2012 ) .
r. watanabe , s. tsuchiya , and y. ohashi , phys .
a * 86 * , 063603 ( 2012 ) . h. hu , x .-
liu , p. d. drummond , and h. dong , phys .
* 104 * , 240407 ( 2010 ) .
p. magierski , g. wlazowski , and a. bulgac , phys .
* 107 * , 145304 ( 2011 ) .
g. wlazowski , p. magierski , j. e. drut , a. bulgac , and k. j. roche , phys .
lett . * 110 * , 090401 ( 2013 ) .
s. nascimbne , n. navon , l. j. jiang , f. chevy and c. salomon , nature * 463 * , 1057 ( 2010 ) .
a. perali , p. pieri , g. c. strinati , and c. castellani , phys .
b * 66 * , 024510 ( 2002 ) . c. j. pethick , and h. smith , _ bose - einstein condensation in dilute gases _ ( cambridge , uk , 2008 ) chap.2 .
v. l. berezinskii , sov .
phys . -jetp * 32 * , 493 ( 1971 ) .
j. m. kosterlitz , and d. j. thouless , j. phys .
c : solid state phys .
* 6 * , 1181 ( 1973 ) . j. m. kosterlitz , j. phys .
c : solid state phys .
* 7 * , 1046 ( 1974 ) . m. randeria , j. -m .
duan , and l. -y .
shien , phys .
rev . lett . * 62 * , 981 ( 1989 ) .
s. s. botelho , and c. a. r. s de melo , phys . rev .
lett . * 96 * , 040404 ( 2006 ) .
m. iskin , and c. a. r. s de melo , phys .
lett . * 103 * , 165301 ( 2009 ) . s. n. klimin , j. tempere , and j. t. devreese , new j. phys . * 14 * , 103044 ( 2012 ) .
w. zhang , g .- d .
lin , and l .- m .
duan , phys .
a * 78 * , 043617 ( 2008 ) .
v. pietil , phys .
a * 86 * , 023608 ( 2012 ) .
n. d. mermin , and h. wagner , phys .
* 17 * , 1133 ( 1966 ) .
p. c. hohenberg , phys
. rev . * 158 * , 383 ( 1967 ) .
s. a. morgan , m. d. lee , and k. burnett , phys .
a * 65 * , 022706 ( 2002 ) .
a. tokumitu , k. miyake , and k. yamada , phys .
b * 47 * , 11988 ( 1993 ) .
d. j. thouless , ann . phys . * 10 * , 553 ( 1960 ) . as shown in fig.[fig1](c ) , the lda particle density at @xmath63 diverges at @xmath3 , so that we can not calculate @xmath126 at this temperature . to avoid this singularity ,
we have calculated the local density of states slightly away from the trap center . when @xmath127 , such a singularity is absent ( see fig.[fig1](c ) .
) , so that we can calculate @xmath68 everywhere in the gas cloud .
|
we investigate strong - coupling properties of a trapped two - dimensional normal fermi gas . within the framework of a combined @xmath0-matrix theory with the local density approximation
, we calculate the local density of states , as well as the photoemission spectrum , to see how two - dimensional pairing fluctuations affect these single - particle quantities . in the bcs ( bardeen - cooper - schrieffer)-bec ( bose - einstein condensation ) crossover region , we show that the local density of states exhibits a dip structure in the trap center , which is more remarkable than the three - dimensional case .
this pseudogap phenomenon is found to naturally lead to a double peak structure in the photoemission spectrum .
the peak - to - peak energy of the spectrum at @xmath1 agrees well with the recent experiment on a two - dimensional @xmath2 fermi gas [ m. feld , _
et al_. , nature * 480 * , 75 ( 2011 ) ] .
since pairing fluctuations are sensitive to the dimensionality of a system , our results would be useful for the study of many - body physics in the bcs - bec crossover regime of a two - dimensional fermi gas .
|
clusters of galaxies are continuously growing through the accretion of galaxies and groups from the field . the star formation activity in the accreting galaxies must be quenched during the assimilation of the galaxies into the cluster .
this transformation is a key process in creating the environmental dependence of galaxy properties , and may also underpin the observed evolution of galaxy properties in distant clusters [ e.g. 1.2 ] however , the physical mechanism which is responsible for these changes has not yet been identified [ 3,4,5 ] . the advent of suprime - cam , a revolutionary wide - field camera on the subaru telescope , has opened a new window in this field .
its 27@xmath1 field of view can trace the variation of galaxy properties from the cluster cores out to the surrounding field in an attempt to identify the environment where the decline in the star formation in accreted galaxies begins .
as a first step towards a systematic study of distant clusters with subaru and suprime - cam , we obtained deep ( @xmath7 ) @xmath8 imaging of the rich cluster a851 at z=0.4 .
we constructed an i - band selected sample which contains 15,055 galaxies brighter than @xmath9 . in order to assign cluster membership across our large field and to faint magnitudes
, we applied photometric redshift technique [ 6 ] to isolate the galaxies in a narrow redshift slice around the cluster redshift . by applying the redshift cut of @xmath10 , we can reduce the field contamination by factor 10 at @xmath9 , while keeping most of the cluster members ( 80% , from the estimate using the spectroscopic members [ 7 ] ) . after this photometric redshift selection ,
it is straightforward to map out the distribution of galaxies which are likely to be associated with the cluster . as shown in fig . 1 , several large scale structures are visible around the cluster core , comprising many clumps and filamentary extensions coming out directly from the core .
importantly , most of these extensions from the core are aligned with the surrounding subclumps .
the structures identified in this region are qualitatively similar to those seen in cosmological simulations of the growth of clusters which exhibit the filamentary / clumpy substructures on similar scales [ e.g. 8 ] it appears therefore that we are witnessing a851 as it assembles through the accretion of galaxies and groups along the filaments onto the cluster core from the surrounding field . by exploiting the striking large scale structure around this cluster , we can investigate the influence of environment on the photometric properties of galaxies .
we define the environment for each galaxy using the local surface number density , @xmath11 , of members , calculated from the 10 nearest neighbors .
we statistically subtract residual field contamination using a subaru blank field data [ 9 ] .
we divide the density distribution into three regimes as indicated in fig .
there is a close correspondence between the local density and structure : the high density region corresponds to the cluster core within @xmath12mpc ; the medium density region includes the structures defining the filaments surrounding the cluster ; and the low density region comprising the rest of the volume ( fig . 1 ) . as shown in fig . 2a
, the color distribution in the high density region is strongly peaked at @xmath13 , the color of an early - type cluster member . however , as we move to lower densities , the distribution becomes dramatically bluer .
notably , this color transition with local density occurs quite abruptly at @xmath14 , indicating a threshold effect in transforming galaxy properties .
the boundary corresponding to this critical density is highlighted in fig . 1 .
surprisingly , this transition occurs in subclumps well outside the core .
our ability to pin - point this environment is a fundamental step towards identifying the dominant mechanism behind the environmental dependence of galaxy properties .
since the ram - pressure stripping is largely supressed in groups [ 4 ] , the effective mechanism at this hierarchy is either galaxy galaxy collisions ( which cause cold disk gas to be driven to the galaxy center creating a star burst [ 3 ] ) ; and ` suffocation ' ( where _ warm _ gas in the galaxy s halo is shock heated by the intra - cluster medium so that it can no longer cool and replenish the cold gas in the disk [ 10 ] ) .
our data do not distinguish directly between these possibilities , but they can be distinguished by tracing the variation in the individual components of galaxies , bulges and disks , between the cluster and the field .
it is important to note that the changes in galaxy properties as a function of local density are most prominent in faint galaxies ( @xmath5 ) .
this is mirrored to the dramatic change in the shape of the luminosity function with local galaxy density : where we see a steepening of the faint end slope of the luminosity function with decreasing density ( fig .
2c ) . assuming that their star formation effectively ceases , the relative absence of low - luminosity galaxies at projected densities of @xmath15 can be understood as they will fade by @xmath16mag as their star formation declines [ 2 ]
this will put many of their descendents below our magnitude limit .
this is also qualitatively consistent with the presence of large numbers of very low luminosity , @xmath17 , passive dwarf galaxies in local clusters [ 11 ] we are grateful to dr y. komiyama for his assistance during our observations .
we acknowledge the suprime - cam team for allowing us to use the blank field data .
we also thank drs .
m.balogh , n. arimoto and k. shimasaku for helpful discussion .
tk and fn acknowledge the japan society for the promotion of science for support through its research fellowships for young scientists .
irs acknowledges support from the royal society and the leverhulme trust .
butcher , h. , oemler , a. , 1984 , apj , 285 , 426 kodama , t. , bower , r. g. , 2001 , mnras , 321 , 18 moore , b. , katz , n. , lake , g. , dressler , a. , oemler , a. , 1996 , nature , 379 , 613 adabi , m. g. , bower , r. g. , navarro , j. f. , 2000 , mnras , 314 , 759 balogh , m. l. , navarro , j. f. , morris , s. l. , 2000 , apj , 540 , 113 kodama , t. , bell , e. f. , bower , r. g. , 1999 , mnras , 302 , 152 dressler , a. , smail , i. , poggianti , b. m. , butcher , h. , couch , w. j. , ellis , r. s. , oemler , a. , 1999 , apjs , 122 , 51 ghigna , s. , moore , b. , governato , f. , lake , g. , quinn , t. , stadel , j. , 1998 , mnras , 300 , 146 ouchi , m. , et al .
, 2001 , apj , 558 , l83 larson , r. b. , tinsley , b. m. , caldwell , c. n. , 1980 , apj , 237 , 692 binggeli , b. , sandage , a. , tammann , g. a. , 1988 , ara&a , 26 , 509
|
we present deep , panoramic multi - color imaging of the distant rich cluster a851 ( @xmath0 ) using suprime - cam on subaru .
these images cover a 27@xmath1 field of view ( @xmath2mpc ) , and by exploiting photometric redshifts , we can isolate galaxies in a narrow redshift slice at the cluster redshift . using a sample of @xmath3 probable cluster members ( @xmath4 ) , we trace the network of filaments and subclumps around the cluster core .
the depth of our observations , combined with the identification of filamentary structure , gives us an unprecedented opportunity to test the influence of the environment on the properties of low luminosity galaxies .
we find an abrupt change in the colors of faint galaxies ( @xmath5 ) at a local density of 100 gal.mpc@xmath6 .
the transition in the color - local density behavior occurs at densities corresponding to subclumps within the filaments surrounding the cluster . identifying the sites where the transition occurs brings us much closer to understanding the mechanisms which are responsible for establishing the present - day relationship between environment and galaxy characteristics . 1.25 in .125 in .25 in
|
large - scale biological network data are increasingly becoming available due to advances in experimental biology .
we analyze protein - protein interaction ( ppi ) networks , where proteins are modeled as network nodes and interactions amongst them as network edges .
since it is the proteins that carry out almost all biological processes and they do so by interacting with other proteins , analyzing ppi network structure could lead to new knowledge about complex biological mechanisms and disease .
additionally , we analyze network representations of protein structures , `` residue interaction graphs '' ( rigs ) , where residues are modeled as network nodes and inter - residue interactions as network edges ; an inter - residue interaction exists between residues that are in close spatial proximity .
understanding rigs might provide deeper insights into protein structure , binding , and folding mechanisms , as well as into protein stability and function . to understand these complex biological network data
, one must be able to successfully reproduce them .
finding an adequate network model that will generate networks that closely replicate the structure of real data is one of the first steps in this direction .
thus , we focus on finding well - fitting network models for biological networks .
the hope is that a good network model could provide insights into understanding of biological function , disease , and evolution .
for example , the use of an adequate network model is vital for discovering network motifs , evolutionary conserved functional modules @xcite , since network motif discovery requires comparing real - world networks with randomized ones @xcite .
a well - fitting network model could also be used to assign confidence levels to existing protein interactions , as well as to predict new interactions that were overlooked experimentally @xcite .
additionally , it could guide biological experiments in a time- and cost - optimal way , thus minimizing the costs of interactome detection @xcite . since discovering ppi and other biological networks
is in its infancy , it is expected that practical application of network models will increase and prove its value in the future .
several network models have been proposed for biological networks . starting with erds - rnyi random graphs @xcite
, various network models have been designed to match certain properties of real - world networks .
early studies published largely incomplete yeast two - hybrid ppi data sets @xcite that were well modeled by scale - free networks @xcite . in a scale - free network
, the distribution of degrees follows a power - law @xcite .
modeling of the data by scale - free networks was based on the assumption that the degree distribution is one of the most important network parameters that a good network model should capture .
however , networks of vastly different structures could have the same degree distributions @xcite .
additionally , it has been argued that currently available ppi network data are samples of the full interactomes and thus the observed power - law degree distributions are artifacts of sampling properties of these networks @xcite . as new biological network data becomes available , we need to ensure that our models continue to fit the data well . in the light of new ppi network data ,
several studies have started questioning the wellness of fit of scale - free models : an evidence has been presented that the structure of ppi networks is closer to geometric random graphs , that model spatial relationships between objects , than to scale - free networks @xcite .
similarly , geometric random graph model has been identified as an optimal network model for rigs @xcite .
a well - fitting network model should generate graphs that closely resemble the structure of real - world networks . to evaluate the fit of a model to the data , one needs to compare model networks with real - world networks .
however , network comparisons are computationally infeasible due to np - completeness of the underlying subgraph isomorphism problem @xcite .
therefore , large network comparisons rely on heuristics , commonly called `` network properties . ''
these properties belong to two major classes : global and local .
global properties include the degree distribution , average clustering coefficient , clustering spectrum , average diameter , and the spectrum of shortest path lengths .
local properties include network motifs , small overrepresented subgraphs @xcite , and graphlets , small connected induced subgraphs of real - world networks ( figure [ fig : heur ] ( a ) ) @xcite .
based on graphlets , two highly sensitive measures of network local structural similarities were designed : the relative graphlet frequency distance ( `` rgf - distance '' ) @xcite and graphlet - based generalization of the degree distribution , called graphlet degree distribution agreement ( `` gdd - agreement '' ) @xcite .
the choice of a network property for evaluating the fit of a network model to the data is non - trivial , since different models might be identified as optimal with respect to different properties . in general , global properties
might not be constraining enough to capture complex topological characteristics of biological networks .
for example , two networks with exactly the same degree distributions can have completely different underlying topologies ( figure [ fig : heur ] ( b ) ) . on the other hand , local properties , rgf - distance and gdd - agreement ,
impose a larger number of constraints , thus reducing degrees of freedom in which networks being compared can differ .
the fit of model networks to real - world data can also be evaluated by using principal component analysis of the vector space whose coordinates are the statistics of network properties @xcite , as well as by counting the number of random walks of a given length in the network and feeding these counts into a probabilistic method @xcite . [
cols="^,^ " , ] filkov _ et al .
_ use seven network properties to describe a real - world network and compare it with model networks @xcite . in comparison
, we use 34 properties , therefore decreasing the number of degrees of freedom in which networks being compared can vary . also , the methodology used by filkov _ et al . _
is different than ours .
first , they evaluate the fit of a model to the data by using principal component analysis of the vector space whose coordinates are the statistics of the seven network properties that they analyzed .
second , they evaluate the fit of two scale - free network models to the data : sf and their new scale - free model of network growth via sequential attachment of linked node groups . in comparison
, we use three network models that have very different network structure : er , sf , and geo . middendorf _ et al .
_ measure the topological structure of a network by counting the number of random walks of a given length @xcite and giving those counts as input into classifiers .
random walks are different than graphlets in several ways .
first , graphlets are induced and random walks are not .
second , nodes and edges can be repeated in a random walk , while a graphlet consists of a unique set of nodes and edges .
use two classifiers , svm @xcite and dt @xcite to discriminate different network models .
that is , in each study , they use a single probabilistic method to predict the best fitting network model for a real - world network . in comparison ,
in this study we use five different probabilistic methods , all supporting geo as the best - fitting model .
we show that dt and svm are the least robust out of the five probabilistic methods that we analyzed ( see section `` robustness and validation '' and table 3 ) .
moreover , the training set of middendorf _ et al .
_ contains model networks of the size of the data only and thus it could be biased by the model properties that are enforced by the chosen network size . in comparison , we train our probabilistic models on random networks of different sizes ( see sections `` techniques '' and `` methods '' ) to allow for a possibility to predict the best fitting network model for any yet unseen real - world network , independent of its size .
_ consider sf , er , and small - world networks and identify sf - based duplication - mutation models as the best - fitting models for biological networks .
given that middendorf _
_ did not consider geo in their studies , and given a low robustness of dt and svm that they used , their reported best fit of sf - based models to the data could be questioned .
we further elaborate on the power of integration of different network properties as opposed to using individual ones to asses the fit of a network model to the data .
we use our graphcrunch software package @xcite to evaluate the fit of er , geo , and sf models to all ppi networks described in data sets section .
graphcrunch evaluates the fit of the models to the data with respect to seven local and global network properties .
when we evaluate the fit of the data to the models with respect to each of the seven properties , we obtain inconclusive results , because each of the properties favors a different model .
for example , as illustrated in figure [ fig : properties ] ( a ) , sf fits yh1 the best with respect to the degree distribution , but geo is the best - fitting network model with respect to the clustering spectrum ( figure [ fig : properties ] ( b ) ) .
this demonstrates the need for a method that finds a consensus between models suggested by different network properties .
we propose such a method in this study .
since our method integrates a variety of network properties , it imposes a large number of constraints on the networks being compared and reduces the number of degrees of freedom in which they can differ , thus increasing the confidence in the fit of a network model .
inclusion of additional network properties could further increase the confidence at the expense of an increased computational complexity .
* ( a ) * * ( b ) * * ( c ) * * ( d ) * although several studies proposed geo as a well - fitting null model for ppi networks @xcite , a recent study questioned this @xcite .
note however , that this conclusion was based on analyzing only one eukaryotic and one prokaryotic ppi network @xcite , each from dip @xcite .
thus , in the light of low quality and incompleteness of the data from large databases @xcite ( also see results section ) , no conclusions about the fit of geo should have been made .
the authors argued that low - dimensional geometric random graphs might not be able to capture high abundance of dense graphlets and bipartite subgraphs observed in real - world networks , neglecting two obvious alternatives for reconciling the differences in the abundance of subgraphs in the data and in geo : ( 1 ) they based their conclusion on the observation that bipartite graphlet 20 can not exist in 2-dimensional ( 2d ) geo , even though it exists in 3d geo as well as in all higher dimensions ( figure [ fig : properties ] ( c ) ) ; and ( 2 ) geo graphs with the same number of nodes , but @xmath0 times more edges than the data have very similar abundance of all graphlets as the data ( figure [ fig : properties ] ( d ) ) .
these observations are important since : ( 1 ) the optimal dimension for the space of ppi networks is unknown and finding it is a non - trivial research problem , but it is highly unlikely that ppi networks exist in a 2d space ; and ( 2 ) the density of real - world data will continue to increase @xcite , most likely in accordance with its network model . also note that over - abundance of graphlets in the currently available ppi networks could be an artefact of the `` matrix '' and `` spoke '' models used to determine ppis in affinity purification followed by mass spectrometry ( ap / ms ) pull - down experiments . in the matrix model , interactions are defined between all proteins in a purified complex , clearly resulting in over - abundance of dense graphlets . in the spoke model
, interactions are defined between a bait and each of its preys , but not between the preys , clearly resulting in over - abundance of sparse graphlets ; an overlap between preys in different purifications results in over - abundance of complete bipartite graphs . finally , it is possible that ppi networks are not completely geometric and that another random graph model would provide a better fit . additionally , it is possible that different parts of ppi networks have different structure .
the two most commonly used high - throughput ppi detection methods , ap / ms and y2h , are fundamentally different : ap / ms detects mostly stable protein complexes , whereas y2h detects mostly transient signalling interactions @xcite .
thus , the two methods examine different , complementary subspaces within the interactome , resulting in networks with different topological and biological properties @xcite . since proteins within a protein complex are close in the cell , it is possible that protein complexes have a geometric structure .
in contrast , transient interactions in signalling pathways might have a different structure , such as that of bipartite graphs or scale - free networks .
we present an integrative approach for identifying the best - fitting network model for rigs and ppi networks .
we use five probabilistic methods to evaluate the fit of a network model to a real - world network with respect to a series of network properties .
all five probabilistic methods confirm that geo is the best - fitting model for rigs .
ppi data sets of high confidence and coverage are also fitted the best by geo , while sparser and lower - confidence ppi networks are fitted the best by sf or er . by testing the robustness of probabilistic methods to noise ,
we demonstrate that our approach is unlikely to predict a real - world network as geo if it had a noisy sf structure , independent of the probabilistic method or noise level . on the other hand
, it could classify a real - world network as sf or er if it had a noisy geo structure , depending on the probabilistic method and noise level .
together , these results suggest that the structure of ppi networks is the most consistent with the structure of noisy geo . since networks have been used to model real - world phenomena in various domains , it would be interesting to apply our method to other types of real - world networks , such as technological or social ones .
we summarize the structure of a complex network by the notion of the `` network fingerprint '' ( of just `` fingerprint , '' for brevity ) .
we define the _ fingerprint _ to be a 34-dimensional vector whose coordinates contain the following network properties : the average degree , average clustering coefficient , average diameter , and frequencies of the appearance of all 31 1 - 5-node graphlets .
the _ degree _ of a node is the number of edges incident to the node ; the _ average degree _ of a network is the average of degrees over all nodes in the network .
the _ clustering coefficient _ of a node is defined as the probability that two neighbors of the node are themselves connected .
the average of clustering coefficients over all nodes in a network is the _ average clustering coefficient _ of the network .
the smallest number of links that have to be traversed in a network to get from one node to another is called the _ distance _ between the two nodes and a path through the network that achieves this distance is called the _ shortest path _ between the nodes ; the average of shortest path lengths over all pairs of nodes in a network is called the _ average network diameter_. _ graphlets _ are small connected non - isomorphic induced subgraphs of a large network @xcite ; we count the occurrences of the only 1-node graphlet , a node , the only 2-node graphlet , an edge , and all 29 3 - 5-node graphlets ( shown in figure 1(a ) ) . because different coordinates of a network fingerprint can differ by several orders of magnitude , we normalize each coordinate to avoid domination of coordinates having large values .
we normalize the @xmath1 coordinate @xmath2 of the network fingerprint _ x _ as @xmath3 , for @xmath4 ; we add 1 to @xmath5 to avoid the logarithm function to go to infinity when @xmath6 .
we consider three random network models : erds - rnyi ( er ) random graphs @xcite , scale - free barabsi - albert ( sf ) networks @xcite , and geometric ( geo ) random graphs @xcite . in
_ erds - rnyi random graphs _
, edges between pairs of nodes are distributed uniformly at random with the same probability @xmath7 @xcite .
_ scale - free networks _
are networks that have power - law degree distributions .
the version of sf networks that we use are generated by barabsi - albert peripheral attachment method @xcite , in which newly added nodes preferentially attach to existing nodes with probabilities proportional to their degrees . in _ geometric random graphs _ , nodes correspond to uniformly distributed points in a metric space and edges are created between pairs of nodes if the corresponding points are close enough in the metric space according to some distance norm @xcite .
we construct geometric random graphs by using 3-dimensional euclidean boxes and the euclidean distance norm @xcite . for each of the three random network models , we generate 10 instances of random networks per model .
we generate random networks of different sizes , both in terms of the number of nodes ( @xmath8 ) and the number of edges ( @xmath9 ) .
we use the following @xmath10 values for @xmath8 : 100 , 200 , 300 , 400 , 500 , 600 , 700 , 800 , 900 , 1,000 , 1,100 , 1,200 , 1,300 , 1,400 , 1,600 , 2,100 , 2,600 , 3,100 , 3,600 , 4,100 , 4,600 , 5,000 , 6,000 , 7,000 , 8,000 , 9,000 , 10,000 , and 11,000 . for each of the @xmath11 values of @xmath8 below 10,000 ,
we vary @xmath12 from 1 to 10 , in increments of 1 . due to the increase in computational complexity with the increase in the number of nodes and edges , for the @xmath13 largest values of @xmath8 , @xmath14 and @xmath15
, we only use @xmath16 .
thus , we analyze @xmath17 different network sizes . in total , for the 3 network models , 10 random network instances per model , and 274 network sizes , we create @xmath18 model networks .
we use five well - known probabilistic methods : backpropagation method ( bp ) , probabilistic neural networks ( pnn ) , decision tree ( dt ) , multinomial nave bayes classifier ( mnb ) , and support vector machine ( svm ) .
both * bp * @xcite and * pnn * @xcite are based on artificial neural networks ( anns ) .
anns are simplified mathematical models of biological nervous systems built of processing units called neurons .
neurons in anns have many input signals and they produce one output signal .
they are organized into the following layers : the input layer , one or more hidden layers , and the output layer .
neurons in the input layer do not perform any processing ; instead , they only distribute the input data to all neurons in the first hidden layer .
the number of hidden layers depends on implementation of an ann .
we use the standard implementations of bp and pnn from neural - network toolbox in matlab . for the completeness of the manuscript ,
we briefly outline them below . in our implementation of * bp *
@xcite , the input layer consists of 34 neurons corresponding to the 34 coordinates of the network fingerprint input vector . to match the length of our input vector , we implement one hidden layer with 15 neurons ;
varying the number of neurons in the hidden layer between 10 and 20 had marginal effect on the results .
the output layer contains three neurons , according to the `` 1-of - n encoding of the output classes '' principle @xcite : the number of neurons in the output layer matches the number of possible `` output classes , '' i.e. , random network models ( er , sf , and geo ) .
thus , for a given output class , the neuron corresponding to the class is set to 1 , whereas the remaining two neurons are set to -1 .
after bp computes the values on the three output neurons for an input vector , it classifies the input into an output class that corresponds to the neuron with the largest value .
all neurons in the input layer are connected with all neurons in the hidden layer .
similarly , all neurons in the hidden layer are connected with all neurons in the output layer .
all of these connections are weighted .
each neuron in the hidden and the output layer produces output by applying a non - linear `` transfer function '' to calculate a weighted sum of its inputs .
we use _ logsig _ and _ tansig _ transfer functions in the hidden layer and the output layer , respectively .
initially , all weights are assigned randomly .
weights are adjusted gradually trough a training ( learning ) process : bp keeps adjusting the weights until the error between the value of each output neuron and its desired value ( i.e. , the value of the class that the input that we are training bp on belongs to ) is @xmath19 .
we use _ trainscg _ `` training function '' and set the learning rate to 0.01 . given these parameters , bp is successfully trained on the training set ( defined in techniques section above ) in 588 epochs . * pnn * that we use consists of the radial basis layer and the competitive layer .
the radial basis layer further consists of the input and pattern sublayers .
similarly , the competitive layer consists of the summation and the output sublayers .
the number of neurons in the input sublayer corresponds to the 34 dimensions of the input vector .
the pattern sublayer consists of three pools of `` pattern '' neurons , where each pool corresponds to one of the three output classes .
the number of neurons in each pool is determined as follows . as each network fingerprint from the training set
is provided as input vector into pnn during the training process , a new neuron is added to the pool that corresponds to the output class ( i.e. , network model ) of the input vector .
after the training phase , when an input vector is presented to the trained pnn , the pattern sublayer computes how close the input vector is to each of the vectors from the training set in each pool .
this information is sent to the summation sublayer .
the summation sublayer consists of three neurons , where each neuron corresponds to one of the three output classes .
input into each neuron in the summation sublayer is the collection of outputs from the corresponding pool in the pattern sublayer .
the output of each summation sublayer neuron is a weighted sum of all its inputs .
each of the three sums represents the probability that the input vector belongs to the corresponding class .
given these probabilities , the output sublayer , consisting of a single neuron , outputs the class having the highest probability .
we use a standard implementation of * dt * @xcite from statistics toolbox in matlab .
interior nodes in the decision tree are queries on certain attributes ; in our case , attributes are the coordinates of the fingerprint vector .
each leaf in the tree corresponds to one of the three output classes .
branches in the tree represent conjunctions of attributes that lead to classification into the output classes .
dt recursively splits the training set of input vectors into subsets based on the values of their coordinates ; this corresponds to branching in the tree .
dt continues to do so until the training input vectors are assigned to their correct classes .
we use a standard implementation of * mnb * @xcite from weka @xcite , a publicly available collection of machine learning algorithms for data mining .
mnb classifies the input data based on the bayes rule by selecting a class that maximizes the posterior probability of the class , given the training set .
mnb does not use the assumption of a nave bayes classifier , that all data attributes are independent of each other .
we use a standard implementation of * svm * @xcite from weka .
svm maps our 34-dimensional input vectors into a high dimensional space ; the space dimension is automatically determined by weka . during the training phase
, svm finds an optimized data division within this space by constructing a hyperplane that optimally separates the data into two classes ; since there are many hyperplanes that might classify the data , the hyperplane is chosen so that the distance from the hyperplane to the nearest data point is maximized .
we generalize this binary classification to the multiclass classification , with three classes corresponding to the three random network models .
we do so by using three binary `` one - versus - all '' svms : for each of the three classes , its corresponding svm either classifies the input data as belonging to the class ( `` positive classification '' ) , or not belonging to the class ( `` negative classification '' ) @xcite . each of these three binary svms produces an output function that gives a relatively large value for a positive classification and a relatively small value for a negative classification .
the input data is classified into the class with the highest value of the output function .
vm participated in the design of the study , carried out most of the experiments , analyzed the results , and helped write the paper .
tm participated in the design of the study , helped run the experiments , analyzed the results , and wrote the paper .
np conceived of the study and participated in its design and coordination and helped write the manuscript .
all authors read and approved the final manuscript .
this project was supported by the nsf career iis-0644424 grant .
ito t , tashiro k , muta s , ozawa r , chiba t , nishizawa m , yamamoto k , kuhara s , sakaki y : * toward a protein - protein interaction map of the budding yeast : a comprehensive system to examine two - hybrid interactions in all possible combinations between the yeast proteins*. _ proc natl acad sci u s a _ 2000 , * 97*(3):11437 .
uetz p , giot l , cagney g , mansfield ta , judson rs , knight jr , lockshon e , narayan v , srinivasan m , pochart p , qureshi - emili a , li y , godwin b , conover d , kalbfleish t , vijayadamodar g , yang m , johnston m , fields s , rothberg jm : * a comprehensive analysis of protein - protein interactions in saccharomyces cerevisiae*. _ nature _ 2000 , * 403*:623627 .
collins s , kemmeren p , zhao x , greenblatt j , spencer f , holstege f , weissman j , krogan n : * toward a comprehensive atlas of the physical interactome of saccharomyces cerevisiae*. _ molecular and cellular proteomics _ 2007 , * 6*(3):439450 .
breitkreutz bj , stark c , , reguly t , boucher l , breitkreutz a , livstone m , oughtred r , lackner dh , bahler j , wood v , dolinski k , tyers m : * the biogrid interaction database : 2008 update*. _ nucleic acids research _ 2008 , * 36*:d637d640 .
giot l , bader j , brouwer c , chaudhuri a , kuang b , li y , hao y , ooi c , godwin b , vitols e , vijayadamodar g , pochart p , machineni h , welsh m , kong y , zerhusen b , malcolm r , varrone z , collis a , minto m , burgess s , mcdaniel l , stimpson e , spriggs f , williams j , neurath k , ioime n , agee m , voss e , furtak k , renzulli r , aanensen n , carrolla s , bickelhaupt e , lazovatsky y , dasilva a , zhong j , stanyon c , finley rj , white k , braverman m , jarvie t , gold s , leach m , knight j , shimkets r , mckenna m , chant j , rothberg j : * a protein interaction map of drosophila melanogaster*. _ science _ 2003 , * 302*(5651):17271736 .
rual j , venkatesan k , hao t , hirozane - kishikawa t , dricot a , li n , berriz gf , gibbons fd , dreze m , ayivi - guedehoussou n , klitgord n , simon c , boxem m , milstein s , rosenberg j , goldberg ds , zhang lv , wong sl , franklin g , li s , albala js , lim j , fraughton c , llamosas e , cevik s , bex c , lamesch p , sikorski rs , vandenhaute j , zoghbi hy , smolyar a , bosak s , sequerra r , doucette - stamm l , cusick me , hill de , roth fp , vidal m : * towards a proteome - scale map of the human protein - protein interaction network*. _ nature _ 2005 , * 437*:117378 .
peri s , navarro jd , kristiansen tz , amanchy r , surendranath v , muthusamy b , gandhi tk , chandrika kn , deshpande n , suresh s , rashmi bp , shanker k , padma n , n iranjan v , harsha hc , talreja n , vrushabendra bm , ramya ma , yatish aj , joy m , s hivashankar hn , kavitha mp , menezes m , choudhury dr , ghosh n , saravana r , chandran s , mohan s , jonnalagadda ck , prasad ck , kumar - sinha c , deshpande ks , pandey a : * human protein reference database as a discovery resource for proteomics*. _ nucleic acids res _ 2004 , * 32 database issue*:d497501 .
[ 1362 - 4962 journal article ] .
middendorf m , ziv e , adams c , hom j , koytcheff r , levovitz c , woods g , chen l , wiggins c : * discriminative topological features reveal biological network mechanisms*. _ bmc bioinformatics _ 2004 , * 5:181*(doi:10.1186/1471 - 2105 - 5 - 181 ) .
|
* background : * networks are used to model real - world phenomena in various domains , including systems biology . since proteins carry out biological processes by interacting with other proteins , it is expected that cellular functions are reflected in the structure of protein - protein interaction ( ppi ) networks .
similarly , the topology of residue interaction graphs ( rigs ) that model proteins 3-dimensional structure might provide insights into protein folding , stability , and function . an important step towards understanding these networks is finding an adequate network model . evaluating
the fit of a model network to the data is a formidable challenge , since network comparisons are computationally infeasible and thus rely on heuristics , or `` network properties . ''
* results : * we show that it is difficult to assess the reliability of the fit of a model using any network property alone .
thus , we present an integrative approach that feeds a variety of network properties into five probabilistic methods to predict the best - fitting network model for ppi networks and rigs . we confirm that geometric random graphs ( geo ) are the best - fitting model for rigs . since geo networks model spatial relationships between objects and
are thus expected to replicate well the underlying structure of spatially packed residues in a protein , the good fit of geo to rigs validates our approach .
additionally , we apply our approach to ppi networks and confirm that the structure of merged data sets containing both binary and co - complex data that are of high coverage and confidence is also consistent with the structure of geo , while the structure of sparser and lower confidence data is not . since ppi data are noisy , we test the robustness of the five classifiers to noise and show that their robustness levels differ . *
conclusions : * we demonstrate that none of the classifiers predicts noisy scale - free ( sf ) networks as geo , whereas noisy geos can be classified as sf .
thus , it is unlikely that our approach would predict a real - world network as geo if it had a noisy sf structure .
however , it could classify the data as sf if it had a noisy geo structure .
therefore , the structure of the ppi networks is the most consistent with the structure of a noisy geo .
* keywords : * network modeling , biological networks , protein - protein interaction networks , residue interaction graphs
|
thanks to the efforts at the @xmath17 factories , the exploration of cp violation is now entering another exciting stage , allowing us to confront the kobayashi maskawa mechanism @xcite with data . after the discovery of mixing - induced cp violation in the `` gold - plated '' mode @xmath25 @xcite , as well as important other measurements , one of the most interesting questions is now to what extent the possible space for new physics ( np ) has already been reduced . in this context , the central target is the unitarity triangle of the cabibbo kobayashi
maskawa ( ckm ) matrix illustrated in fig .
[ fig : ut](a ) , where @xmath5 and @xmath6 are the generalized wolfenstein parameters @xcite .
the usual fits for the allowed region for the apex of the unitarity triangle in the @xmath5@xmath6 plane the `` ckm fits '' seem to indicate that no np is required to accommodate the data @xcite .
however , this is not the complete answer to this exciting question . in order to fully address it ,
the interpretation of the data on the mixing - induced cp asymmetry @xmath26 where @xmath7 is the cp - violating weak @xmath0@xmath1 mixing phase , requires a more involved analysis . within the standard model ( sm ) , @xmath7 is given by @xmath27 .
however , due to the possible impact of physics beyond the sm , @xmath7 takes the following general form : @xmath28 where @xmath29 is due to non - standard contributions to @xmath0@xmath1 mixing , which is the preferred mechanism for np to manifest itself in ( [ amix - bdpsiks ] ) . in principle , physics beyond the sm may also affect the @xmath30 decay amplitudes ; however , in this case the new contributions have to compete with sm tree - level amplitudes and their relative impact is expected to be much smaller .
so far , a set of `` smoking - gun '' observables to search for such kind of np does not indicate any deviation from the sm @xcite . obviously ,
because of the @xmath29 term in ( [ phid - np ] ) , we may not convert the experimental information on @xmath7 into a constraint on @xmath11 in the presence of np contributions to @xmath31@xmath32 mixing . moreover
, we may not use the sm interpretation of the @xmath33 mass differences @xmath34 to determine the unitarity - triangle side @xmath35 as recently discussed in @xcite . on the contrary
, the determination of the side @xmath36 by means of exclusive and inclusive transitions of the type @xmath37 and @xmath38 , which are dominated by sm tree - level amplitudes , is very robust as far as the impact of np is concerned @xcite .
a determination of the `` true '' unitarity triangle in the presence of a completely general np model , i.e. with arbitrary flavour - mixing terms , is almost impossible .
however , we may still perform useful predictive analyses within certain scenarios for physics beyond the sm , for example within models with `` minimal flavour violation '' ( mfv ) , where the only source for flavour mixing is still given by the ckm matrix ( see , for instance , @xcite ) . in the present paper we go one step beyond the mfv models : we analyse a scenario with large generic np contributions to @xmath0@xmath1 mixing ( @xmath39 ) , eventually also to @xmath40 ( @xmath41 ) , but _ not _ to the @xmath2 and @xmath42 decay amplitudes .
as we shall demonstrate , this scenario is well motivated by simple dimensional arguments in a wide class of models , including supersymmetric frameworks .
if we assume np of this kind , we may complement the experimental information on @xmath7 with data on cp violation in @xmath9 to extract @xmath12 , and may then fix the `` true '' apex of the unitarity triangle with the help of the side @xmath43 . needless to note , we may then also extract @xmath10 and @xmath11 .
several years ago , assuming also such a np scenario but neglecting penguin effects , it was already pointed out in @xcite that it is actually possible to determine the unitarity triangle up to discrete ambiguities by combining the cp - violating observables of @xmath25 and @xmath9 decays with @xmath43 .
since we have now strong experimental and theoretical indications for large penguin effects in @xmath9 , we must definitely care about them . to this end , we follow @xcite , and use data on the cp - averaged @xmath13 branching ratio , which allows us to control the penguin contributions with the help of plausible dynamical assumptions and @xmath44-spin flavour - symmetry arguments . employing `` qcd factorization '' to deal with the penguins topologies @xcite , a similar analysis was also performed in @xcite .
further alternative strategies to extract physics information from @xmath7 , @xmath43 , and cp violation in @xmath9 can be found , for instance , in @xcite .
another important aspect of our analysis are the very rare decays @xmath21 and @xmath20 . using the allowed ranges for the generalized wolfenstein parameters obtained from the strategy sketched above , we are able to predict the branching ratios for these modes independently of possible np contributions to @xmath0@xmath1 mixing . as we shall show , these predictions could be very different from the sm expectations .
= 7.0 truecm = 7.0 truecm the outline of this paper is as follows : a closer look at @xmath7 , @xmath43 and @xmath11 is presented in section [ sec : phid ] . in section [ sec : np ] , we motivate the scenario where np manifests itself only through contributions to @xmath0@xmath1 mixing . in section
[ sec : ut - det ] , we discuss how we may determine the `` true '' unitarity triangle for this kind of np with the help of cp violation in @xmath9 .
the implications for the rare decays @xmath21 and @xmath20 are presented in section [ sec : rare ] .
finally , we conclude in section [ sec : concl ] with a short summary of our results and a brief outlook .
the present world average for @xmath45 , determined from mixing - induced cp violation in @xmath25 and similar modes , is @xmath46 @xcite , which implies @xmath47 here the former solution would be in perfect agreement with the `` indirect '' range implied by the ckm fits , @xmath48 @xcite , whereas the latter would show a significant discrepancy .
measuring the sign of @xmath49 , both solutions can be distinguished .
there are several strategies on the market to accomplish this important task @xcite .
for example , in the @xmath30 system , @xmath50 can be extracted from the time - dependent angular distribution of the decay products of @xmath51 k^\ast[\to\pi^0k_{\rm s}]$ ] , if the sign of a hadronic parameter @xmath52 , involving a strong phase @xmath53 , is fixed through factorization @xcite .
this analysis is already in progress at the @xmath17 factories @xcite . in this context , it is important to note that the ckm factor @xmath43 introduced in ( [ rb - def ] ) allows us to obtain the following bounds ( see fig .
[ fig : ut]a ) : @xmath54 using the rather conservative experimental range @xmath55 which corresponds to @xmath56 , we obtain latexmath:[\[\label{beta - bound } @xmath43 from semileptonic @xmath17 decays , which originate from tree - level sm processes , appears to be very robust as far as the impact of np is concerned , we may consider ( [ beta - bound ] ) as an upper bound for the `` true '' angle @xmath11 of the unitarity triangle . consequently
, as we have @xmath58 within the sm , ( [ beta - bound ] ) implies latexmath:[\[\label{phid - sm - bound } former solution @xmath14 in ( [ phid - exp ] ) satisfies this bound nicely , this is definitely not the case for @xmath15 .
the latter solution can not be accommodated in the sm and requires np contributions to @xmath0@xmath1 mixing , i.e. we now need @xmath60 in ( [ phid - np ] ) . at this point ,
the important question of how we may represent this second solution in the @xmath5@xmath6 plane arises . in the literature , this is usually simply done through a second branch , corresponding to @xmath61 as we have shown in fig .
[ fig : ut](b ) .
however , because of ( [ phid - np ] ) , this is actually _ not _ correct .
moreover , since @xmath15 is associated with np contributions to @xmath0@xmath1 mixing , we may _ no longer _ use the sm interpretation of @xmath62 and @xmath63 to determine the side @xmath64 of the unitarity triangle .
in particular , we may no longer conclude from the experimental lower bound on @xmath65 that @xmath66 .
it is important to note that ( [ phid - np ] ) applies to @xmath67 as well , so that np may at least in principle hide itself in this case , despite the striking consistency with the sm interpretation of other observables .
as already pointed out , one of the key ingredients to determine the `` true '' apex of the unitarity triangle without using information on @xmath0@xmath1 mixing is the side @xmath43 .
if we complement it with a measurement of @xmath12 , we may determine the coordinates of the apex of the unitarity triangle straightforwardly through @xmath68from a theoretical point of view , certain pure `` tree '' decays would be best suited for the determination of @xmath12 , allowing theoretically clean extractions of this angle that would also be very robust under the impact of np ( see , for instance , @xcite).@xmath69 mixing .
however , it could then be taken into account through a measurement of the corresponding mixing parameters @xcite . ]
a similar comment applies to the new strategies that were recently proposed in @xcite .
here decays of the kind @xmath70 and @xmath71 are employed .
if we use the @xmath72@xmath73 mixing phase @xmath74 as an input , and observe the neutral @xmath75 mesons through their decays into cp - even and cp - odd eigenstates , certain `` untagged '' and mixing - induced observables allow a very efficient , theoretically clean determination of @xmath12 in an essentially unambiguous manner @xcite ; alternatively , we may also employ decays of the neutral @xmath75 mesons into cp non - eigenstates to this end @xcite .
these strategies appear to be particularly interesting for the next generation of dedicated @xmath17 experiments , lhcb and btev , as well as those at super-@xmath17 factories , although important steps may already be made at babar and belle .
since we can not yet confront these methods with data , we have to employ alternative strategies to extract @xmath12 . in this context
, @xmath9 offers a very interesting tool .
if we use @xmath7 as an input , and employ the cp - averaged @xmath13 branching ratio to control the penguin effects @xcite , we may determine @xmath12 from the cp - violating @xmath9 observables for each of the two solutions given in ( [ phid - exp ] ) @xcite ; using `` qcd factorization '' @xcite to deal with the penguin effects , such an exercise was also performed in @xcite . in the future , the @xmath9 observables can be fully exploited with the help of @xmath76 @xcite .
in contrast to the pure tree decays mentioned above , these modes may in principle well be affected by new physics , as they receive contributions from loop - induced flavour - changing neutral - current ( fcnc ) amplitudes .
however , as we shall discuss in the next section , there is a particularly interesting class of extensions of the sm where new physics yields sizeable contributions to @xmath0@xmath1 mixing only , leaving the decay amplitudes including the loop - induced ones almost unaffected .
as far as flavour physics is concerned , extensions of the sm can be classified into two wide categories : models with minimal flavour violation and models with new sources of flavour mixing . within the highly - constrained class of mfv models , the only source of flavour - symmetry breaking terms is given by the sm yukawa couplings @xcite .
as a consequence , all flavour - changing transitions are still ruled by the ckm matrix .
for this reason , many of the standard ckm constraints hold also in mfv models : the determination of @xmath64 in terms of @xmath63 is still valid @xcite , and the relation between @xmath45 and @xmath77 can differ at most by an overall sign @xcite , i.e. @xmath78 or @xmath79 .
it is then easy to realize that within mfv models the structure of the unitarity triangle is fixed up to a twofold ambiguity even without @xmath9 data , and that there is no room for the non - standard solution @xmath80 .
in the wide class of models with new sources of flavour mixing , it is rather natural to expect extra @xmath81 contributions to the @xmath82@xmath83 amplitude , with arbitrary phases , so that we may accommodate any value of @xmath29 .
it is also very natural to assume that these np effects have a negligible impact on @xmath2 amplitudes dominated by tree - level sm contributions ; this happens essentially in all realistic models . on the other hand
, it is less obvious why these new sources of flavour - symmetry breaking being able to induce @xmath81 corrections to @xmath82@xmath83 mixing should have a small impact on @xmath2 amplitudes arising only at the loop level within the sm .
this hypothesis , which is one of the main assumptions of the present analysis , certainly does not represent the most general np scenario .
however , as we shall discuss in the following , it can be realized under rather general conditions .
the generic np scenario we shall advocate is a model where non - standard effects are negligible in all amplitudes that receive tree - level sm contributions ( independently of possible ckm suppressions ) .
moreover , in order to protect the effects on @xmath2 fcnc amplitudes , we shall add the following two general requirements : 1 .
the effective scale of np is substantially higher than the electroweak scale ; 2 . the adimensional effective couplings ruling @xmath39 transitions can always be expressed as the square of two @xmath2 effective couplings . employing an effective - theory language
, what we mean under these two hypotheses is that the generic dimension - six operators encoding np contributions to @xmath82@xmath83 and @xmath84 transitions can be written as q^np_b=2 = ( d ) ( d ) , q^np_b=1 = ( d ) ( q ) , [ eq : npops ] where @xmath85 denotes the new @xmath2 effective flavour - changing coupling , @xmath86 indicates generic dirac and/or colour structures , and possible coefficient functions of @xmath81 have been ignored . coherently with the requirement ii ) , we shall also assume that @xmath87-breaking operators of dimension less than six , such as the chromomagnetic operator , play a negligible role .
the overall normalization of the two operators in ( [ eq : npops ] ) , or the definition of the effective scale @xmath88 , has been chosen such that the corresponding sm @xmath39 term is q^sm_b=2 = ( d ) ( d ) .
[ eq : normal2 ] choosing this normalization , we have implicitly factorized out an overall coefficient of @xmath89 $ ] in the effective hamiltonian , both in the @xmath39 and in the @xmath2 cases . as a result , within the sm
, the loop - induced @xmath2 operators can be written as q^sm_b=1 = c ( d ) ( q ) , [ eq : normal1 ] where the coefficient function @xmath90 is of @xmath81 in the case of pure short - distance - dominated electroweak operators ( such as those generated by @xmath91-penguin and @xmath92-box diagrams ) and substantially larger than unity for those that receive large logarithmic corrections via rge .
note that the normalization of the np contributions in ( [ eq : normal2 ] ) does not necessarily imply that they are generated only by loop amplitudes at the fundamental level : their natural size is comparable to that of sm loop amplitudes , but they could well be induced by tree - level processes .
for instance , a model with a heavy @xmath93 boson with a fcnc coupling to the @xmath94 current would also perfectly fit in this scheme , and in this case _ both _ @xmath95 and @xmath96 would receive tree - level contributions .
what we can not accommodate in this scheme is the possibility that @xmath95 receives large tree - level contributions and @xmath96 only loop - induced ones : in our language , this case would correspond to a violation of the condition ii ) .
since the measurement of the @xmath0@xmath1 mass difference falls in the ballpark of the sm expectations , the new flavour - changing coupling @xmath85 can not be arbitrarily large : barring fine - tuned scenarios with severe cancellations among different terms , we can allow at most @xmath81 corrections to the sm amplitude .
this implies ( see also @xcite ) 1 , [ eq : cond ] where possible @xmath81 factors associated with the matrix elements of the operators have been neglected .
owing to the different parametric dependence from scale factor and flavour - changing coupling of @xmath39 and @xmath2 operators in ( [ eq : npops ] ) , if the condition ( [ eq : cond ] ) is fulfilled the corresponding non - standard effects induced in @xmath2 flavour - changing neutral - current transitions turn out to be suppressed at least by a factor @xmath97 relative to the sm level : 1 .
[ eq : cond2 ] since the coefficient @xmath90 is substantially larger than 1 for qcd - penguin amplitudes , the suppression is even more severe in this case .
we have thus obtained a natural justification for the smallness of non - standard effects in @xmath2 loop - induced amplitudes in the well - motivated scenario of a heavy np scale ( @xmath98 ) .
although rather qualitative , the above argument has the great advantage of being almost independent of the details of the np model .
indeed , it can be realized in very different frameworks , from low - energy supersymmetry to models with large extra dimensions .
as can be easily understood , this argument does not apply only to @xmath0@xmath1 mixing : it is characteristic of any type of @xmath99 meson antimeson mixing versus the corresponding @xmath100 loop - induced amplitudes , provided the corresponding assumption ii ) is fulfilled . for this reason , in the phenomenological determination of the ckm matrix of section [ sec : ut - det ] , we shall try to avoid the use of observables such as @xmath40 or @xmath65 , which are sensitive to the @xmath101@xmath102 and @xmath103@xmath104 mixing amplitudes , respectively . in principle
, the case of @xmath105 transitions is somehow different from the @xmath106 and @xmath107 ones , since the @xmath103@xmath104 mass difference has not yet been measured .
this is indeed one of the reasons why speculations about possible large np effects in penguin - mediated @xmath105 transitions , such as @xmath108 and especially @xmath109 are still very popular ( see , for instance , @xcite and references therein ) .
however , we recall that the available data on @xmath65 already show a preference for this observable to be close to its sm expectation @xcite .
if we assume that np effects in @xmath65 can be at most of @xmath81 , we can accommodate large np effects in @xmath105 transitions only by means of violations of the conditions i ) and ii ) @xcite , or by fine - tuning .
therefore , in order to understand the consistency of our scenario , it will be very interesting to follow the evolution of future measurements of @xmath110 and @xmath109 decays .
so far , the available data on the @xmath108 observables fall well into the sm - allowed regions in observable space @xcite . in particular , they do not indicate any anomalous behaviour of the @xmath13 mode , which will be used in section [ sec : ut - det ] to control the penguin effects in @xmath9 . to conclude this general discussion
, it is worth stressing that this mechanism is not representative of all possible models with new sources of flavour mixing .
as already mentioned , the condition ii ) is not necessarily fulfilled . moreover
, a large hierarchy of matrix elements could invalidate the conditions ( [ eq : cond ] ) and ( [ eq : cond2 ] ) .
this happens , for instance , in supersymmetry with specific choices of the soft - breaking terms .
however , as we shall discuss in the following , it is very natural to assume that this mechanism works also within supersymmetric models , so that the largest np effects appear only in @xmath99 amplitudes . among specific extensions of the sm , low - energy supersymmetry is certainly one of the most interesting and well - motivated possibilities . in the absence of a mfv pattern for the soft - breaking terms ,
sizeable modifications of fcnc amplitudes are naturally expected within this framework , and the mass - insertion approximation provides a very efficient tool to describe them @xcite .
supersymmetric contributions to @xmath82@xmath83 mixing have been widely discussed in the literature ( see @xcite and references therein ) , and there exist several possibilities to accommodate arbitrary values of @xmath29 .
for instance , we can simply adjust the coupling @xmath111 ( or , equivalently , @xmath112 ) to produce the desired modification of @xmath82@xmath83 mixing via gluino - mediated box diagrams . in this case
the sum of supersymmetric and sm contributions to @xmath82@xmath83 mixing can be written as m_d e^-i_d & & _ b s_0(x_t ) | ( _ l _ d_l)^2 | b_d^0 + & & \
{ ^2 + r _ ^2 } , [ eq : bb_susy ] where @xmath113 and @xmath114 denote the initial condition and leading qcd corrections of the sm wilson coefficient , respectively ( see , e.g. , @xcite ) .
the explicit expression for @xmath115 the supersymmetric loop function depending on the ratio @xmath116 of squark and gluino masses and the corresponding qcd correction factor @xmath117 can be found in @xcite . as a reference figure , note that @xmath118 , and that we can set , to a good approximation , @xmath119 . expression ( [ eq : bb_susy ] ) provides a simple realization of the general scenario discussed in the previous subsection , with the adimensional effective flavour - changing coupling given by @xmath120 , and the effective np scale given by ^susy_eff = | |^1/2 1.4 tev ( ) . according to condition ( [ eq : cond ] ) ,
corrections to @xmath82@xmath83 mixing of @xmath81 should be obtained for @xmath121 : this expectation is fully confirmed by the detailed analysis of @xcite . as pointed out in @xcite ,
since sm and supersymmetric loop functions have the same sign , ; however , it holds for @xmath122 , which is suggested by rge constraints in grand - unified scenarios @xcite . ]
@xmath123 must have a large imaginary part if the sm term gives too large a @xmath124 , as in the case where the `` true value '' of @xmath12 lies in the second quadrant .
interestingly , as we shall see in the following section , this scenario is the favoured one for @xmath80 .
owing to the general dimensional argument in ( [ eq : cond2 ] ) , a supersymmetric framework with @xmath125 , where @xmath126 is the only new source of flavour - symmetry breaking , leads to negligible effects in @xmath127 transitions .
this can be explicitly verified by means of @xcite in the case of electroweak penguin amplitudes of the type @xmath128 , or by means of @xcite in the case of @xmath129 transitions . according to these analyses , if @xmath125 and if we have only left
left mass insertions , the supersymmetric corrections to electroweak penguin amplitudes reach at most the level of a few per cent with respect to the sm contributions .
these results are particularly important for the discussion of rare decays in section [ sec : rare ] : they show that it is perfectly conceivable to have a scenario where the np corrections to @xmath0@xmath1 mixing are large , but the direct np contributions to the rare decay amplitudes are negligible . within this framework ,
i.e. with heavy squark masses and non - standard sources of flavour - symmetry breaking induced only by left
left mass insertions , the relative impact of non - standard effects is even smaller for non - leptonic qcd - penguin amplitudes , such as those contributing to @xmath130 and @xmath13 decays . indeed , in this case the sm contribution
is substantially enhanced by the coupling constant and by large ( infra - red ) logarithms @xcite . on the other hand ,
as already mentioned in the general discussion , we recall that this framework is not representative of all possible supersymmetric scenarios .
for instance , it is well known that by means of flavour - non - diagonal @xmath131 terms , i.e. mass insertions of the left
right type , we can generate sizeable effects in @xmath100 transitions in particular rare decays without contradicting the @xmath99 bounds @xcite .
this happens because we may generate by means of left
right mass insertions sizeable contributions to @xmath132-breaking @xmath100 operators , which allow us to evade condition ii ) @xcite .
similarly , condition ii ) is badly violated by higgs - mediated fcnc amplitudes in large @xmath133 scenarios ( see , in particular , @xcite ) . also in this case the reason is intimately related to an interplay between flavour- and electroweak - symmetry breaking @xcite . to summarize
, we can state that the scenario with large non - standard effects only in @xmath99 amplitudes is characteristic of supersymmetric models with : i ) a heavy scale for the soft - breaking terms , ii ) new sources of flavour - symmetry breaking only ( or mainly ) in the soft - breaking terms which do not involve the higgs fields , iii ) yukawa interactions very similar to the pure sm case .
let us now assume that we have np of the kind specified in the previous section .
we may then complement the experimentally determined @xmath0@xmath1 mixing phase @xmath7 and the ckm factor @xmath43 with data on cp violation in the @xmath17-factory benchmark mode @xmath9 to fix the apex of the `` true '' unitarity triangle in the @xmath5@xmath6 plane .
the decay @xmath134 originates from @xmath135 quark - level transitions . within the sm and the scenario for
np introduced in section [ sec : np ] , we may write the corresponding decay amplitude as follows @xcite : @xmath136,\ ] ] where the cp - conserving strong parameter @xmath137 measures sloppily speaking the ratio of penguin to tree contributions in @xmath9 .
if we had negligible penguin contributions , i.e. @xmath138 , the corresponding cp - violating observables were simply given by @xmath139 where we have , in the last identity , also used the sm relation @xmath58 and the unitarity relation @xmath140 .
we observe that actually the phases @xmath7 and @xmath12 enter directly in the @xmath9 observables , and not @xmath10 .
consequently , since @xmath7 can be fixed straightforwardly through @xmath25 , we may use @xmath9 to probe @xmath12 , which has important advantages when dealing with penguin and np effects @xcite .
this procedure was also adopted in the `` qcd factorization '' analyses performed in @xcite .
we shall come back to this point below .
measurements of the cp - violating @xmath9 observables are already available : @xmath141 @xmath142 the babar and belle results are unfortunately not fully consistent with each other . hopefully , the experimental picture will be clarified soon .
if we nevertheless form the weighted averages of ( [ adir - exp ] ) and ( [ amix - exp ] ) , using the rules of the particle data group ( pdg ) @xcite , we obtain @xmath143 where the errors in brackets are the ones increased by the pdg scaling - factor procedure .
direct cp violation at this level would require large penguin contributions with large cp - conserving strong phases .
interestingly , a significant impact of penguins on @xmath9 is also indicated by data on the @xmath144 branching ratios @xcite , as well as by theoretical considerations @xcite .
consequently , it is already evident that the penguin contributions to @xmath9 _ can not _ be neglected . over the recent years
, many approaches to control the impact of the penguin contributions on the extraction of weak phases from the cp - violating @xmath9 observables were proposed ( see , for example , @xcite ) .
let us here follow the method suggested in @xcite , which is a variant of the @xmath9 , @xmath76 strategy proposed in @xcite .
if we apply ( [ bpipi - ampl ] ) , we may write ( for explicit expressions , see @xcite ) : @xmath145 which are _ exact _ parametrizations within the sm , and hold also for the np scenario specified in section [ sec : np ] .
if we fix @xmath7 through ( [ phid - exp ] ) , these two observables depend on three unknown parameters , @xmath146 , @xmath147 and @xmath12 . in order to extract these quantities
, it would be ideal to measure the following observables : @xmath148 where @xmath149 can be assumed to be negligibly small in the sm , or can be fixed through cp - violating effects in @xmath150 .
since @xmath76 is related to @xmath9 through an interchange of all strange and down quarks , i.e. through the @xmath44-spin flavour symmetry of strong interactions , we may derive the @xmath44-spin relation @xmath151 which allows us to determine @xmath146 , @xmath147 , @xmath152 and @xmath12 from the cp - violating observables of the @xmath76 , @xmath9 system @xcite .
unfortunately , @xmath76 is not accessible at the @xmath153 @xmath17 factories operating at the @xmath154 resonance ; this decay can be observed for the first time at run ii of the tevatron @xcite , and can be ideally studied in the era of the lhc @xcite .
however , since @xmath76 is related to @xmath13 through an interchange of spectator quarks , we have @xmath155 @xmath156 and may approximately use @xmath13 , which has already been measured at the @xmath17 factories , to deal with the penguins in @xmath9 @xcite .
the key quantity is then the following ratio of the cp - averaged @xmath9 and @xmath13 branching ratios : @xmath157= \left\{\begin{array}{cc } 7.4\pm2.5 & \mbox{(cleo \cite{cleo - bpik})}\\ 7.6\pm1.2 & \mbox{(babar \cite{babar - bpik})}\\ 7.1\pm1.9 & \mbox{(belle \cite{belle - bpik } ) } \end{array}\right\}=7.5\pm0.9,\ ] ] where the factor @xmath158 involving the kaon and pion decay constants takes into account factorizable @xmath44-spin - breaking corrections , and @xmath159 .
if we employ in addition to ( [ u - spin - rel1 ] ) another @xmath44-spin relation , @xmath160 and make plausible dynamical assumptions to replace the @xmath76 channel through @xmath13,-spin relation ( [ u - spin - rel1 ] ) is sufficient in the case of the @xmath76 , @xmath9 strategy . ]
we may write @xmath161 consequently , ( [ cp - bpipi - par ] ) and ( [ h - par ] ) allow us now to determine @xmath12 , as well as @xmath146 and @xmath147 .
the explicit formulae to implement this strategy can be found in @xcite , taking also into account possible corrections to ( [ u - spin - rel1 ] ) and ( [ u - spin - rel2 ] ) . as discussed in detail in @xcite , additional information is also provided by the direct cp asymmetry of the @xmath13 modes , which is related to the one of @xmath9 through the following @xmath44-spin relation @xcite : @xmath162
if we insert @xmath163 and @xmath164 into this expression , we obtain @xmath165 . on the other hand ,
the present average of the @xmath17-factory results for this cp asymmetry is given by @xmath166 @xcite . in comparison with the old result @xmath167 considered in @xcite ,
this cp asymmetry has interestingly moved towards larger values , and its sign is in accordance with the expectation following from ( [ u - spin - cp ] ) . in view of the considerable experimental uncertainties ,
the feature that @xmath168 still seems to favour somewhat smaller values than @xmath169 can not be considered as a discrepancy . moreover ,
( [ cp - rel ] ) and ( [ br - rel ] ) rely not only on the @xmath44-spin symmetry , but also on dynamical assumptions , as `` exchange '' and `` penguin annihilation '' topologies contribute to the @xmath76 channel , but not to @xmath170 .
although such topologies are expected to play a minor rle , they may in principle be enhanced through large rescattering effects .
once the @xmath76 channel is measured , we have no longer to rely on this dynamical assumption , and may check how well ( [ cp - rel ] ) and ( [ br - rel ] ) are actually satisfied .
in particular a measurement of the cp - averaged @xmath76 branching ratio , which will hopefully be soon available from the cdf collaboration , would already be a very important step ahead , allowing a considerably more solid determination of @xmath171 . as pointed out in @xcite
, there is a transparent way to characterize the @xmath9 mode in the space of its cp - violating observables , allowing a simple comparison with the experimental data .
let us focus here on the unitarity triangle and the allowed region for its apex in the @xmath5@xmath6 plane .
if we follow @xcite and employ the quantity @xmath171 to control the penguin effects in @xmath9 , we may convert the direct and mixing - induced cp asymmetries of this decay into values of @xmath12 . using @xmath172 and the experimental averages given in ( [ bpipi - cp - averages ] ) and ( [ bpipi - cp - averages2 ] ) , we obtain @xmath173
as pointed out in @xcite , these solutions are related to each other through @xmath174 because of the unsatisfactory experimental situation concerning the cp asymmetries of @xmath9 , the ranges in ( [ gam - res ] ) should mainly be considered as an illustration of how our strategy is working . indeed , in order to obtain ( [ gam - res ] ) , we have just used the `` ordinary '' errors in ( [ bpipi - cp - averages ] ) and ( [ bpipi - cp - averages2 ] ) , and not the enlarged ones given there in brackets .
interestingly , the information on the _ positive _ sign of @xmath175 , which is favoured by the belle measurement @xcite , already implies that @xmath176 _ can not _ be accommodated in the case of @xmath15 , as emphasized in @xcite .
the experimental uncertainties will be reduced considerably in the future , thereby providing significantly more stringent results for @xmath12 .
it should be noted that we have assumed in ( [ gam - res ] ) as is usually done that @xmath177 $ ] .
this range is implied by the interpretation of @xmath40 , provided that : i ) np does not change the sign of the @xmath41 amplitude with respect to the sm ; ii ) the `` bag '' parameter @xmath178 is positive ( as indicated by all existing non - perturbative calculations ) . a similar assumption about the `` bag '' parameter @xmath179 the @xmath180-meson counterpart of @xmath178 enters also in ( [ phid - np ] )
; for a discussion of the very unlikely @xmath181 , @xmath182 cases , see @xcite .
if we relaxed these assumptions about np and/or `` bag '' parameters , we would need to double the solutions and consider the specular case @xmath183 $ ] .
.input values used to draw the @xmath40 , @xmath62 and @xmath63 constraints in figs .
[ fig : first ] and [ fig : second ] , and to analyse the branching ratios of rare decays in section [ sec : rare ] . as usual , we define @xmath184 . [ cols="<,<",options="header " , ] the results shown in figs . [
fig : first ] and [ fig : second ] are complemented by table [ expval ] , where we collect the ranges for the angles @xmath10 , @xmath11 and @xmath12 of the unitarity triangle , as well as those for the generalized wolfenstein parameters @xmath5 and @xmath6 . in this table , we illustrate also the impact of @xmath44-spin - breaking effects on ( [ u - spin - rel1 ] ) , which are described by the following parameter : @xmath185 as noted in @xcite , these effects are potentially much more important than possible corrections to ( [ u - spin - rel2 ] ) . nevertheless , we observe that the impact of a variation of @xmath186 within @xmath187 $ ] is quite moderate .
interestingly , we may well accommodate values of @xmath10 around @xmath188 in the case of @xmath189 , in contrast to the situation of @xmath190 , which would favour @xmath10 to lie around @xmath191 .
theoretical arguments for @xmath192 were given in @xcite .
let us now compare our results with the analysis performed by the belle collaboration in @xcite , yielding the range @xmath193 the starting point of this study , which follows closely @xcite , is also the parametrization ( [ bpipi - ampl ] ) of the @xmath194 decay amplitude . however , there are important differences in comparison with our analysis : 1 .
the range in ( [ belle - range ] ) corresponds to the 95.5% c.l .
interval of the belle asymmetries , constrained to lie within the physical region , whereas the results in table [ expval ] were obtained for the averages of the babar and belle measurements in ( [ bpipi - cp - averages ] ) and ( [ bpipi - cp - averages2 ] ) ( with `` ordinary '' errors ) , which fall well into the 95.5% c.l . belle region .
2 . in the belle analysis , @xmath146 was varied within @xmath195 $ ] to explore the penguin effects , whereas we use the observable @xmath171 to control the penguin contributions .
3 . a crucial difference arises , since the sm relation @xmath58 was used in @xcite , in combination with the unitarity relation @xmath196 , to eliminate @xmath12 ( see ( [ bpipi - cp0 ] ) ) .
following these lines , the attention is implicitly restricted to the standard solution @xmath189 with @xmath197 , so that the case of @xmath190 _ can not _ be explored . despite the different input values for the cp asymmetries and
the different treatment of the penguin contributions , ( [ belle - range ] ) is in good agreement with the ranges for @xmath10 collected in table [ expval ] for @xmath189 . and insert the @xmath189 range for @xmath12 in ( [ gam - res ] ) into @xmath198 , we obtain @xmath199 . as ( [ belle - range ] ) , this range does not depend on @xmath43 , but relies on the sm relation @xmath58 . ] on the other hand , all the interesting features related to the @xmath190 solution were not addressed in the analysis in @xcite .
interestingly , flavour - symmetry strategies to determine @xmath12 with the help of charged @xcite and neutral @xcite @xmath18 modes show some preference for values of @xmath12 larger than @xmath188 ( for recent overviews , see @xcite ) . moreover
, as pointed out in @xcite , the theoretical predictions for the penguin parameter @xmath201 obtained within the `` qcd factorization '' @xcite and `` pqcd '' @xcite approaches can be brought to better agreement with the measured value of @xmath171 in the case of @xmath16 .
it should also be noted that the global fits to all available @xmath18 , @xmath202 data show a similar picture ( see , for instance , @xcite ) .
since the @xmath18 analyses are rather involved , we shall not discuss them here in further detail ; let us just emphasize that their preferred values for @xmath12 could be conveniently accommodated in fig .
[ fig : second ] , i.e.for @xmath190 , but not in the case of the conventional @xmath189 solution shown in fig .
[ fig : first ] .
because of the large experimental uncertainties , it is of course too early to draw definite conclusions , but the experimental situation is expected to improve continuously in the future . in this context , it is also helpful to construct a set of robust sum rules @xcite , which are satisfied by the @xmath18 observables .
in this section , we shall explore the implications of the allowed regions in the @xmath5@xmath6 plane shown in figs .
[ fig : first ] and [ fig : second ] for the branching ratios of the very rare decay processes @xmath21 and @xmath20 . needless to note ,
the predictions corresponding to the @xmath190 scenario are of particular interest . as in the previous section
, we assume that np does not affect the amplitudes of these modes , i.e. that it manifests itself only through the different allowed ranges for @xmath5 and @xmath6 shown in figs .
[ fig : first ] and [ fig : second ] .
as is well known , the rare kaon decays @xmath203 and @xmath204 offer valuable tools to explore flavour physics .
these modes originate from @xmath91 penguins and box diagrams , and are theoretically very clean .
let us first focus on @xmath203 , which has already been observed by the e787 experiment at bnl through two clean events , yelding the following branching ratio @xcite : @xmath205 within the sm and the np scenarios specified above , the `` reduced '' branching ratio @xcite @xmath206 takes the following form : @xmath207 ^ 2 + \left[\frac{\mbox{re}\lambda_t}{\lambda^5}x(x_t)+ \frac{\mbox{re}\lambda_c}{\lambda } p_c(\nu\overline{\nu})\right]^2,\ ] ] where @xmath208 and @xmath209 are appropriate coefficient functions , which encode top- and charm - quark loop contributions , respectively @xcite .
taking into account @xmath210 and @xmath211 , we obtain @xmath212 and @xmath213 respectively , where @xmath214 .
the following ranges for the @xmath203 branching ratio are obtained by scanning @xmath5 and @xmath6 within the black regions shown in figs .
[ fig : first ] and [ fig : second ] , and varying simultaneously @xmath215 , @xmath216 and the relevant parameters in table [ tab : inputs ] within their allowed ranges . in the case
when @xmath189 , we obtain @xmath217 whereas the @xmath190 scenario favours a larger branching ratio , @xmath218 which is due to the sign - change of @xmath5 in ( [ sym2 ] ) . despite the large present uncertainties , it is interesting to note that ( [ kpnn2 ] ) is in better agreement with the experimental result ( [ kpnn - exp ] ) , thereby favouring the @xmath190 solution . on the other hand ,
( [ kpnn1 ] ) agrees well with the sm range @xmath219 given in @xcite , which is not surprising .
note , however , that ( [ kpnn1 ] ) relies only on the data on @xmath7 , cp violation in @xmath220 , the observable @xmath171 , and the measurement of the side @xmath43 of the unitarity triangle . on the other hand ,
the usual ckm fits were used in ( [ kpnn - sm ] ) .
the observation that the central value of the bnl - e787 result does not necessarily imply a non - standard contribution to the @xmath203 amplitude , but that it could well be accommodated by np effects in @xmath0@xmath1 mixing only , has already been made in @xcite .
indeed , the sm interpretation of the measured br@xmath19 defines a region in the @xmath5@xmath6 plane which is perfectly consistent with the @xmath43 circle , but favours @xmath16 . adding to this analysis the @xmath130 information
, we can now conclude that the measured @xmath221 favours a @xmath0@xmath1 mixing phase of @xmath190 .
let us now have a brief look at the decay @xmath204 , which is characterized by @xmath222 with @xmath223 ^ 2 = \eta^2a^4x(x_t)^2.\ ] ] in contrast to @xmath224 , this reduced branching ratio does not depend on @xmath5 .
consequently , since the allowed ranges for @xmath6 are equal in figs .
[ fig : first ] and [ fig : second ] because of ( [ sym2 ] ) , @xmath204 does not allow us to distinguish between these two cases , i.e. we obtain the same range for @xmath189 and @xmath225 : @xmath226 which overlaps well with the sm expectation @xcite @xmath227 let us now , finally , turn to two rare @xmath17 decays , @xmath229 and @xmath230 . within the sm and the kind of np specified above , they are mediated by box diagrams and @xmath91 penguins
. we may write the branching ratio for @xmath229 as follows @xcite : @xmath231 ^ 2 \left [ \frac{|v_{td}|}{0.008 } \right]^2 \left [ \frac{\tau_{b_d}}{1.5 \ , \mbox{ps } } \right ] \left [ \frac{{\overline m}_t(m_t ) } { 167 \ , \mbox{gev } } \right]^{3.12},\end{aligned}\ ] ] where @xmath232 is given by @xmath233.\end{aligned}\ ] ] we may now calculate , in analogy to our discussion of the @xmath21 modes given in subsection [ subsec : kpinunu ] , the range for this branching ratio corresponding to the black regions shown in figs .
[ fig : first ] and [ fig : second ] . in the case of @xmath189 , we obtain @xmath234 whereas the @xmath190 scenario favours the following , larger branching ratio : @xmath235 as in the case of @xmath203 , the enhancement is due to the sign - change of @xmath5 in ( [ sym2 ] ) . on the other hand , we have @xmath236 ^ 2 \left [ \frac{|v_{ts}|}{0.040 } \right]^2 \left [ \frac{\tau_{b_s}}{1.5 \ , \mbox{ps } } \right ]
\left [ \frac{{\overline m}_t(m_t ) } { 167 \ , \mbox{gev } } \right]^{3.12},\end{aligned}\ ] ] where @xmath237 is given up to tiny corrections entering at the @xmath238 level as follows : @xmath239 consequently , since @xmath240 does not depend on @xmath5 and @xmath6 , if we neglect these corrections , it has no sensitivity on the allowed ranges in the @xmath5@xmath6 plane shown in figs .
[ fig : first ] and [ fig : second ] , i.e. we obtain the same prediction for @xmath189 and @xmath190 : @xmath241 at first sight , @xmath240 does therefore not appear to be of great interest for our analysis .
however , this is actually not the case , since the ratio @xmath242\left [ \frac{m_{b_d}}{m_{b_s}}\right]\left[\frac{f_{b_d}}{f_{b_s}}\right]^2 \left|\frac{v_{td}}{v_{ts}}\right|^2\ ] ] is affected to a much smaller extent by the hadronic uncertainties associated with @xmath243 and @xmath244 than the individual branching ratios .
these decay constants now enter in a ratio , which equals 1 in the @xmath245 limit , i.e. we have only to deal with the @xmath245-breaking corrections to this quantity @xcite : @xmath246 whereas the ranges for the individual decay constants can be found in table [ tab : inputs ] . using ( [ vts ] ) and taking ( [ rt - def ] ) into account , we obtain from ( [ mumu - rat ] ) the following expression : @xmath247\left [ \frac{m_{b_s}}{m_{b_d}}\right]\left[\frac{f_{b_s}}{f_{b_d}}\right]^2 \left[\frac{\mbox{br}(b_d \to \mu^+\mu^-)}{\mbox{br}(b_s \to \mu^+\mu^-)}\right]=\lambda^2 r_t^2.\ ] ] since the side @xmath64 of the unitarity triangle takes quite different values for the situations shown in figs .
[ fig : first ] and [ fig : second ] because of ( [ sym2 ] ) , we may nicely probe them through @xmath248 . in the case of @xmath189 , we find @xmath249 while our @xmath190 scenario corresponds to @xmath250 unfortunately , the sensitivity of the existing experiments is still far from the level necessary to probe the ratio @xmath248 .
the present experimental upper bounds read as follows : @xmath251 @xmath252 however , owing to clean experimental signatures , these processes are among the benchmark modes of future @xmath17-physics experiments at hadron colliders .
the main points of this paper can be summarized as follows : * we have emphasized that the cp - violating @xmath0@xmath1 mixing phase @xmath253 , which has been determined from @xmath4 , can not be represented straightforwardly in the @xmath5@xmath6 plane in the presence of np contributions to @xmath0@xmath1 mixing .
this feature affects in particular the `` unconventional '' solution of @xmath190 , which definitely requires such kind of physics beyond the sm .
we have pointed out that it is therefore very misleading to represent this solution simply as a second branch in the @xmath5@xmath6 plane , as is usually done in the literature . * in the solution of this problem , @xmath43 and @xmath12 play a key rle . from a theoretical point of view
, certain pure tree decays , for example @xmath254 or @xmath255 modes , would offer an ideal tool to determine @xmath12 , providing theoretically clean results that are , in addition , very robust under the influence of np .
unfortunately , these strategies can not yet be implemented .
however , an interesting alternative is offered by the @xmath17-factory benchmark mode @xmath9 , where first experimental results on cp - violating effects are already available from babar and belle .
* we have considered a specific scenario for physics beyond the sm , where we have large np contributions to @xmath0@xmath1 mixing , but not to the @xmath2 and @xmath42 decay processes . within this framework
, we may then also straightforwardly accommodate the @xmath190 solution .
we have given general conditions for such a kind of np , and have argued that it is well motivated in supersymmetry and several specific frameworks .
* as widely discussed in the literature , the cp - violating @xmath9 observables provide valuable information about the `` true '' apex of the unitarity triangle for such a scenario of np .
however , the true challenge in this game is the control of penguin effects . to this purpose
, we employ the cp - averaged @xmath256 branching ratio , and make use of @xmath44-spin and plausible dynamical assumptions . following these lines
, we may determine @xmath12 for each of the two solutions for @xmath7 . complementing this information with the experimental range for @xmath43
, we may fix the apex of the unitarity triangle , and thus determine also @xmath10 and @xmath11 . in the case of @xmath67 , we arrive at the first quadrant of the @xmath5@xmath6 plane , and obtain an allowed region , which is in perfect agreement with the constraints arising from the sm interpretation of @xmath7 , @xmath62 and @xmath63 , as well as of @xmath40 . on the other hand , for @xmath257 , we obtain an allowed region in the second quadrant , corresponding to @xmath16 .
interestingly , this range is still consistent with the @xmath40 hyperbola , whereas we may now no longer use @xmath62 and @xmath63 to determine @xmath64 , as these quantities would receive np contributions .
* we have also explored the implications of these two solutions for very rare @xmath258 and @xmath17 decays , @xmath21 and @xmath20 , respectively , assuming that their decay amplitudes take the same form as in the sm . in this case , np would manifest itself only indirectly , through the determination of @xmath5 and @xmath6 . consequently , for @xmath189 we obtain the same picture for these modes as in the sm . however , in the case of @xmath257 , the branching ratio for @xmath203 that would be favoured is about twice as large as in the sm , and would be in better agreement with the present measurement .
similarly , if @xmath257 , a branching ratio about twice as large as in the sm is expected for @xmath259 . on the other hand , the impact of this non - standard solution on both br@xmath260 and br@xmath261 would be negligible . at the present stage
, we may not yet draw any definite conclusion , because of the large experimental uncertainties and the unsatisfactory situation of the measurement of the cp - violating @xmath9 observables ; hopefully , the discrepancy between babar and belle will be resolved soon .
interestingly , alternative analyses of the data for the @xmath262 branching ratios show some preference for @xmath16 , thereby favouring the @xmath190 scenario .
a similar comment applies to the rare kaon decay @xmath203 . on the other hand , it should not be forgotten that the allowed region for @xmath189 is consistent with the stringent constraints from @xmath31@xmath32 mixing .
as far as @xmath40 is concerned , both the @xmath189 and the @xmath190 scenario are consistent with the sm interpretation of this quantity . in view of these observations
, it would be very important to distinguish directly between the two solutions for @xmath7 , which are implied by their extraction from @xmath45 , through a measurement of the sign of @xmath49 .
m. kobayashi and t. maskawa , _ prog .
_ * 49 * ( 1973 ) 652
. babar collaboration ( b. aubert _ et al .
_ ) , _ phys . rev .
lett . _ * 87 * ( 2001 ) 091801 ; + belle collaboration ( k. abe _
_ ) , _ phys
. rev .
_ * 87 * ( 2001 ) 091802 .
l. wolfenstein , _ phys .
lett . _ * 51 * ( 1983 ) 1945 ; + a.j .
buras , m.e .
lautenbacher and g. ostermaier , _ phys .
_ * d50 * ( 1994 ) 3433 .
buras , tum - hep-435 - 01 [ hep - ph/0109197 ] ; + a. ali and d. london , _ eur .
j. _ * c18 * ( 2001 ) 665 ; + d. atwood and a. soni , _ phys .
_ * b508 * ( 2001 ) 17 ; + m. ciuchini _ et al .
_ , _ jhep _ * 0107 * ( 2001 ) 013 ; + a. hcker _
_ , _ eur .
j. _ * c21 * ( 2001 ) 225 .
y. nir , wis-35 - 02-dpp [ hep - ph/0208080 ] .
r. fleischer , _ phys .
_ * 370 * ( 2002 ) 537 .
r. fleischer and t. mannel , _ phys .
_ * b506 * ( 2001 ) 311 .
g. dambrosio and g. isidori , _ phys .
_ * b530 * ( 2002 ) 108 .
d. becirevic _ et al .
_ , _ nucl .
_ * b634 * ( 2002 ) 105 .
y. grossman , y. nir and m.p .
worah , _ phys .
_ * b407 * ( 1997 ) 307 .
_ , _ phys .
_ * b500 * ( 2001 ) 161 .
g. dambrosio , g.f .
giudice , g. isidori and a. strumia , _ nucl .
_ * b645 * ( 2002 ) 155 .
r. fleischer , _ eur .
j. _ * c16 * ( 2000 ) 87 .
r. fleischer and j. matias , _ phys .
_ * d66 * ( 2002 ) 054009 .
m. beneke _
_ , _ nucl .
_ * b606 * ( 2001 ) 245 .
m. neubert , clns-02 - 1794 [ hep - ph/0207327 ] .
j.p . silva and l. wolfenstein ,
_ * d49 * ( 1994 ) 1151 ; + a.l .
kagan and m. neubert , _ phys .
_ * b492 * ( 2000 ) 115 ; + g. eyal , y. nir and g. perez , _ jhep _ * 0008 * ( 2000 ) 028 ; + f. botella , g. branco , m. nebot and m. rebelo , _ nucl .
_ * b651 * ( 2003 ) 174 ; + a.j .
buras , f. parodi and a. stocchi , _ jhep _ * 0301 * ( 2003 ) 029 .
azimov , v.l . rappoport and v.v .
sarantsev , _ z. phys .
_ * a356 * ( 1997 ) 437 ; + y. grossman and h.r .
quinn , _ phys .
_ * d56 * ( 1997 ) 7259 ; + j. charles _ et al .
_ * b425 * ( 1998 ) 375 ; + b. kayser and d. london , _ phys .
_ * d61 * ( 2000 ) 116012 ; + h.r .
_ , _ phys .
_ * 85 * ( 2000 ) 5284 .
dighe , i. dunietz and r. fleischer , _ phys .
_ * b433 * ( 1998 ) 147 ; + i. dunietz , r. fleischer and u. nierste , _ phys .
_ * d63 * ( 2001 ) 114015 .
r. itoh , kek - preprint-2002 - 106 [ hep - ex/0210025 ] .
m. gronau and d. wyler , _ phys .
_ * b265 * ( 1991 ) 172 ; + d. atwood , i. dunietz and a. soni , _ phys .
. _ * 78 * ( 1997 ) 3257 ; + r. fleischer and d. wyler , _ phys . rev .
_ * d62 * ( 2000 ) 057503 .
+ y. grossman , z. ligeti and a. soffer , lbnl-51630 [ hep - ph/0210433 ] ; + m. gronau , cern - th/2002 - 331 [ hep - ph/0211282 ] . j.p .
silva and a. soffer , _ phys .
_ * d61 * ( 2000 ) 112001 ; + d. atwood , i. dunietz and a. soni , _ phys .
_ * d63 * ( 2001 ) 036005 .
r. fleischer , cern - th/2003 - 010 , to appear in _ phys .
_ * b * [ hep - ph/0301255 ] .
r. fleischer , cern - th/2003 - 011 , to appear in _ nucl .
_ * b * [ hep - ph/0301256 ] .
r. fleischer , _ phys .
_ * b459 * ( 1999 ) 306 .
buras and r. fleischer , _ phys .
_ * d64 * ( 2001 ) 115010 .
m. ciuchini , e. franco , a. masiero and l. silvestrini , hep - ph/0212397 ; + r. harnik , d.t .
larson , h. murayama and a. pierce , hep - ph/0212180 .
r. fleischer and t. mannel , _ phys .
_ * b511 * ( 2001 ) 240 .
hagelin , s. kelley and t. tanaka , _ nucl .
_ * b415 * ( 1994 ) 293 ; + f. gabbiani _
_ , _ nucl .
_ * b477 * ( 1996 ) 321 .
g. buchalla , a.j .
buras and m.e .
lautenbacher , _ rev .
_ * 68 * ( 1996 ) 1125 .
d. choudhury _ et al .
_ , _ phys .
_ * b342 * ( 1995 ) 180 .
e. lunghi , a. masiero , i. scimemi and l. silvestrini , _ nucl .
_ * b568 * ( 2000 ) 120 .
buras , a. romanino and l. silvestrini , _ nucl .
_ * b520 * ( 1998 ) 3 .
r. barbieri and a. strumia , _ nucl .
_ * b508 * ( 1997 ) 3 .
g. colangelo and g. isidori , _ jhep _ * 09 * ( 1998 ) 009 ; + a. masiero and h. murayama , _ phys .
_ * 83 * ( 1999 ) 907 .
_ * b566 * ( 2000 ) 3 .
babu and c.f .
kolda , _ phys .
_ * 84 * ( 2000 ) 228 ; + a.j .
_ , _ nucl .
_ * b619 * ( 2001 ) 434 ; + g. isidori and a. retico , _ jhep _ * 0111 * ( 2001 ) 001 .
b. aubert _
et al . _ ( babar collaboration ) , _ phys . rev .
_ * 89 * ( 2002 ) 281802
et al . _ ( belle collaboration ) , belle preprint 2003 - 1 [ hep - ex/0301032 ] .
particle data group ( k. hagiwara _ et al .
_ ) , _ phys . rev .
_ * d66 * ( 2002 ) 010001 .
m. gronau and j.l .
rosner , _ phys .
_ * d65 * ( 2002 ) 093012 .
a.i . sanda and k. ukai ,
* 107 * ( 2002 ) 421 ; + y .- y . keum , dpnu-02 - 30 [ hep - ph/0209208 ] m. gronau and d. london , _ phys . rev .
_ * 65 * ( 1990 ) 3381 ; + r. fleischer and t. mannel , _ phys . lett .
_ * b397 * ( 1997 ) 269 ; + y. grossman and h.r .
quinn , _ phys .
_ * d58 * ( 1998 ) 017504 ; + j. charles , _ phys .
rev . _ * 59 * ( 1999 ) 054007 ; + d. london , n. sinha and r. sinha , _ phys . rev .
_ * d63 * ( 2001 ) 054015 ; + m. gronau , d. london , n. sinha and r. sinha , _ phys .
_ * b514 * ( 2001 ) 315 ; + m. gronau and j.l .
rosner , _ phys .
_ * d66 * ( 2002 ) 053003 .
k. anikeev _
et al . _ , fermilab - pub-01/197 [ hep - ph/0201071 ] .
et al . _ , cern - th/2000 - 101 [ hep - ph/0003238 ] , in cern report on standard model physics ( and more ) at the lhc ( cern , geneva , 2000 ) p. 305 .
cleo collaboration ( a. bornheim _
_ ) , cleo 03 - 03 [ hep - ex/0302026 ] . babar collaboration ( b. aubert _ et al .
_ ) , hep - ex/0207065 ; hep - ex/0206053 .
r. fleischer and j. matias , _ phys .
_ * d61 * ( 2000 ) 074004 .
y. grossman , b. kayser and y. nir , _ phys .
_ * b415 * ( 1997 ) 90 .
h. fritzsch and z .- z .
xing , _ nucl .
_ * b556 * ( 1999 ) 49 .
m. neubert and j.l .
rosner , _ phys .
_ * b441 * ( 1998 ) 403 ; _ phys .
_ * 81 * ( 1998 ) 5076 .
buras and r. fleischer , _ eur .
j. _ * c16 * ( 2000 ) 97 ; * c11 * ( 1999 ) 93 .
rosner , efi-02 - 96 [ hep - ph/0207197 ] ; + r. fleischer , cern - th/2002 - 293 [ hep - ph/0210323 ] .
hou , j.g .
smith and f. wrthwein , ntu - hep-99 - 25 [ hep - ex/9910014 ] .
m. gronau and j.l .
rosner , _ phys .
_ * d59 * ( 1999 ) 113002 ; + m. neubert , _ jhep _ * 9902 * ( 1999 ) 014 ; + j. matias , _ phys .
_ * b520 * ( 2001 ) 131 and hep - ph/0209331 .
e787 collaboration ( s. adler _ et al .
_ ) , _ phys . rev .
lett . _ * 88 * ( 2002 ) 041803
. g. buchalla and a.j .
buras , _ phys .
_ * b333 * ( 1994 ) 221 ; _ phys .
_ * d54 * ( 1996 ) 6782 .
kettell , l.g .
landsberg and h.h .
nguyen , fermilab - fn-0727 [ hep - ph/0212321 ] .
buras , tum - hep-402 - 01 [ hep - ph/0101336 ] .
l. lellouch , cpt-2002-p-4443 [ hep - ph/0211359 ] .
cdf collaboration ( f. abe _
_ ) , _ phys .
_ * d57 * ( 1998 ) 3811 .
babar collaboration ( b. aubert _ et al .
_ ) , babar - conf-02/028 [ hep - ex/0207083 ] .
|
in a wide class of new - physics models , which can be motivated through generic arguments and within supersymmetry , we obtain large contributions to @xmath0@xmath1 mixing , but not to @xmath2 processes .
if we assume such a scenario , the solutions @xmath3 for the @xmath0@xmath1 mixing phase implied by @xmath4 can not be converted directly into a constraint in the @xmath5@xmath6 plane .
however , we may complement @xmath7 with @xmath8 and the recently measured cp asymmetries in @xmath9 to determine the unitarity triangle , with its angles @xmath10 , @xmath11 and @xmath12 . to this end
, we have also to control penguin effects , which we do by means of the cp - averaged @xmath13 branching ratio .
interestingly , the present data show a perfectly consistent picture not only for the `` standard '' solution of @xmath14 , but also for @xmath15 . in the latter case ,
the preferred region for the apex of the unitarity triangle is in the second quadrant , allowing us to accommodate conveniently @xmath16 , which is also favoured by other non - leptonic @xmath17 decays such as @xmath18 .
moreover , also the prediction for br@xmath19 can be brought to better agreement with experiment .
further strategies to explore this scenario with the help of @xmath20 decays are discussed as well .
cern - th/2003 - 039 + uab - ft-541 + hep - ph/0302229 * * shedding light on the `` dark side '' of @xmath0@xmath1 mixing through @xmath9 , @xmath21 and @xmath20 robert fleischer,@xmath22 gino isidori,@xmath23 joaquim matias@xmath24 @xmath22 _ theory division , cern , ch-1211 geneva 23 , switzerland _ @xmath23 _ infn , laboratori nazionali di frascati , i-00044 frascati , italy _
@xmath24 _ ifae , universitat
autnoma de barcelona , 08193 bellaterra , barcelona , spain _ february 2003
|
single - molecule manipulation @xcite has opened new perspectives on understanding of the mechanical processes taking place in a biological cell and may offer insights into design of nanostructures and nanomachines .
examples of manipulation of a protein include stretching @xcite , mechanically controlled refolding @xcite , knot untying @xcite and knot tightening @xcite .
the experimental studies pertained to only a handful of systems and yet demonstrated richness of possible mechanical behaviors .
experiments on stretching generate information on mechanostability .
it can be captured by providing the value of @xmath1 the largest force that is needed to unravel the tertiary structure of a protein .
@xmath1 can be as large as 480 pn
the value measured for scaffoldin c7a @xcite .
the structural motif that is the core of resistance to stretching is known as the mechanical clamp . in most cases of large mechanostability ,
the mechanical clamp consists primarily of parallel @xcite ( and sometimes antiparallel @xcite ) @xmath2-strands that get sheared on pulling at two points of attachment , which are often the terminal amino acids .
this mechanism is operational , for instance , in scaffoldin @xcite and titin @xcite .
recently , we have made stretching simulations for 17 134 proteins within a coarse grained structure - based model @xcite .
this survey pertained to proteins built of no more than 250 amino acids .
its results are deposited at the bsdb database ( biomolecule stretching database , www://info.ifpan.edu.pl / bsdb/ ) as described in ref .
the survey not only identified about 35 proteins of larger mechanostability than the scaffoldin but it also discovered an entirely new type of the mechanical clamp : the cystine slipknot ( csk ) .
in fact , the 13 top strongest model proteins are endowed with this mechanism .
it involves forced dragging of a piece of the backbone , known as the slip - loop , through a cystine ring which is a special case of a knot - loop .
thus csk does not exist in the native state but it develops through pulling .
the cystine ring arises by connecting two backbone segments by two disulfide bonds between two pairs of cysteins .
the ring is a part of what is known as the cystine knot ( ck ) motif @xcite and has been first discovered in the superfamily of growth factors @xcite and is responsible for the high mechanical stability of collagen @xcite .
it typically contains between 8 and 15 residues @xcite and it has emerged probably as a result of convergent evolution @xcite .
the pulling - induced dragging takes place because of the presence of a third disulfide bond that pierces the ring ( figure [ clamps ] ) .
this csk mechanism has been predicted @xcite to yield @xmath1 exceeding even 1000 pn .
such large forces would arise because of jamming resulting from the slip - loop getting stuck in the cystine ring .
an identifiable force peak maximum appears if the steric hindrances are overcome and the clinch is released .
it should be noted that the acronym csk has been initially meant to stand for `` cysteine slipknot '' @xcite as the motif involves cysteins .
however the term `` cystine slipknot '' used in this paper seems more apt as it brings the role of the disulfide bonds into fore . at present , there are three well established families within the cystine knot superfamily : growth factor cystine knots ( gfck , e.g. vascular endothelial growth factor @xcite ) , inhibitor cystine knots ( ick , e.g. scorpion venom proteins @xcite ) and cyclic cystine knots ( cck ) which have no terminal amino acids @xcite .
the cystine knot proteins studied in this work are members of the first family .
the ck proteins are mostly extracellular factors and they are known for their remarkable stability against enzymatic cleavage @xcite . these 35 strongest proteins , with the csk mechanical clamp and without , have not yet been studied in single molecule stretching experiments . furthermore , our predictions are based on a simple model that accounts for the geometry and size of the side groups in a very approximate manner ( primarily through the length parameters in the contact potentials ) whereas the success in dragging through the cystine ring crucially depends on the actual size and structure of the moving segments involved . in this paper , we perform all - atom steered molecular dynamics ( smd ) simulations and confirm the existence of the csk mechanical clamps with the values of @xmath1 being substantially larger than those associated with the shear - based mechanical clamps .
we focus on the top 20 proteins identified in the survey as listed in table 1 .
we also consider titin which serves as a benchmark @xcite .
we find that even though the coarse grained model captures the essence of the csk clamp and , in particular , that it involves jamming , it misses several interesting atomic - level components .
the first component is the different origin of two possible stretching pathways : in the all - atom model what matters is the direction of approach of the pulled backbone to the cystine ring whereas it is the number of amino acids ( one or two ) that cross the ring simultaneously in the coarse grained model .
another component is that the tendency of developing multiple force peaks on a given pathway is stronger in the all - atom model because size modulations in the side groups may split the force peaks .
these modulation are also relevant for thermal stability of the csk - containing proteins ( csk proteins , for short ) .
the smd simulations need to fit in the time scale of several tens of ns and thus have to be performed at pulling speeds which are orders of magnitude greater than what can be achieved in experiments or in coarse grained simulations .
the basic pulling velocity used here is 0.04 / ps . this necessarily leads to the values of @xmath1 which are much larger than experimental , typically an order of magnitude larger @xcite .
thus our results on @xmath1 are also overblown for this reason and have to be scaled down .
we show that the functional form of this scaling down is complicated as it involves a crossover between a logarithmic and linear speed dependence .
the latter is expected at high speeds when the drag force dominates .
nevertheless , the values of @xmath1 obtained suggest that csk comes with a much larger mechanostability than the shear - based clamp also at the experimental speeds .
for many csk proteins there is a considerable variation between trajectories in the values of @xmath1 .
pinpointing a specific force - based ranking of the proteins is difficult without generating significant statistics of the trajectories .
it need not coincide with the one obtained by using the coarse grained model ( and based on the dominant pathways ) . on the other hand , the ranking of the proteins with the shearing mechanical clamp , and listed in table 1 , is the same .
in addition to providing the force - displacement ( @xmath3 ) trajectories for typically two trajectories for each protein studied , we also characterize the geometry and sequential make - up of the resulting mechanical clamps and study rupturing of the hydrogen bonds .
table 1 lists 21 proteins that are studied in this paper .
the proteins there are arranged according to the value of @xmath1 as obtained in the coarse grained model @xcite .
the energies and forces of that model are determined in terms of an energy parameter @xmath4 which measures the depth of the potential well in the lennard - jones - like native contacts .
the forces written in table 1 are converted to the experimental units using @xmath4/=110@xmath530 pn the result obtained by comparing the theoretically determined values of @xmath1 to the established experimental data @xcite .
notice a considerable error bar in the conversion factor .
another reason for not treating the protein ranking in table 1 in a verbatim way is that for the csk proteins at the high end of the force , there is a considerable trajectory dependence in the values of @xmath1 both in the coarse grained and all - atom models .
even though the coarse grained values in table 1 are obtained through averaging over 10 trajectories , a substantial dispersion in the csk cases remains .
the important point is that among the 21 proteins listed in table 1 , 15 have the csk mechanical clamp as determined in the coarse grained model .
they are ranked 1 - 13 and then 16 and 18 .
there are more than 100 other csk proteins in the full set of 17 134 considered in ref . @xcite
but they are expected to correspond to lower forces , at least within the coarse grained model .
the protocol of our all atom simulations is explained in the methods section .
the resulting @xmath3 plots are displayed in figs.[three ] through [ shear ] .
full extension is never reached as this would require breaking disulfide bonds .
figures [ three ] and [ next ] refer to the csk proteins whereas fig.[shear ] to the proteins with the shear mechanical clamp . in the latter case , the values of @xmath1 are an order of magnitude smaller and the trajectory dependence is substantially weaker .
previous all - atom studies @xcite for titin indicate that @xmath1 obtained at such speeds is an order of magnitude higher than the experimental value of 204 pn @xcite . among the 15 csk proteins , five ( 1cz8 , 2gh0 , 1rew , 1m4u , and 3bmp ) have both trajectories corresponding to @xmath1 of order 15 nn whereas the remaining ten have at least one in the range of 30 nn as in the case of 1bmp ( bone morphogenic protein ) which is the mechanostability leader in the coarse grained evaluation .
larger statistics of ten trajectories have been obtained for two proteins , 1fzv and 2gyz , as shown in panels ( g ) and ( h ) of fig.[next ] .
they indicate that @xmath1 takes values in the whole range in which the upper reach is about twice as large as the lower reach .
2gyz is seen to have more trajectories with forces lower than 1fzv whereas in the coarse grained model the two systems appeared to behave similarly .
all of these force values are significantly larger than those associated with the shearing clamp shown in fig.[shear ] .
the largest value of @xmath1 in this group is @xmath6 3 nn as observed for the homologous pair 1c4p and 1qqr ( panels ( a ) and ( b ) in fig.[shear ] ) . despite the spread in @xmath1 between the trajectories for the csk proteins , many @xmath3 plots look essentially similar indicating existence of a single pathway .
however , some proteins , like 2gyz ( panels ( a ) and ( h ) of fig.[next ] ) come with two distinctive patterns and thus two pathways : one with two force peaks and @xmath1 exceeding 30 nn and another with four peaks and @xmath1 of order 15 nn
. we shall discuss why these differing patterns arise later on .
fig.[speed ] addresses the issue of the pulling rate dependence for titin ( 1tit ) and human placenta growth factor ( 1fzv ) .
the bottom panel shows the @xmath3 curves obtained at several speeds : the bigger the speed the higher the curve and the higher the second force maximum .
we focus on the first maximum as its physics is governed by the well studied shear mechanical clamp .
the corresponding values of @xmath1 are displayed in the top panel .
they suggest existence of a gradual crossover from the logarithmic to linear dependence on the speed in the regime of the pulling rates studied . at the slowest speeds ,
the data points appear to be consistent with a proper logarithmic extrapolation to forces obtained experimentally .
nothing is known about the speed dependence of @xmath1 for the csk proteins .
the top panel of fig.[speed ] displays results for 1fzv based on one ( for smaller speeds ) or two trajectories .
the statistics involved are too poor to guess the functional dependence , however , the titin - like logarithmic dependence on going to the experimental speeds remains a possibility . thus scaling down of our results for the csk proteins using
the titin - like logarithmic dependence seems to be a reasonable first guess .
if so then @xmath1 of 30 nn may probably correspond to the experimental 1500 pn .
such forces are comparable to those needed to rupture covalent n - c and c - c bonds : 1500 and 4500 pn respectively @xcite . on the other hand , the coarse grained simulations suggested values which would not exceed 1100 pn . at this stage ,
the coarse grained model is probably more reliable in this respect .
it should be noted that the coarse grained simulations have been performed at speeds much closer to those expected at the experimental conditions . for the csk proteins ,
the largest @xmath1 obtained is about 5 times larger than the @xmath1 for titin . at the speeds considered in the all - atom simulations , the derived forces are typically 10 - 20 times larger than for titin .
the extra factor of 2 - 4 compared to the coarse grained model should be attributed to the different resolution of the two models and thus to the different ways in which the cystine ring is penetrated . as will be discussed later
, the atomically represented ring distorts to an oblique form significantly and allows for penetration to occur only at an acute angle .
this kind of penetration is sensitive to what atoms are facing the plane of the ring .
in contrast , the cystine ring in the coarse grained model remains more circular and penetration is more vertical .
we do not expect any other changes in the physics of dragging a segment of the backbone through the cystine ring to be relevant . in order to set the stage for the discussion of the working of the csk mechanical clamp
, we first discuss the shear clamp .
the largest value of @xmath1 within this mechanism is predicted to arise in the blood clotting streptokinase @xmath2-domain 1c4p ( we consider chain a ) and its very close companion hydrolase activator streptokinase domain b 1qqr .
a sequential alignment indicates that 1c4p has two more residues at the n terminus , 1qqr has three more at the c terminus and there are four sites at which the sequences differ .
the high sequence identity results in a very close structural similarity .
the @xmath3 curves for the two proteins are shown in the first two panels in fig.[shear ] .
the values of @xmath1 show only a small spread between two trajectories .
they are 3171 and 3315 pn for 1c4p and 3048 and 3037 for 1qqr indicating a bit larger mechanostability of the former . in both cases
, the mechanical clamp corresponds to the schematic representation as shown in the top panel of fig.[clamps ] .
the way this mechanical clamp works is shown in more details in fig.[1c4pstory ] for 1c4p at four stages of unraveling .
the major shearing action takes place between the parallel and long @xmath2-strands @xmath7 ( 11 amino acids ) and @xmath8 ( 13 amino acids ) .
there is extra shear between strands @xmath7 and @xmath9 as well as between @xmath10 and @xmath8 .
the top left panel of fig.[wodor ] and fig.[hydro1c4p ] illustrate the nearly monotonic disappearance of the hydrogen bonds in 1c4p as a result of pulling . at @xmath1 ,
there are no obvious indicators of rupture in the total number of the hydrogen bonds .
however , such indicators arise when one looks into the number of the hydrogen bonds in specific pairs of the strands ( fig.[wodor ] ) .
it is seen that the pairs @xmath7-@xmath8 and @xmath7-@xmath9 lose their couplings at @xmath1 , though some of them reappear temporarily later . on the other hand ,
the sheet @xmath10-@xmath8 loses fewer couplings at @xmath1 suggesting their stabilizing role .
they are still seen as partly operational in the lower panel of fig.[hydro1c4p ] which shows the placement of the hydrogen bonds at @xmath11=80 . at this stage ,
the hydrogen bonds in the 15-residue long @xmath12-helix remain nearly intact .
the mechanical behavior of the remaining proteins shown in fig.[shear ] is similar to that found in 1c4p even though the corresponding structures are different and the values of @xmath1 are smaller . here
, we discuss the diadenosine tetraphosphate hydrolase 1f3y for which the maximum force ( of 2553 2624 pn ) arises not at the first but at the third force peak .
there are five helices in 1f3y ( 53 - 65 , 85 - 95 , 116 - 118 , 139 - 145 , and 148 - 164 ) and seven @xmath2-strands ( @xmath13 is 36 - 41 , @xmath14 is 44 - 46 , and the remaining ones are listed in the caption to fig.[wodor ] .
the two terminal ones unfold at the beginning and generate a small force peak at @xmath11 around 50 .
these events are followed by rupturing the sheet formed by the antiparallel strands @xmath15 and @xmath16 .
the second small peak ( around @xmath11=160 ) comes from unfolding the bonds related to residues in the segment between 113 and 130 which involves loops and the third short helix .
panel ( b ) of fig.[wodor ] demonstrates that the primary shear mechanical clamp is formed by two parallel strands @xmath17 and @xmath18 with some contribution from the antiparallel @xmath19 and @xmath18 .
the bonds between @xmath15 and @xmath16 are seen to disappear much earlier .
we can distinguish two characteristic groups of the @xmath3 curves : with one force peak ( 1fzv , 1qty , 1cz8 , 1wq9 , 1flt , 1vpf , 1wq8 ) and with multiple peaks .
this feature originates primarily from the size of the slipknot loop . in all of the single maxima cases ,
the proteins have a short and stiff slipknot loop comprising of order six amino acids .
larger slipknot loops ( typically of order 30 amino acids ) are more flexible and unfavorable steric clashes with the cytine ring get split into several events in various ways and are protracted in time as seen in figs.[three ] and [ next ] .
such long loops arise in 1bmp , 1lxi , 1m4u , 1rew , 2bhk , 2gh0 , 2gyz , and 3bmp .
however , multiple peaks may sometimes also arise with short loops , as discussed in the next section .
this effect relates to sizes of the amino acids in the loop .
we first consider two examples of proteins for which we have observed only single force peaks : placenta growth factor 1fzv and vascular endothelial growth factor 1vpf ( both human ) .
in the case of 1fzv , the ring is formed from eight amino acids cys66thr67gly68cys69cys70cys113glu112cys111 that make a ring because of the disulfide bonds cys66-cys111 and cys70-cys113 .
the slip - loop involves the segment leu75his76cys77val78pro79 ( see also table 2 ) .
dragging is carried out by the disulfide bond cys35-cys77 . in the case of 1vpf
, the ring is created by the disulfide bonds cys57-cys102 and cys61-cys104 that link the backbone segments gly58gly59cys61 and glu103 into a tight ring .
the slip - loop is formed by another segment of the backbone which consists of amino acids leu65glu67cys68val69pro70 ( see also table 2 ) .
dragging is carried out by the cys36-cys68 disulfide bond .
the stages of unfolding of 1fzv are illustrated in fig.[stages ] for the trajectory with @xmath1 of 30 390 pn ( the other one corresponds to 26 980 pn ) .
dragging through the cystine ring is seen to take place abruptly it is accomplished when @xmath11 varies by less than about 3 . in the native state ,
the slip - loop is stabilized by 12 hydrogen bonds between two long parallel @xmath2-strands ( amino acids 7591 and 99115 ) .
almost all of them get ruptured on dragging the slip - loop through the ring as illustrated in fig.[hydro1fzv ] .
we have demonstrated within the coarse grained model @xcite that replacing all of the slip - knot related contacts ( inside the slip - loop and between the slip - loop and the ring ) by repulsive contacts does not affect @xmath1 in a significant manner , indicating that the csk mechanical clamp operates mostly through overcoming the sterical constraints .
panel ( c ) in fig.[wodor ] shows that the total number of hydrogen bonds undergoes a sudden drop when the csk clamp gets ruptured because of a major transformation of the whole structure .
panel ( d ) in this figure shows that the same observation also applies to the multiple - peak csk proteins like 1bmp .
this has not been the case in the shear mechanical clamps ( panels ( a ) and ( b ) of fig.[wodor ] ) because the unraveling at @xmath1 is relatively more local and much more gentle .
we now consider the neurotrophic growth factor 2gyz .
it is an example of a csk protein which has ( at least ) two unfolding pathways , each with several force peaks ( another such example is 1m4u ) .
on one pathway @xmath1 is nearly twice as large as on the other .
the cystine ring consists of eight amino acids : cys32ser33gly34ser35cys36cys100gly99cys98cys32 .
it is indicated by an ellipse in figs.[slabas ] and [ ellipse ] .
fig.[slabas ] shows two possible ways of crossing the ring by the pulled cysteine whereas fig.[ellipse ] , corresponding to the right - hand panels of fig.[slabas ] , defines the geometrical parameters of the csk that will be discussed when describing temporal changes in the structure of the csk mechanical clamp .
the slip - loop is long and is pulled by cys70 which forms a disulfide bond with cys5 which is just near the n - terminus .
one branch of the slip - loop is 27-residue long ( from arg37 to cys70 ) and the other is 33-residue long ( from cys70 to ala97 ) .
the relevant segment of the slip - loop that has to squeeze through the ring is then gln67pro68cys69cys70arg71pro72thr73 .
the steric hindrance arises when this segment of the slip - loop passes near the cys32ser33gly34 fragment of the ring .
depending on the history of thermal fluctuations , there are only two ways in which the slip - loop is facing the ring : either the first neighbor cys32 is closer to the ring or the other first neighbor arg97 is the one which is closer .
the two ways differ by a rotation of 180@xmath20 .
if the approach to the plane of the ring the were vertical , the two amino acids would enter the ring simultaneously .
instead , the symmetry is broken because the approach is at an acute angle since the pulling direction becomes increasingly parallel to the longer axis of the ellipse that represents the narrowing ring .
the first way , shown in the left panel of fig.[slabas ] , yields the smaller force of 15 nn since cysteine is the smaller sized amino acid of the two . the second way , shown in the right panel , yields @xmath1 of 30 nn .
the small force case must necessarily come with the striated @xmath3 pattern since the bigger amino acid must follow immediately next which involves further forced expansion of the ring .
the separation into large force and small force pathways , independent of the number of the force peaks , should be observed for each of the csk proteins studied here .
however , for short slip - loops one way is expected to dominate , as determined by its sequential makeup , but both possibilities are seen in stretching of 1vpf and 1wq9 . for long slip - loops , there is no particular preference for selection of the initial orientation on approaching the ring .
it should be noted that the role of fluctuations depends on the speed of pulling . at small pulling speeds
, conformations are expected to get equillibrated better which should reduce choices between various trajectories .
we first consider proteins with short slipknot loops such as 1fzv and 1vpf .
figs.[cleat ] and [ cleat1 ] illustrate the jamming process through the evolution of the clinched fragment of 1fzv in the immediate vicinity of the force peak .
the relaxed , native structure of the fragment is shown at the top for comparison .
the fragment comprises the full cystine ring ( in green ) and five amino acids of the slip - loop near its dragging head
cys75 ( in yellow ) .
the three - residue long fragments of the slip - loop on the left and on the right of the pulling cysteine are listed in table 2 .
the table makes use of the sequential alignments as in refs .
note that for each of the entries listed in the table , the second sequential neighbor of the pulling cysteine , when counting away from the n - terminus , is always a proline which initiates a @xmath2-sheet . in all but three cases , the proline is followed by threonine . between the pulling cysteine and the proline
, the residues can be either small or large .
the cysteine is a relatively small amino acid so it slides into the cystine ring fairly easily .
this step , however , bends the slip - loop and raises tension . on further pulling , the neighboring residues ( his76 and val78 shown in brown and pink in the figures respectively ) have to penetrate the ring either his76 first or val78 first with the other of the two following immediately afterwards . the pathway with his75 entering first is dominant because the corresponding conformation is more native - like .
it is shown in fig.[cleat ] whereas the other pathway in fig.[cleat1 ] .
the penetration is resisted by the ring and the system is jammed .
this stage determines the value of @xmath1 .
the dominant pathway yields a larger @xmath1 since histidine is bigger in size than valine because of its aromatic ring . in this pathway , when his76 stays jammed , its neighbors leu75 and asn74 together with val78 ( which is flanking cys77 from the other side ) form a compact complex .
this complex has an attached `` hook '' , pro79 ( in blue ) , which is pivoting on the ring . immediately after his76 passes through the cystine ring , this complex passes the ring as well .
this event determines the largest force and is very quick .
the whole process generates a single maximum .
once the bulky leu75 , asn74 and val78 complex crosses the loosened ring , the remaining slipknot residues slide through easily in a swinging motion . when the full length of the slip - loop is exhausted and stretching affects only covalent bonds from the protein backbone , the tension grows again .
the pathway in which val78 is the head amino acid which penetrates the cystine ring ( fig.[cleat1 ] ) is similar with the role of the neighboring amino acids interchanged .
for instance , it is leu75 which forms a hugging hook on the ring now . the single peak mechanism involving a passage of a bulky plug is also observed in 1qty , 1wq8 , 1flt , 1fzv , 1vpf , and 1wq9 .
in each of these proteins , the other second neighbor of the pulling cysteine is bulkier than proline and , in the dominant pathway , it is the larger second residue which penetrates the first and paves the way for the smaller companion .
it should be pointed out that despite the multitude of the possible values of @xmath1 , as seen , e.g. in fig.[next]g , there are only two pathways of the ring penetration .
the differences in the values arise merely from minute differences in the angle of approach that get emphasized by the very magnitude of the stresses involved .
multiple peaks arise when the slipknot loop is long and when the amino acids enter in order in which their geometrical sizes keep increasing because the subsequent residues keep forcing the cystine ring ever more open .
the resulting sawtooth - like @xmath3 pattern is encountered , for instance , in 1bmp , 2gyz , 2bhk , 2gh0 , 1rew , 3bmp .
again , there are basically only two pathways , albeit multipeaked .
it is interesting to observe that all of these multiple - peak proteins contain another cysteine .
it should be noted that this neighboring cysteine may be involved in forming a dimer state with a similar companion . in the dimeric state , the slip - loop motion through the ring would be prohibited .
we have found that the passage of the slipknot through the cystine ring is generally associated with a rapid rise in the bond and bond - angle contributions to the total energy .
all remaining contributions to the total energy , such as electrostatic , van der waals and those associated with the dihedral and improper angles remain fairly unaffected by crossing of the force peaks .
finally , we discuss the geometry of the csk mechanical clamp in terms of geometry as characterized by the effective curvatures .
previously @xcite , we have proposed that the slip - loop can be driven through the cystine ring provided @xmath21 where @xmath22 and @xmath23 are the radii of curvature of the slip - loop and the ring respectively .
we have estimated these radii to be of order 7 and 3 correspondingly .
symbol @xmath24 denotes the effective thickness of the slip - loop and @xmath25 that of the ring .
both have been assessed to be around 2.5 . here , we reexamine this condition by determining the local radii of curvature in a way used in the context of the tube picture of proteins @xcite , i.e. by finding a circle which goes through the corners of a triangle set by three consecutive c@xmath26 atoms .
fig.[curvature ] shows the thus determined @xmath22 and @xmath23 for 1bmp and 1fzv , together with the semimajor axis , @xmath27 , of the ellipse assigned to the cystine ring . in the native state ,
@xmath27 and @xmath23 are both close to 5 for both proteins .
@xmath22 is found to be about 3 . on pulling ,
all of these parameters evolve . until reaching @xmath1
, @xmath27 generally grows whereas @xmath23 and @xmath22 go down . at @xmath1 ,
the radii jump upward , indicating an escape from the constriction , and @xmath27 jumps downward . at each stage ,
the condition given by eq .
[ condition ] is satisfied if the tube thickness is neglected . in order to estimate the effect of the thickness in a description which does not invoke an explicit tube , we replace one of the corners of the triangle by another atom . for a new effective @xmath22
we take the backbone oxygen atom associated with the pulled amino acid ( the line denoted by o ) . for a new effective @xmath23
we take the sulphur atom ( the line denoted by s ) residing on the next coming amino acid ( see also fig.[ellipse ] ) .
these displaced - atom lines generally follow the behavior of the c@xmath26 based curvatures , at least until reaching @xmath1 .
however , they illustrate the magnitude of the effects associated with the tube thickness .
the shifts are seen to be substantial and indicate that the condition to go through is borderline or even prohibitive if viewed from the tube - like perspective .
we conclude that the atomic - level features and structure are important for the csk mechanism to work when @xmath22 and @xmath23 are comparable in size .
our all - atom simulations confirm existence of huge mechanostability of the csk proteins that has been predicted by using the coarse grained model .
it should be pointed out that there are differences between the way the slip - knot passes through the cystine ring as described by the two models .
the all - atom ring is much tighter because of the bigger excluded volume generated by the atoms in the side groups .
thus the passage process involves significant distortion of the ring so that it becomes a narrow ellipse . as a result
, only one amino acid can squeeze through at a time . in contrast , the c@xmath26-built ring in the coarse - grained model remains fairly circular allowing for the simultaneous passage of up to two amino acids that are the first neighbors of the pulling cysteine . the variations in the magnitude of @xmath1 between the trajectories in the coarse grained model , also even up to a factor of 2 , arise now not that much from the selection of the amino acid that goes through first but from the possibility of dragging of one or two neighboring amino acids initially .
our all - atom simulations validate the results obtained through coarse grained modeling in a qualitative way as they support possible existence of the csk mechanical clamp .
the ultimate test , however , should come from experimental pulling studies and it is the experiments that should establish the true values of @xmath1 .
our theoretical estimates of @xmath1 are not sufficiently precise and should be considered just as indicative of large forces that are expected to be found .
the difficulty that such experiments might encounter is the resolvability of the csk - related force peaks from the background generated by stretching of the peptide bonds .
this stability of the ck proteins appears to be related to the high mechanical stability found in this paper .
specifically , we have shown that the buildup in the tension results from jamming .
the sequential neighbors of the pulling cysteine are of a size that prevents them from penetrating the cystine ring without forcing .
thus we expect that thermal fluctuations would not be able to form a slipknot and change the conformation .
similar stability features should also be operational in the cck and ick proteins .
thus even though mechanical stability , in general , need not be related to stability against thermal fluctuations ( see , e.g. @xcite ) there is a relationship in the case of the ck proteins . in particular , it has been shown @xcite that a removal of the disulfide bonds in the ck - containing vegf protein through mutations lowers the melting temperature significantly . in a recent survey of multidomain proteins @xcite
still new mechanical clamps have been identified by using the coarse grained model . among them are two kinds of tensile clamps
one simple , involving contacts between two domains , and another in which a part of the clamp is immobilized by a non - cystine knot - loop .
it would be interesting to investigate atomic aspects of the workings of such clamps by using all - atom simulations as done in this paper .
all simulations have been performed using namd 2.6 code @xcite with the all - atom charmm27 force field @xcite .
all of the simulations have been carried out using the same computational protocol .
the initial structure is downloaded from the pdb database @xcite and then placed within a box with rigid water molecules .
the layer of the water molecules is at least 8 thick . na@xmath28 and
cl@xmath29 ions are introduced into the system at concentration of 0.5 mol / l ( between 10 and 20 ions in the box ) . in order to neutralize the system , additional several ions of one sign
are added using a script embedded in the vmd code @xcite .
the molecules of water are then moved for 100 steps per atom in order to minimize the energy and then equilibrated at the temperature of 300 k during 100 ps of the time evolution . after this stage ,
the protein atoms are allowed to move and the whole system undergoes 1000 steps of the energy minimization followed by heating from 0 k to 300 k during 50 ps in a stepwise fashion .
the last step before stretching consists of 1 ns equilibration at 300 k by using the langevin dynamics .
the time step is set at 1 fs . in analogy to the coarse - grained approach @xcite , the smd simulations @xcite are implemented by placing the n - terminal c@xmath26 atom at a fixed location and by attaching an elastic spring to the c@xmath26 atom at the c - terminal amino acid .
the elastic constant is equal to 4 kcal / mol / @xmath30 ( 277.9 pn / ) .
the other end of the spring is made to move with the pulling velocity @xmath31 of 0.04 / ps ( 4 m / s ) .
the pulling direction is set to go through the line connecting the first and last c@xmath26 atoms . in all cases , at least two 10 ns smd simulations that evolve from the same starting configuration are carried out .
this is the configuration obtained at the end of the initial langevin dynamics .
various trajectories arise due to differences in the time dependence of the langevin noise applied .
graphical analysis are performed with the vmd 1.8.7 software @xcite .
forces and local geometries are analyzed by using homemade scripts . in order to determine whether two heavy atoms are connected by a hydrogen bond we check whether the distance between them does not exceed 3.3 and the angle donor ( like o )
acceptor ( like n ) does not exceed 25@xmath20 .
these bonds rupture often and then reform .
the forces and the numbers of the hydrogen bonds are time averaged over the pulling distances of 0.5 . in the plots
, these data undergo further smoothing out .
very useful discussions with j. i. sukowska at the beginning of the project are appreciated .
this work has been supported by the grant n n202 0852 33 from the ministry of science and higher education in poland , the ec funmol project under fp7-nmp-2007-small-1 and through the innovative economy grant ( poig.01.01.02 - 00 - 008/08 )
. 99 a. galera - prat , a. gomez - sicilia , a. f. oberhauser , m. cieplak , and m. carrion - vazquez , curr . op .
. biol . * 20 * , 63 - 69 ( 2010 ) .
m. carrion - vazquez , m. cieplak , and a. f. oberhauser , single molecule biophysics and unfolding forces , in : encyclopedia of complexity and systems science , editor - in - chief r. a. meyers , springer , new york ( 2009 ) , isbn:978 - 0 - 387 - 75888 - 6 .
n. crampton and d. j. brockwell , curr .
. biol . * 20 * , 508 - 517 ( 2010 ) .
m. carrion - vazquez , a. f. oberhauser , t. e. fisher , p. e. marszalek , h. b. li et al .
biol . * 74 * , 63 - 91 ( 2000 ) .
j. m. fernandez and h. b. li , science * 303 * , 1674 - 1678 ( 2004 ) .
m. t. alam , t. yamada , u. carlsson , and a. ikai febs lett . * 519 * , 35 - 40 ( 2002 ) .
t. bornschloegl , d. m. anstrom , e. mey , j. dzubiella , m. t. rief , and k. forest , biophys .
j. * 96 * , 1508 - 1514 ( 2009 ) .
a. valbuena , j. oroz , r. hervas , a. m. vera , d. rodriguez , et al .. proc .
u s a. * 106 * , 13791 - 13796 ( 2009 ) .
h. lu and k. schulten , chem .
phys . * 247 * , 141 - 153 ( 1999 ) .
d. j. brockwell , e. paci , r. c. zinober , g. beddard , p. d. olmsted , d. a. smith , r. n.perham , and s. e. radford , nat .
10 * , 731 - 737 ( 2003 ) .
j. i. sukowska and m. cieplak , j. phys .
cond . mat .
* 19 * , 283201 ( 2007 ) .
m. sikora , j. i. sukowska , and m. cieplak , plos comp .
biol . , * 5 * , e1000547 ( 2009 ) .
m. sikora , j. i. sukowska , b. s. witkowski , and m. cieplak , nucl .
( 2010 ) ( published on - line ) .
d. j. craik , n. l. dally , and c. waine , toxicon * 39 * , 43 - 60 ( 2001 ) .
k. berry , m. thomas , b. langley , m. sharma , and r. kambadur , am .
j. physiol . cell .
physiol . * 283 * , c135-c141 ( 2002 ) .
k. j. rosengren , n. l. daly , m. r. plan , c. waine , and d. j. craik , j. biol . chem .
* 278 * , 8606 - 8616 ( 2003 ) .
n. q. mcdonald and w. a. hendrickson , cell * 73 * , 421 - 424 ( 1993 ) .
p. k. pallaghy , k. j. nielsen , and d. j. craik , protein sci , * 3 * , 1833 - 1839 ( 1994 ) .
d. j. craik , n. l. daly , t. j. bond , c. waine , j. mol . biol .
* 294 * , 1327 - 1336 ( 1999 ) .
d. barth , h. j. musiol , m. schuett , s. fiori , a. g. milbradt , c. renner , and l. moroder , chemistry * 9 * , 3692 - 702 ( 2003 ) .
o. carugo , s. lu , j. luo , x. gu , s. liang , s. strobl , and s. pongor , prot .
* 14 * , 639 - 646 ( 2001 ) .
s. iyer , p. d. scotney , a. d. nash , and k. r. acharya , j. mol . biol . * 359 * , 76 - 85 ( 2006 ) .
r. m. stroud and j. a. wells , sci stke 231 : re7 ( 2004 ) . c. combelles , j. gracy , a. heitz , d. j. craik , and l. chiche , proteins : struct .
funct . bio . * 73 * , 87 - 103 ( 2008 ) .
c. w. gruber , m. cemazar , m. a. anderson , and d. j. craik , toxicon * 49 * , 561 - 575 ( 2007 ) .
h. p. erickson , _ science _ * 276 * , 1090 - 1092 ( 1997 ) .
w. a. linke , m. ivemeyer , n. olivieri , b. kolmerer , j. c. ruegg , and s. labeit , _ j. mol .
biol . _ * 261 * , 62 - 71 ( 1996 ) .
l. tskhovrebova , k. trinick , j. a. sleep , and m. simmons , _ nature _ * 387 * , 308 - 312 ( 1997 ) .
m. s. z. kellermayer , s. b. smith , h. l. granzier , and c. bustamante , _ science _ * 276 * , 1112 - 1116 ( 1997 ) .
m. rief , m. gautel , f. oesterhelt , j. m. fernandez , and h. e. gaub , _ science _ * 276 * , 1109 - 1112 ( 1997 )
. m. carrion - vasquez , a. f. oberhauser , s. b. fowler , p. e. marszalek , s. e. broedel , j. clarke , and j. m. fernandez , _ proc .
( usa ) _ * 96 * , 3694 - 3699 ( 1999 ) .
a. f. oberhauser , p. k. hansma , m. carrion - vazquez , and j. m. fernandez , _ proc .
( usa ) _ * 98 * , 468 - 472 ( 2001 ) .
a. f. oberhauser , p. e. marszalek , m. carrion - vazquez , and j. m. fernandez , _ nature struct .
* 6 * , 1025 - 1028 ( 1999 ) .
s. b. fowler , r. b. best , j. l. toca herrera , t. j. rutherford , a. steward , e. paci , m. karplus , and j. clarke , _ j. mol
_ * 322 * , 841 - 849 ( 2002 ) .
m. grandbois , m. beyer , m. rief , h. clausen - schaumann , and h. gaub , science * 283 * , 1727 - 1730 ( 1999 ) .
o. avisian kretchmer and a. j. w. hsueh ,
endocrinology * 18 * , 1 - 12 ( 2004 ) .
j. r. banavar , t. x. hoang , a. maritan , f. seno , and a. trovato , phys .
e * 70 * , 041905 ( 2004 ) .
t. x. hoang , a. trovato , f. seno , j. r. banavar , and a. maritan , proc .
u s a * 101 * , 7960 - 7964 ( 2004 ) .
m. cieplak and j. i. sukowska , j. chem . phys . *
123 * , 194908 ( 2005 ) .
y. a. muller , c. heiring , r. misselwitz , k. welfle , and h. welfle , j. biol .
chem . * 277 * , 43410 - 43416 ( 2002 ) .
m. sikora and m. cieplak , nucl .
39 , d443 ( 2011 ) j. c. phillips , r. braun , w. wang , j. gumbart , e. tajkhorshid , c. chipot , r. d. skeel , and l. s. k. kale , j .
comput . chem .
* 26 * , 1781 - 1802 ( 2005 ) .
a. d. mackerell , d. bashford , m. bellot , et al .
, j. phys .
b , * 102 * , 3586 - 3616 ( 1998 ) .
h. m. berman , j. westbrook , z. feng , g. gilliland , t. n. bhat , et al.,nucl .
acids res .
* 28 * , 235 - 242 ( 2000 ) .
w. humphrey , a. dalke , and k. schulten , j. mol . graph .
* 14 * , 27 - 8 , 33 - 8 ( 1996 ) .
m. cieplak , t. x. hoang , and m. o. robbins , proteins : struct .
funct . bio .
* 56 * , 285 - 297 ( 2004 ) .
b. isralewitz , j. baudry , j. gullingsrud , d. kosztin , and k. schulten , j. mol .
model * 19 * , 13 - 25 ( 2001 ) .
h. lu , b. isralewitz , a. krammer , v. vogel , and k. schulten , biophys .
j. * 75 * , 662 - 671 ( 1998 ) .
e. paci and m. karplus , j. mol
. biol . * 288 * , 441 - 459 ( 1999 ) .
l. peplowski , k. kubiak , and w. nowak , chem .
* 467 * , 144 - 149 ( 2008 ) .
w. nowak and p. marszalek , molecular dynamics simulations of single molecule atomic force microscope experiments . in current trends in computational chemistry , ed .
j. leszczynski , 47 - 83 ( 2005 ) , world scientific , singapore .
= 3.2 in = 3.5 in = 4.6 in = 4.5 in = 4.0 in
table 1 : the predicted list of the strongest proteins .
@xmath32 [ ] & @xmath33 [ ] & cath & scop + 1 & * 1bmp & 104 & * 1120 & 23.2 & 176.0 & 2.10.90.10 & g.17.1.2 + 2 & * 1qty & 95 & * 980 & 72.1 & 108.3 & 2.10.90.10 & b.1.1.4 + 3 & * 2bhk & 119 & * 800 & 26.5 & 129.3 & & + 4 & * 1lxi & 104 & * 800 & 22.5 & 126.4 & & g.17.1.2 + 5 & * 1cz8 & 107 & * 700 & 76.5 & 149.4 & 2.10.90.10 & b.1.1.1 + 6 & * 2gh0 & 219 & * 640 & 25.9 & 104.6 & & + 7 & * 1wq9 & 100 & * 610 & 72.0 & 131.8 & 2.10.90.10 & g.17.1.1 + 8 & * 1flt & 107 & * 610 & 75.6 & 128.6 & 2.10.90.10 & b.1.1.4 + 9 & * 1fzv & 117 & * 590 & 90.4 & 130.3 & 2.10.90.10 & g.17.1.1 + 10 & * 2gyz & 100 & * 590 & 14.4 & 93.3 & 110 & & + 11 & * 1rew & 103 & * 580 & 21.7 & 92.7 & 2.10.90.10 & g.7.1.3 + 12 & * 1m4u & 139 & * 580 & 52.1 & 114.5 & 2.10.90.10 & g.17.1.2 + 13 & * 1vpf & 94 & * 580 & 68.1 & 121.6 & 2.10.90.10 & g.17.1.1 + 14 & * 1c4p & 137 & * 560 & 106.0 & 138.8 & 3.10.20.180 & d.15.5.1 + 15 & * 1qqr & 138 & * 550 & 110.3 & 134.8 & 3.10.20.180 & d.15.5.1 + 16 & * 3bmp & 114 & * 550 & 33.0 & 96.1 & 2.10.90.10 & g.17.1.2 + 17 & * 1j8s & 193 & * 540 & 77.9 & 99.3 & 2.60.40.1370 & b.2.3.3 + 18 & * 1wq8 & 96 & * 540 & 82.6 & 109.8 & 2.10.90.10 & g.17.1.1 + 19 & * 1j8r & 193 & * 530 & 77.7 & 97.0 & 2.60.40.1370 & b.2.3.3 + 20 & * 1f3y & 165 & * 530 & 284.7 & 336.0 & 3.90.79.10 & d.113.1.1 + 3580 &
* 1tit & 89 & * 230 & 55.3 & 43.3 & 2.60.40.10 & b.1.1.4 + * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * table 1 .
@xmath1 is obtained within the structure - based coarse grained model in ref .
the model is defined in terms of the energy parameter @xmath4 that determines the depth of the potential well in the native contacts .
the conversion to pn is by taking the average relationship @xmath34 @xmath35 pn . the pulling velocity velocity is @xmath60.005 / ns . the first column indicates the ranking of a model protein , the second the pdb code , and the third the number of the amino acids that are present in the structure used .
@xmath32 denotes the end - to - end distance at which the maximum force arises and @xmath33 the corresponding tip displacement .
the last two columns give the leading cath and scop codes .
bmp-7 & 1bmp & + bmp-7 & 1m4u & + bmp-7 & 1lxi & + bmp-2 & 3bmp & + bmp-2 & 1rew & + vegf & 1qty & + vegf & 1cz8 & + vegf & 1flt & + vegf & 1vpf & + vegf & 1wq9 & + vegf toxin & 1wq8 & + human gdf & 2bhk & + gdnf family receptor @xmath12 - 3 & 2gh0 & + plgf & 1fzv & + artn isoform 3 & 2gyz & + table 2 .
abbreviations used in the table : bmp - bone morphogenetic protein , vegf - vascular endothelial growth factor , gdf - growth and differentiation factor , gdnf - glial cell line - derived neurotrophic factor , plgf - placenta growth factor , artn - neurotrophic factor artemin .
|
a recent survey of 17 134 proteins has identified a new class of proteins which are expected to yield stretching induced force - peaks in the range of 1 nn .
such high force peaks should be due to forcing of a slip - loop through a cystine ring , i.e. by generating a cystine slipknot .
the survey has been performed in a simple coarse grained model . here , we perform all - atom steered molecular dynamics simulations on 15 cystine knot proteins and determine their resistance to stretching . in agreement with previous studies within a coarse grained structure based model ,
the level of resistance is found to be substantially higher than in proteins in which the mechanical clamp operates through shear .
the large stretching forces arise through formation of the cystine slipknot mechanical clamp and the resulting steric jamming .
we elucidate the workings of such a clamp in an atomic detail .
we also study the behavior of five top strength proteins with the shear - based mechanostability in which no jamming is involved .
we show that in the atomic model , the jamming state is relieved by moving one amino acid at a time and there is a choice in the selection of the amino acid that advances the first .
in contrast , the coarse grained model also allows for a simultaneous passage of two amino acids .
@xmath0 e - mail : [email protected] 40 pt
|
the distribution of groups , clusters , and superclusters of galaxies represents a fundamental testing ground for theories of the origin and evolution of structure in the universe . until recently ,
the intersection between observation and theory has largely relied on statistical properties of galaxy spatial / kinematic clustering , along with x - ray determined global icm temperatures and azimuthally averaged radial brightness and temperature profiles .
in contrast , much of the existing observational data indicates that large - scale asymmetric , probably filamentary , structures , which are not easily subjected to statistical definition _ on an individual basis _ , are present in clusters and superclusters ( e.g. , gregory & thompson 1978 ; shandarin 1983 ; de lapparent , geller , & huchra 1986 ; west , jones , & forman 1995 ; west & blakeslee 2000 ) . multi - fiber spectroscopy of galaxies with 400-fiber positioners deployed over 2 fields , as well as the unprecedented combination of spatial resolution and sensitivity in x - rays provided by the _
chandra _ and xmm - newton observatories , are rapidly advancing the observational view of clusters of galaxies .
in addition , numerical simulations , carried out within the framework of a cold dark matter dominated universe in which structure is built up in a hierarchical fashion , have reecently reached the point where large areas , on scales comparable to the largest superclusters , can be simulated in some detail ( pearce 2001 and references therein ) .
furthermore , using parametrized physics of the baryonic component , simulations within the framework of a cold dark matter dominated universe now predict the time evolution of the baryonic material ( kauffmann 1999a , b ; somerville & primack 1999 ; benson 2001 and references therein ) , thus providing a more direct link to the observational data . hence the opportunity now presents itself to make a comprehensive comparison between observations and simulations in the case of _ specific _ clusters / superclusters , thereby providing a more sensitive probe of structure evolution than can be gained from purely statistical considerations .
recent studies of the shapley supercluster , an extremely massive concentrations of galaxies and clusters first pointed out by shapley ( shapley 1930 ; raychaudhury 1989 ) , represent one example of the possibilities provided by an intensive campaign of multi - fiber spectroscopy , as well as of x - ray and radio imaging .
these studies have produced a wealth of data on the spatial / kinematic structures present in the shapley supercluster among the galaxies ( e.g. , quintana 2000 ; drinkwater 1999 ; bardelli , zucca , & baldi 2001 ) and on the hot icm ( e.g. , ettori 2001 ) , as well as results on the the emission line and radio properties of cluster galaxies ( baldi , bardelli , & zucca 2001 ; venturi 2001 ) . in the process , a variety of structures have been found , and the influence of environment on both galaxy and radio source evolution has been partially elucidated . in this paper
we present new observational data on a double galaxy cluster , a3128/a3125 , which is itself embedded in the massive horologium - reticulum ( h - r ) supercluster . as is discussed later , the h - r supercluster , and the above - mentioned shapley supercluster , represent the two largest cluster concentrations within the local 300 h@xmath2 mpc ] .
the observational data consist of multi - fiber spectroscopy of the double cluster a3128/a3125 with the anglo - australian telescope and 2 field multi - fiber positioner .
in addition , we present x - ray imaging of the the central regions of a3128 and a3125 with the chandra x - ray observatory .
radio observations obtained with the molonglo observatory synthesis telescope and the australia telescope compact array are also presented .
these observations have revealed a variety of substructures , both in galaxy groupings and in the multiple x - ray peaks detected , which in responding to the large gravitational acceleration provided by the h - r supercluster , reach unusually high infall velocities into a3128 .
thus the connection between a3128/a3125 and the surrounding h - r supercluster provides an opportunity to characterize substructure merging on a variety of scales and unusual conditions . in
2 we present the multi - fiber spectroscopy and x - ray and radio data . in
3 we discuss the kinematic structures revealed by the multi - fiber spectroscopy , while in 4 we discuss the observed state of the icm , as inferred from the chandra observations , and in 5 we present the most and atca radio imaging . a discussion of our results is given in 6 ;
specifically , we attempt to produce a unified view of the merging events occurring in a3128/a3125 based on the full set of observations .
multi - fiber spectroscopy in a3128 and a3125 ( also known as dc0329 - 53 and dc0326 - 52 from the dressler ( 1980 ) study ) has been carried out using the 400 fiber positioner and double spectrograph system ( 2df ) on the anglo - australian telescope ( aat ) .
the field center for the observations is at 3:29:24 -52:49:45 ( j2000 ) , so as to fully cover the a3128/a3125 double cluster system with the 2 field of 2df .
we obtained a complete catalog of all galaxies down to a magnitude of b@xmath3 @xmath4 18.5 in this field from the ukst survey plates scanned by the supercosmos machine at the royal observatory edinburgh ( hambly 1998 , 2001 ) .
discrimination between stars and galaxies is carried out by the supercosmos software , thus we restricted our sample to those objects identified as galaxies in the supercosmos catalog .
a total of 698 galaxies to the limit b@xmath5 @xmath4 18.5 were found .
however , spectroscopy of 72 galaxies has previously been obtained in the double cluster a3128/a3125 by caldwell & rose ( 1997 ) , and 193 galaxies have been observed in the eso nearby abell cluster survey ( enacs ; katgert 1996 , 1998 ) . since 42 galaxies
are common to caldwell & rose ( 1997 ) and enacs , a total of 223 velocities were available prior to our study .
the enacs velocities are concentrated in a3128 , hence it was highly desired to increase the number of velocities in a3125 to be able to study its kinematic properties .
we excluded from our supercosmos - based catalog the 72 galaxies previously observed by caldwell & rose ( 1997 ; where the galaxies are referred to by their dressler ( 1980 ) numbers ) , as well as 22 background and foreground galaxies from the enacs survey .
we kept the other 171 enacs galaxies , since we desired higher s / n spectra for them , but gave them a lower priority in the fiber assignment process . because 42 of the caldwell & rose ( 1997 )
galaxies are also in the enacs list , those galaxies were also included in the final catalog .
altogether , the number of galaxies in the final catalog was reduced to 640 .
the aat/2df system has 2 identical spectrographs each receiving 200 fibers , for a total of 400 fibers ( lewis 2000 ) . to allocate the 400 fibers , including sky fibers
, we used the 2df configuration program configure , which gave a configuration at the telescope allowing 371 object and 24 sky fibers ( 5 fibers on the a spectrograph were unusable ) .
the a3128/a3125 field was observed in service mode on 21 january , 2001 ( ut ) .
five separate exposures of 30 minutes were taken , using the 1200 lines mm@xmath6 blue grating in first order , yielding a dispersion of 1.11 / pix at the tek5 1024 x 1024 pixel ccd , and a spectral resolution of @xmath02.8 ( or @xmath0200 ) fwhm .
the wavelength range covered is 3750 4850 .
a total of 371 spectra were obtained resulting in 358 measured redshifts ; of these 23 turned out to be misclassified galactic stars .
hence we obtained velocities for a total of 335 galaxies . for 58 of these galaxies , and for two of the stars , velocities have been published previously by either the enacs collaboration ( katgert 1998 ) or caldwell & rose ( 1997 ) . including the other previously available velocities from katgert 1998 and caldwell & rose ( 1997 ) we have in our sample redshifts for 532 galaxies in our supercosmos - derived catalog , for 76% completeness to b@xmath5 @xmath4 18.5 .
the data from each spectrograph were reduced independently using the 2df data reduction system 2dfdr .
the standard automatic reduction procedure was followed which automatically does the debiassing , trim - line map generation , fiber extraction , cosmic ray removal , wavelength calibration and sky subtraction .
the reduced files from each exposure were then combined into single spectra of improved s / n .
heliocentric radial velocities from the absorption line spectra were obtained for all galaxies using the fxcor routine in the iraf rv radial velocity package .
the spectrum of a star that was included in the sample was used as the radial velocity template . the radial velocity zero - point for that star
was determined by cross - correlating its spectrum against that of several stars with known radial velocities in the coude feed spectral library ( jones 1999 ) , which is available from the noao ftp archive . where possible , radial velocities were also measured from emission line features by two of us ( jar and aeg ) independently , by fitting a gaussian profile to all measurable emission lines .
an average offset of 129 between emission and absorption line velocities was found for cases where good velocities can be obtained for each .
the reason for such an offset is unknown .
however , uncertainties of @xmath0100 in the radial velocities are small compared to the large systematic motions found in this paper .
a final radial velocity for each object was thus obtained using the following approach . for cases in which the emission spectrum dominates ,
the final velocity is based entirely on the emission line measurements with a correction of + 129 . in the case that both reliable emission and absorption velocities can be obtained ( i.e. , a standard old - galaxy spectrum is clearly visible underneath the emission ) , we average the emission and absorption measurements , after first correcting the emission measurements by 129 . in table
[ velocities_table ] all radial velocity measurements are listed . in columns ( 1 ) and ( 2 ) are the reference number from our 2df catalog and reference numbers from previous observations respectively .
this is followed by the ra and dec , j2000 , for each galaxy , the b@xmath5 magnitude , and the radial velocity . in column
( 7 ) we tabulate whether the velocity was based exclusively on emission lines ( `` e '' ) , exclusively on absorption lines ( `` a '' ) , or on a combination of the two ( `` ae '' ) . in column
( 8) we give the error in the radial velocity , as given by the cross - correlation routine .
note that this error estimate is invalid for spectra whose velocity determinations are based on emission lines only . for the latter
, error estimates are difficult to produce , especially since they are very dependent on the strength of the emission lines and the s / n ratio in the spectra . in most cases
, however , in which emission dominates , we have been able to locate a line with sufficient s / n ratio that a velocity error in excess of 100 is unlikely .
finally , in column ( 9 ) we tabulate whether any emission line is found in the spectrum . as a test for a zero point uncertainty in our velocities we compared our 2df velocities with those in common with both the enacs ( katgert 1998 ) and caldwell & rose ( 1997 ) samples .
there are 57 galaxies for which we have duplicate measures with enacs . omitting the two cases with large velocity discrepancies
we find a mean zero point difference of -38 between the aat/2df and enacs velocities , with the 2df velocities lower than enacs .
the rms dispersion between the two samples is 72 .
there are also three galaxies in common between the 2df data and that of caldwell & rose ( 1997 ) , for which there is no enacs velocity as well .
our mean 2df velocity is 33 lower than caldwell & rose ( 1997 ) for those three galaxies .
there are also 42 galaxies in common between caldwell & rose ( 1997 ) and enacs .
the mean velocity difference is -21 , in the sense that the caldwell & rose ( 1997 ) velocities are lower than enacs .
thus small velocity zero point errors may exist in the data , but at a level insignificant compared to the large group motions discussed in 3 .
we also measured [ oii]@xmath73727 emission line equivalent widths to compare emission line strengths for galaxies in different parts of the cluster .
unfortunately , for the fainter galaxies in the sample we encountered considerable problems in the continuum levels in the sky - subtracted spectra , especially in the blue end of the spectrum occupied by [ oii]@xmath73727 . in many cases
we find the continuum level is actually negative , leading to an unphysical negative equivalent width for [ oii]@xmath73727 .
we tried to remedy the situation by correlating the supercosmos b@xmath5 magnitude with the flux in the continuum of the spectra near 4500 .
our goal was to find a mean normalization factor between the b@xmath5 magnitude and the continuum flux in the spectra for the brighter galaxies , and then use that normalization factor to allow substitution of the more reliable b@xmath5 magnitude in calculating the [ oii]@xmath73727 equivalent width from the observed instrumental flux in the line and the blue magnitude .
however , there is such a large scatter in the plotted relation between b@xmath5 magnitude and continuum flux in the spectra that we were unable to follow this plan .
thus although we are able to search to sensitive limits ( typically better than 2 fwhm ) in the [ oii]@xmath73727 line , we are unable to produce reliable equivalent widths . as a result , we list only whether or not emission has been detected in column ( 9 ) , but no quantitative information in regard to equivalent width of [ oii]@xmath73727 .
for the same 55 galaxies in common between our sample and the enacs sample we find the following emission line statistics . for 30 galaxies
both enacs and aat/2df spectra show no emission lines , while for 13 galaxies both enacs and aat spectra record emission .
in addition , there are 12 galaxies for which we find emission lines , but enacs does nt .
there are no galaxies for which enacs reports emission but we do not .
hence , our aat/2df spectra are finding emission lines at approximately twice the frequency of the enacs spectra . to further assess this discrepancy , for two galaxies we report emission lines
, enacs does not , and caldwell & rose ( 1997 ) also do not report emission . specifically , the two galaxies are aat#477=enacs#148=cr#80a , for which we find an equivalent width in [ oii]@xmath73727 of 9 , and aat#481=enacs#151=cr#53a , for which we find an equivalent width of 6 .
we have reinspected the argus spectra obtained in caldwell & rose ( 1997 ) , and now see evidence for [ oii]@xmath73727 emission in both these spectra .
both argus spectra in caldwell & rose ( 1997 ) are clearly of substantially lower s / n ratio in the vicinity of [ oii]@xmath73727 , while the aat/2df detections are robust . a comparison between the 2df and argus spectra of the two galaxies is made in fig .
[ emission ] , where the clear detection of [ oii]@xmath73727 emission in the 2df spectra is evident . in short , we conclude that the better s / n ratio ( especially in the blue at [ oii]@xmath73727 ) of the aat/2df spectra has allowed us to probe lower emission levels than in previous analyses .
abell 3128 was observed on 5 may 2000 for a total of 19,513 seconds using the _ chandra _ x - ray observatory with the advanced ccd imaging spectrometer ( acis ) in its default imaging mode ( acis - i ) .
the field center of the observation is 03:30:21.93 -52:31:51.5 , i.e. , approximately midway between the two x - ray peaks previously identified on rosat archival images . in acis - i imaging mode , the primary field
is imaged onto four front - illuminated ccd chips ( referred to as i0-i3 ) , arranged in a 2 x 2 array so as to cover a field of 16x 16 .
the telescope is dithered ( by 16peak to peak ) during the exposure , to provide exposure in the ( @xmath011 ) gaps between the ccd chips .
the results presented here are based on reprocessed data ( revision 2 ) , and the events are filtered to include events with flight grades 0,2,3,4 and 6 and energy @xmath4 10 kev . for the imaging analysis the lower energy range
was restricted to 0.5 to 10 kev .
the background during this observation was free of flares or enhancements , so no data was rejected because of high background .
the imaging analysis was performed after correcting the data for vignetting and position dependent gain effects ( primarily due to charge transfer inefficiency resulting from radiation damage ) with the current calibration files in 5 energy bands , 0.5 1.5 , 1.5 2.5 , 2.5 4.0 , 4.0 6.0 , and 6.0 10.0 kev .
these were then combined into a single broadband image from 0.5 to 10.0 kev .
background determination for extended sources can be notoriously difficult when the source fills most of the field of view , which is the case for this observation ( see fig 17 ) . by examining the data in several bands and at various smoothing scales we found two regions that are free of cluster emission , one on the extreme nw corner of the i0 chip and another on the far sw corner of the i3 chip .
these regions are used as the background for the imaging analysis and for the spectral analysis of the cluster emission .
the molonglo observatory synthesis telescope ( most ) is an east - west synthesis array comprising two colinear cylindrical paraboloids each 12 m wide by 778 m long , separated by a 15 m gap ( robertson 1991 ) .
it operates at a frequency of 843 mhz with a 3 mhz bandwidth . for the observation of a3128
the synthesized fwhm beamwidth was @xmath8 and the field of view was @xmath9 .
the data reduction used standard in - house software and the rms noise in the cleaned image was 0.85 mjy / beam .
the australia telescope compact array ( atca ) is a 6-element east - west array with a maximum baseline of 6 km .
simultaneous observations were made in two frequency bands , 1384 and 2496 mhz , each of bandwidth 128 mhz .
short observations were made with different antenna configurations in 1999 february , 2000 march and november .
these data sets were co - added after calibration , giving a total integration time of 3.6 hours .
all data reduction was performed using the miriad software package ( sault et al .
the primary flux density calibrator was pks b1934@xmath10638 , and the phase calibrator pks b0334@xmath10546 .
the resultant beamshapes were @xmath11 in pa @xmath12 at 1384 mhz ( natural weighting ) , and @xmath13 in pa @xmath14 at 2496 mhz ( uniform weighting ) .
full details of the radio observations are given in table [ radio_obs ] .
kinematic signatures of groups and other structures within a cluster can be used , in principle , to assess whether clusters are indeed formed through a merger hierarchy . in this section
, we make an empirical examination of the spatial - kinematic distribution of the galaxies in a3128/a3125 , to look for the presence of substructures on various scales . in what follows
we begin with the largest scale features found in our galaxy database , and then work toward smaller scale groups , before turning in 4 to substructure in the icm . a key challenge will be to assess the statistical reality of apparent groupings identified in the galaxy data .
an overall impression of the a3128/a3125 spatial - kinematic structure can be obtained from the dec versus ra plot in fig .
[ cluster ] . the ra and dec axes are plotted in units of arcminutes from the center of a3128 . we have plotted all galaxies from our sample within the redshift range of 9000 @xmath4 cz @xmath4 29000 , but with those galaxies in the narrower velocity range 16000 @xmath4 cz @xmath4 19500 plotted as filled squares .
the locations of the two bright x - ray peaks , discussed in 4 , are plotted as large six - pointed stars .
there is a marked contrast between a3128 , which exhibits a clear central galaxy concentration , and the less massive a3125 , which has a dispersed appearance .
the latter cluster is primarily located between 0and -50 in @xmath15@xmath16 and between -80and -25 in @xmath15@xmath17 in fig .
[ cluster ] . on the basis of n - body simulations , caldwell & rose ( 1997 ) argue that the dispersed nature of a3125 is due to the tidal effect of its recent passage through a3128 ( i.e. , we are now observing a3125 _ after _ a close passage through a3128 ) . in figs .
[ allv ] and [ foba_xyv ] we plot position - velocity diagrams in both ra and dec . in fig .
[ allv ] the large velocity range from 5000 to 70000 is included , while in fig .
[ foba_xyv ] we emphasize the more restricted velocity region between 9000 and 29000 .
the nonuniform distribution in velocity is readily apparent . aside from the obvious a3128/a3125 cluster peak at @xmath018000
, there are concentrations as well at @xmath013000 , @xmath023000 , and @xmath033000 , i.e. , at -5000 , + 5000 , and + 15000 with respect to the the mean cluster velocity . due to our interest in a3128/a3125 , and its relation to the surrounding h - r supercluster , in what follows we restrict our attention to the velocity peaks at @xmath013000 ( or double peak at @xmath012000 and @xmath014000 ) , @xmath018000 , and @xmath023000 , which are better seen in fig .
[ foba_xyv ] . to further emphasize the uneven distribution of redshifts in the direction of the a3128/a3125 system
, we plot the histogram of velocities in fig .
[ velhist ] . the most striking aspect of the histogram is the fact that while the velocity distribution drops off sharply , as expected , at @xmath01500 on both sides of the central cluster redshift at @xmath017775 , the distribution increases again substantially at @xmath04000 on both high and low velocity sides , i.e. , at @xmath014000 and @xmath022000 . on one hand ,
the dip in the histogram , followed by a subsequent rise at both high and low velocity , is basically in accord with the expectation that a cluster will carve out a local velocity void around it , and that the velocity distribution then returns to the unperturbed hubble flow beyond the cluster turnaround radius . on the other hand ,
the turnaround radius for a cluster with velocity dispersion of 1000 is expected to be at 1500 ( kaiser 1987 ) , while the rebound in the a3128/a3125 velocity distribution is at 4000 . as will be discussed further in 5
, we attribute the abnormally large gap in the hubble flow that is inferred from the velocity histogram in fig .
[ velhist ] to the gravitational influence of the h - r supercluster .
the implication is that not only is there an abnormally large void in the hubble flow produced by the h - r supercluster , but that _ groups of galaxies falling into a3128/a3125 after detaching from the hubble flow are accelerated to unusually high infall velocities , due primarily to the h - r supercluster , and secondarily to the acceleration caused by a3128/a3125 itself . _ a second interesting aspect of the position - velocity plots shown in fig .
[ foba_xyv ] is the apparently non - uniform distribution of galaxies in certain areas of the plots , which indicate the presence of spatial - kinematic substructures in a3128/a3125 .
the apparent groupings are most noticeable on the high velocity side of the cluster .
we have identified several groupings , selected either on the basis of their clumping in ra-@xmath18 or dec-@xmath18 space .
these candidate groups are identified in position - velocity diagrams in fig .
[ groups2_xyv ] , where they are plotted with different colors and symbols .
we have identified two types of groups ; ( 1 ) those which are fairly symmetrically distributed in position and velocity ( such as the group , with magenta open circles , centered at dec=-60and cz=18600 ) , and ( 2 ) those which appear to form filaments in position - velocity space ( such as the group highlighted by red squares which stretch from dec=-10and cz=19300 to dec=20and cz=20300 ) .
the locations of these groupings in ra - dec space are plotted in fig .
[ groups2_xy ] , where the same colored symbols of the groups identified in fig .
[ groups2_xyv ] are carried over into the ra - dec plot .
specifically , we have selected two candidate groups ( plotted as magenta and green colored annuli ) and two candidate filaments ( plotted as red and blue colored filled squares ) .
the typical spatial dimension of these systems is @xmath010 - 20 , or @xmath01 - 2 mpc for our assumed value of 50 km sec@xmath6mpc@xmath6 for .
the kinematic length of the filaments , however , at @xmath15cz@xmath01000 , implies a spatial length in the hubble flow of @xmath020 mpc , an implication which is discussed in 6.3 .
are the groups and filaments identified in fig .
[ groups2_xyv ] truly physical groups or just chance associations selected by eye ? unfortunately , testing the statistical reality of such groupings , given that our analysis is clearly ex post facto ( we first select apparent groupings in the plots , and then statistically analyze whether they are indeed real ) , is notoriously suspect . with any large sample of data points ,
it is not difficult to find a suggestive grouping in the data , for which an ex post facto analysis produces a high degree of confidence that the grouping is not random . to make a more realistic assessment of the reality of the group
, one needs to determine what is the likelihood in a data sample ( of a particular overall size ) of finding at least one grouping that is as large as the one selected .
such an analysis in principle could be performed in the following manner .
one could simulate a large number of clusters similar to a3128/a3125 by randomly selecting galaxies from the observed redshift distribution of fig .
[ velhist ] and then distributing them at random in ra and dec over the region covered by our data sample .
one could then assess how frequently groups as large and as concentrated as the ones selected in fig .
[ groups2_xyv ] show up in these random simulations .
it should be clear to the reader , however , that the distribution of galaxies in , especially , the redshift region 18500 @xmath4 22000 is indeed non - random and will indicate clumping at a high statistical level .
the problem is that by setting up a baseline of _ complete _ randomness in ra and dec , we are in effect setting up a statistical strawman to knock down .
we use , instead , the following statistical approach . in principle
, the ra versus @xmath18 plot is statistically independent from the dec versus @xmath18 plot .
therefore , if we select a candidate group or filament that is localized in both position and velocity in the ra versus @xmath18 plot , but that is actually just a false statistical group , the data points should appear randomly spread in dec in the corresponding dec versus @xmath18 plot .
an example of such behavior is shown in figs .
[ control_xyv ] and fig .
[ control_xy ] , where an apparent tight group of points localized between -30and -22 in @xmath16 and between 17500 and 18300 in redshift is spread widely in @xmath17 . in contrast , the groups and filaments selected in either ra or dec in fig .
[ groups2_xyv ] are more localized in the corresponding coordinate than the pseudo group shown in fig .
[ control_xyv ] .
the ra versus dec plot of this pseudo group in fig .
[ control_xy ] confirms the non - association of the galaxies . to provide a quantitative footing to the above approach , in each case that we find a candidate group localized in ra ( or dec ) and @xmath18
, we create a control sample containing all data points within the identifed @xmath18 limits for that group , excluding the group members themselves .
we then apply three statistical measures to assess the reality of the candidate group or filament in three dimensional @xmath16-@xmath17-@xmath18 space .
first , we apply the kolmogorov - smirnoff ( k - s ) two - sample test to the group versus control sample in the following manner .
if the candidate group has been defined in @xmath16 and @xmath18 , we apply to the k - s test to the @xmath17 values , to assess the likelihood that the @xmath17 values of the group and control samples come from the same parent population .
if the group has been selected in @xmath17 , then we apply the k - s test to the @xmath16 values .
second , we make a linear least squares analysis for a correlation between ra and @xmath18 , and also test for a correlation between dec and @xmath18 .
third , we measure the rms dispersion in both the @xmath16 and @xmath17 coordinates for each group , and for each control sample .
the results of the statistical tests are given in table [ groups_table ] . in column
( 1 ) the group identification is given , where g1 and g2 are the two candidate groups ( the green and magenta colored annuli respectively ) ; f1 and f2 are the two candidate filaments ( the red and blue colored squares respectively ) ; c1 is the control group from fig .
[ control_xyv ] . in column
( 2 ) is given the number of group members , while in columns ( 3 ) , ( 4 ) , and ( 5 ) are given the limits in @xmath18 , @xmath16 , and @xmath17 for each group respectively . in column ( 6 ) are listed the results of applying the kolmogorov - smirnov two - sample test to each group versus its respective control sample ; the number given is the likelihood for the hypothesis that the group and its control are drawn from the same parent sample . in columns ( 7 ) and (
8) are listed the results the linear least squares regression analysis for ra versus @xmath18 and dec versus @xmath18 respectively .
the numbers represent the probability that the variables are uncorrelated .
finally , in columns ( 9 ) and ( 10 ) are given the rms deviations in the ra and dec coordinates for the groups .
immediately below each group in table [ groups_table ] is given the data for its specific control sample , for which selection is made in @xmath18 only .
the two groups , g1 and g2 , stand out clearly in having low @xmath19 and @xmath20 compared to their control samples .
in addition , the k - s test indicates a very high confidence that they are different from their control samples . on the other hand ,
there is no indication of correlation in @xmath21 or @xmath22 , which is expected since neither group appears to have a filamentary elongation .
in contrast , the control group c1 , while defined to have a small spread in @xmath16 , shows a large spread in @xmath17 .
surprisingly , the k - s test rejects at a 97.5% likelihood the hypothesis that the c1 group and its own control group are selected from the same sample .
however , the c1 group was defined to have a low mean ra , which places it away from the main concentration of a3128 .
therefore , the distribution of dec values in this pseudo group do not reflect that of the main a3128 cluster .
the control group for c1 is defined in @xmath18 only , hence the distribution in dec shows the expected concentration around a3128 . as a consequence
, the k - s test indicates that the samples are different .
thus although the dec values in the c1 `` group '' are clearly spread quite randomly ( see figs .
[ control_xyv ] and [ control_xy ] ) , the k - s test does not actually test that fact .
the case of the two filaments , f1 and f2 , is less clear .
the principal evidence that f1 and f2 are real structures lies in the correlation statistics between @xmath16 and @xmath18 and between @xmath17 and @xmath18 . for f1 , the probability of a correlation is high in _ both _ coordinates , while for f2 the probability is only significant in @xmath17 versus @xmath18 . a high correlation is also found in @xmath16 versus @xmath18 for the two control samples .
this is due to the fact that the control samples for both f1 and f2 contain parts of both filaments . since f1 and f2 are widely separated in @xmath17 and also have a mean offset in @xmath18 , a high correlation coefficient is found .
this result reinforces the above example regarding the k - s result for the c1 control sample , namely , that blindly applied statistical tests can be misleading .
overall , we consider the case for the reality of f1 to be strong , given the high correlation probability in both ra and dec .
the statistical case for f2 , on the other hand , rests entirely on the correlation found in @xmath17 versus @xmath18 , and thus needs to be taken with caution .
we also note that g1 joins onto f1 , as does g2 onto f2 .
thus we have perhaps artificially separated a group component from a filament component .
in addition to the g1 , g2 , f1 , and f2 groups shown in figs . [ groups2_xyv ] and [ groups2_xy ] , we have selected five additional small groups , each with less than 10 members , which are plotted in figs . [ groups3_xyv ] and [ groups3_xy ] . for these latter groups ,
the only statistical test that we have applied is to measure the dispersions in ra and dec , as well as for their corresponding control samples .
the results of that analysis is summarized in table [ groups3_table ] . in all cases , the small rms dispersion in one coordinate
is matched in the other coordinate , particularly when a single discordant galaxy is removed .
in contrast , the rms dispersion in the corresponding control sample is always substantially larger , except in the case of group g6 , for which the control sample of only 4 galaxies appears to constitute a small group of its own .
a final comment about the statistical reality of groups concerns the linear velocity gradient in the a3125 cluster reported in caldwell & rose ( 1997 ) , on the basis of 19 galaxies , which is typical of the numbers in g1 , g2 , f1 , and f2 . that linear velocity gradient is no longer evident in a3125 , now that a much larger sample of galaxies has been observed .
in particular , the presence of g2 and f2 in the sw , which due to the smaller area sampled by caldwell & rose ( 1997 ) was missed by them , has swamped the original gradient .
it appears likely that caldwell & rose ( 1997 ) detected a gradient produced by sampling primarily from two galaxy clumps with different mean redshift . in short , when small numbers of galaxies are involved , it is difficult to distinguish between a real filamentary gradient and an apparent gradient produced by the offset properties of two groups .
turning now to the concentrations of galaxies in the foreground and background of a3128/a3125 , we have subdivided the foreground and background into various velocity slices and plotted them in position - velocity space in fig .
[ foba_xyv ] , with different colors referring to the different velocity slices , and annuli and filled squares representing foreground and background galaxies respectively .
the positions of the color - coded galaxies are repeated in an @xmath23 plot in fig .
[ foba_xy ] . an intriguing aspect of figs .
[ foba_xyv ] and [ foba_xy ] is the nonuniform distribution of foreground versus background galaxies .
there is an overall tendency for the foreground galaxies ( colored unfilled circles ) to populate the ne region , i.e. , to concentrate around a3128 , while the background galaxies ( colored squares ) lie primarily in the southern portion of the plot .
caution must again be exercised in making a statistical argument about this offset between foreground and background , since the background galaxies , in particular , appear to be somewhat concentrated into groups .
the existence of background group g7 in the se has already been pointed out in figs .
[ groups3_xyv ] and [ groups3_xy ] . as a result
, the foreground and background galaxy distributions might well result from just a few independent groups .
thus statistical tests for differences between the distribution of foreground versus background galaxies , based on assuming all galaxies to be independent points , will not accurately reflect that situation . to further illustrate the differences in spatial distribution between foreground and background
, we plot the galaxy redshifts versus their radial distance from the center of a3128 in fig .
[ radius ] .
the large ( @xmath04000 ) gap in @xmath18 between cluster and both foreground and background is again clearly evident .
in addition , while the foreground galaxies tend to have radial distances in the range 0to 50 , the background galaxies largely avoid the central 20 . as previously mentioned , due to problems with the continuum levels near [ oii]@xmath73727 for the fainter galaxies , it was not possible to obtain reliable emission line equivalent widths .
however , we are able to search for [ oii]@xmath73727 emission in the galaxies to typically 0.5 levels in equivalent width . as a result
we have compiled statistics of the percentage of galaxies with observed emission , as a function of their environment .
here we have restricted the discussion to only those galaxies with aat/2df spectra , since those spectra provide the most homogeneous set to work from .
we find a surprisingly high percentage of galaxies in the main a3128 cluster have detectable [ oii]@xmath73727 emission .
specifically , 41 of 85 galaxies , or 48% , with radial velocity between 16000 and 19200 , and with declination north of -53:01:30 ( the center of a3128 is at @xmath17 = -52:31:30 , j2000 ) have emission .
this is an unusually high level of emission , given that most surveys of emission in rich clusters report emission detection rates below 15% ( biviano 1997 ) .
in contrast , biviano ( 1997 ) report emission line detections for only 30 out of 152 galaxies in their data on a3128 , i.e. , an @xmath020% detection rate .
thus the high fraction of galaxies found with emission lines is partly the result of the high s / n ratio achieved for the 2df spectra , but also partly due to the fact that a3128 has an unusually high level of emission , since the 20% detection rate reported by biviano ( 1997 ) is at the high end for the enacs cluster sample . while the main cluster emission - line galaxy fraction is high , the detection rate in our 2df data of galaxies in foreground and background flows is substantially higher .
a total of 38 out of 54 galaxies , i.e. , 70% , of the galaxies in the background within the velocity range 22100 to 28500 have detected [ oii]@xmath73727 emission , while 15 out of 16 , or 94% , of the foreground galaxies in the velocity range 10000 to 15000 are also detected .
the statistics of the group and filament galaxies is that 32 out of 48 of them , i.e .
, 67% , have detected emission .
thus , while the emission line statistics of the main a3128 cluster are remarkably high , indicating perhaps an overall dynamical youth of the cluster , the emission levels are still lower than in the immediate field and in infalling groups . finally , of the 19 galaxies in the background redshift peak at cz@xmath033000 , 16 of them (
i.e. , 84% ) have detected [ oii]@xmath73727 emission , hence indicative of a field population .
note that the detection of [ oii]@xmath73727 emission alone does not allow us to distinguish between emission due to star formation and that caused by an active galactic nucleus .
to conclude at this point , data on the distribution of galaxies in the field of a3128/a3125 indicates that spatial / kinematic structures are present on a variety of scales .
the most evident structure is the @xmath04000 underpopulated velocity region on both sides of the cluster velocity .
much smaller structures are evident as well , ranging from groups of 4 galaxies to groups with a couple of dozen members , and even the foreground and background galaxies at 4000 on either side of the a3128 systemic velocity are nonuniformly distributed .
the majority of galaxies in the underpopulated velocity regions appear to be members of groups and/or filaments .
however , the statistical reality of these groups ( and particularly of the filaments ) is difficult to establish .
most of the groups identified have high velocities with respect to the a3128/a3125 double cluster .
such high velocities of these infalling and merging substructures should in principle lead to unusual transient physical conditions in the hot intracluster medium , a subject which we turn to in the next section , where we consider the chandra x - ray images of a3128 .
aside from spatial - kinematic substructures evident in the galaxy distribution , substructure in the hot icm also provides clues to past and ongoing cluster merger events .
the icm has the particular advantage of not suffering from the discreteness problem that afflicts the statistical selection of small galaxy groups .
consequently , we now turn to the _ chandra _ acis - i observations of the central region of a3128 .
the x - ray morphology of the central 16x 16of a3128 is illustrated in figs .
[ ds9_3 ] and [ ds9_2 ] , where the 20 ks chandra image is displayed at two different contrast levels , for photons in the energy range 0.5 kev to 10 kev . in all cases the original image has been 4 x 4 binned into 2x 2pixels , and then further smoothed with a gaussian of @xmath24=4 pixels . the most evident feature of the cluster x - ray morphology is the presence of two bright x - ray peaks , aligned in the ne - sw axis , and separated by 12 , or 1.2 h@xmath2mpc .
these two peaks have been marked previously on figures such as fig .
[ cluster ] .
we hereafter refer to these two components as the ne and sw peaks .
also evident is a lower surface brightness component , located between the two bright peaks .
this third component to the x - ray flux has a center that , to within the uncertainties , is indistinguishable from the center of the galaxy distribution in a3128 . to further assess the morphology of the x - ray emission , we present in fig .
[ adapt_1 ] an adaptively smoothed version of the chandra image .
specifically , a maximum smoothing of @xmath24=5 pixels was applied .
furthermore , to illustrate the connection between the x - ray emission and the galaxy distribution , we have plotted the chandra x - ray contours on top of a grayscale representation of the optical b@xmath5 image in fig .
[ xrscos ] .
it is evident that while the sw x - ray peak is coincident with a bright galaxy , and with an apparent compact galaxy group , there is no such general coincidence between the ne x - ray peak and the galaxy distribution , a point that we return to in 6 .
a second noteworthy feature of the x - ray morphology is the fact that the two bright peaks have small core radii and are offset from the lower surface brightness emission on which they are superimposed , as can be seen qualitatively in , e.g. , fig .
[ adapt_1 ] . to quantify the small core radii we have plotted in fig .
[ profs ] the _ azimuthally averaged _ radial surface brightness profiles of the ne and sw x - ray components , plotted both in terms of linear radial distance and logarithmic radial distance . also plotted are the best @xmath25 model fits to the ne and sw components , following the prescription of cavaliere & fusco - fumiano ( 1978 ) , i.e. , we fit the surface brightness distribution according to @xmath26^{-3\beta+1/2},$ ] where @xmath27 is the core radius of the fit .
we obtain a core radius of 26 h@xmath2kpc and 40 h@xmath2kpc for the ne and sw components respectively .
as will be further discussed in 5 , these values are at the low end of the large range in core radii observed in clusters of galaxies ( e.g. , jones & forman 1984 ; peres 1998 ; mohr , mathiesen , & evrard 1999 ) , where a typical cluster has a core radius of @xmath0250 kpc .
in addition , the @xmath25 values of @xmath00.3 derived from the fits are low compared to typical values for clusters of @xmath00.7 .
to better illustrate the fact that the bright ne and sw cores are offset from their underlying lower surface brightness emission , we have fit elliptical isophotes to the ne and sw components at various surface brightness levels , using the ` ellipse ' routine in the stsdas ` isophote ' package .
the fits return the x - y positions of the center of the fitted ellipse at each contour level , and the results have been plotted in fig .
[ centers ] .
the x ( ra ) and y ( dec ) centers change by 55 pixels ( 110 ) and 20 pixels ( 40 ) respectively in the ne peak as one moves from the central peak out to a maximum of 200 in the semi - major axis of the azimuthally averaged radial profile . in the sw peak the x and y offsets are 25 pixels and 12 pixels respectively .
thus the effect is considerably more pronounced in the ne component , but clearly is present in the sw component as well . moreover , while the shift in isophote center is fairly linear with semi - major axis , in the case of the ne component , the shift is very small within the central 20 pixels ( 40 ) , then ramps up at larger radii .
note that the displacement of the core emission in both peaks relative to the lower surface brightness emission underlying them is in the sense that the emission falls off sharply on the side pointing away from the cluster and trails off more slowly in the cluster center direction .
a final morphological feature of interest is the resolved inner structure of the ne component , seen in figs .
[ ds9_3 ] and [ adapt_1 ] .
the most striking feature of the ne emission peak is its strong elongation _ along _ the axis connecting the ne and sw components .
this elongation factor is substantial , as can be seen from the ellipticity and position angle profiles plotted in fig .
[ ellpa ] .
while the mean ellipticity for the fainter contours is @xmath00.4 , it rises to 0.65 in the inner 5 pixels ( 17 h@xmath2 kpc ) in radius .
furthermore , there is a substantial twist in the position angle of the fitted ellipse , from 60@xmath28 e of n in the inner 10to 70@xmath28 e of n at a semi - major axis of 50 .
in short , transient morphological features are present in the x - ray emission from the icm of a3128 on a variety of scales , from less than 20 h@xmath2 kpc in the core of the ne component to the @xmath01 mpc projected separation between the ne and sw components .
another key diagnostic of the physical conditions in the icm of a3128 is the temperature structure of the x - ray emitting gas .
we have derived temperatures from the observed x - ray emission by using xspec ( arnaud 1996 ) to fit a mekal ( mewe 1985 , 1986 ; liedahl , osterhedl , & goldstein 1995 ) thin plasma model with foreground absorption to the spectral data .
the abundances are given relative to the solar values ( anders & grevesse 1989 ) .
in these fits the hi absorption , temperature , and abundance are allowed to vary , and we restrict our analysis to the 0.5 10 kev band .
the ne and sw components have global temperatures of 3.9 kev and 3.6 kev respectively , while the third , lower surface brightness , component has a temperature of @xmath04 kev . moreover , while the sw component appears to be isothermal , the data on the ne component suggests that the temperature is higher in the inner 45 in radius .
the data on temperature and metal abundance , along with their 90% confidence limits , are summarized in table [ temps_table ] .
while the uncertainties are substantial , the data are suggestive of a temperature increase in the core of the ne component , and are in any case inconsistent with a substantial temperature drop in either the ne or sw component .
thus there is no indication of a cooling flow in either peak . from the @xmath25 model fit to the radial surface brightness in the ne and sw components , and assuming the temperatures from the `` global '' fits to each component , we have calculated an estimated `` virial '' mass enclosed within a radius of 0.3 h@xmath2 mpc for both components , by assuming hydrostatic equilibrium . for the ne and sw components
we then derive a virial mass of 1.6 x 10@xmath29 m@xmath30 and 1.5 x 10@xmath29 m@xmath30 , respectively .
on the other hand , the gas mass in the ne and sw components ( inside of 3 or 0.3 h@xmath2 mpc in radius ) derived from the central electron densities of 1.1 x 10@xmath31 @xmath32 and 2.4 x 10@xmath31 @xmath32 is 6.5 x 10@xmath33 m@xmath30 and 7.2 x 10@xmath33 m@xmath30 , respectively , for the ne and sw components ( the x - ray luminosities of the two components are 5.4 x 10@xmath34 erg / sec and 3.4 x 10@xmath34 erg / sec in the 0.5 10 kev band ) .
thus the implied gas - mass to virial - mass ratios are @xmath04 - 5% in the two components , which is very low compared to more typical values of @xmath020 - 30% usually found in rich clusters ( e.g. , mohr , mathiesen & evrard 1999 )
. seen another way , for the observed x - ray emitting gas masses , the virialized velocity dispersion in the galaxies for the ne and sw components should be 437 and 425 respectively . while the observed velocity dispersion of the galaxies can not be readily separated into ne and sw components , and while in fact the presence of so much substructure in a3128 makes it difficult to define a global cluster velocity dispersion , it is clear in any case that the velocity dispersion of the galaxies in the center of a3128 is far in excess of 430 .
thus the discrepant values extracted from a dynamical analysis indicates that the x - ray emitting gas is _ not _ virialized and therefore must be well removed from hydrostatic equilibrium .
the third , diffuse emission component , seen between the two x - ray peaks ( fig .
[ ds9_2 ] and fig .
[ adapt_1 ] ) is noteworthy in that its extent is approximately that of a normal cluster ( @xmath03 mpc ) , it is hot ( kt@xmath04 kev ) , and is also enriched with an abundance of @xmath0 0.6 solar .
the fact that the diffuse gas is enriched implies that it is or was bound to a system with stars which can supply the metals . since there are two peaks with similar properties , determining which one might be associated with the diffuse emission component must be based on indirect evidence . from the global abundances in table [ temps_table ]
we see that the ne component has a very low global abundance , @xmath00.13 solar , much lower than the best fit abundance of the diffuse component , which on the other hand is consistent with the abundance of the sw clump .
so , while we can not rule out that the diffuse component is associated with the ne component , it seems more likely to be associated with the enriched sw component .
thus the third , diffuse , component appears unlikely to have a common origin with the ne component .
finally , we have obtained a direct estimate of the redshift of the x - ray gas in the ne , sw , and diffuse components using the x - ray emission lines present in the acis - i spectra . for the sw component , our best fit model , in which we allow the redshift to be a free parameter ,
produces a best fit redshift of z=0.065 , with 90% confidence limits between 0.050 and 0.078 . for the diffuse emission component , the best fit
redshift is z=0.059 , with a 90% confidence range of 0.047 to 0.072 . due to the lower counts and lower metal abundance in the ne component , the redshift of that gas is less well constrained .
the best fit redshift is 0.10 , with a 90% confidence range of 0.037 to 0.23 .
thus we can conclude that the gas in the sw and diffuse components are unambiguously associated with a3128 , while the redshift found for the ne component is consistent with a3128 , but the association is not conclusive .
further insight into the events transpiring in a3128/a3125 can be gained from examination of radio images of the double cluster , since radio sources can be reenergized by icm shocks produced in mergers ( ensslin 1998 ; roettiger , burns , & stone 1999 ; govoni 2001 ) .
furthermore , the morphology of tailed sources can be useful in assessing the relative direction of motion of the source with respect to any gas dynamical movement of the local icm .
we begin by displaying in fig .
[ xrmost ] the most 843 mhz radio contours superposed on the chandra acis - i x - ray image .
four of the radio sources have x - ray counterparts , three of which appear to be cluster members , as revealed by the overlaid radio - optical plot in fig.[mostscos ] .
we will focus here on the two radio sources aligned with the ne and sw x - ray peaks .
the weak unresolved source coincident with the sw x - ray core is identified with enacs#75 , one of the brightest galaxies in the cluster . on the other hand ,
the radio source at the ne peak is both stronger and more remarkable .
it appears to consist of a bright point source and a much fainter arc - like feature , displaced to the ne of the bright source . at the higher resolution of the atca 20 cm and 13 cm images
, however , the bright radio source resolves into an elongated north - south double , as shown in fig .
[ atca_20_13 ] , where the atca 20-cm and 13-cm contours are superposed on a grayscale rendition of the ctio 0.9-m r bandpass image .
this figure shows that the radio source is identified with a faint galaxy that is undoubtedly background to a3128 .
the spatial alignment of optical galaxy , most and atca radio contours , and x - ray contours is illustrated in fig.[overlay_ne ] ctio 0.9-m b and r images of the region in fig .
[ negal ] show that the faint galaxy has a concentrated red nucleus . in the deeper r - band image a somewhat irregular low surface brightness fuzz
surrounds the bright nucleus , while in the b image , a long curved structure is seen primarily to the southwest .
a hint of this curved structure is seen in the r image as well , but it clearly has a very blue color relative to the nucleus of the galaxy . in the unlikely circumstance that the galaxy is a member of a3128 , it would be a low - luminosity dwarf , and hence extremely unlikely to harbor a compact luminous double radio source .
thus we infer that the bright double is in fact a superposed background source that has no physical relation to a3128 , consistent with the galaxy s r magnitude ( see section 6.4.6 ) , the integrated radio power at 1.4 ghz is @xmath35 w hz@xmath6 , making it a powerful source , near the boundary between fanaroff - riley classes fri and frii ] . on the other hand , the faint extended radio arc seen in the most image ( but not in the higher resolution atca 20 cm image ) , is more intriguing .
it is situated slightly to the ne of the core emission in the ne x - ray component , about where one might expect a bow shock to be present if the ne component were plowing supersonically through a3128 .
the fact that the radio arc is displaced from the bright background source suggests that it truly is a structure associated with a3128 , but deeper radio imaging is needed to better assess that conjecture .
the single most compelling feature of the combined 2df spectroscopy and chandra x - ray imaging is the wide variety of structures observed in both the spatial / kinematic distribution of the galaxies and the hot icm . on the largest scales
there are peaks in the velocity distribution , separated by @xmath04000 , i.e. , over inferred distances of @xmath080 h@xmath2 mpc , while on the smallest scales the x - ray image of a3128 shows transient structures @xmath3620
kpc in size . in this section
we attempt to draw a coherent picture encompassing the structures found on all scales , and to establish a link between galaxy substructure and the disturbed state of the icm . in what follows ,
we begin with the largest structures , to establish the overall potential field in which the smaller substructures are embedded , and then work toward the smaller scales . as has been pointed out before , the velocity histogram in fig .
[ velhist ] indicates that on either side of the huge velocity peak at @xmath017750 , representing the velocity center of a3128/a3125 , there is a large `` gap '' of @xmath04000 , in which the galaxy density is low , picking up again at @xmath014000 on the low velocity side , and at @xmath022000 on the high velocity side .
it is natural to assume that the increase in galaxy density at @xmath014000 and @xmath022000 represents the resumption of the hubble flow on either side of the double cluster .
while this assumption might be compatible with the galaxy distribution in redshift on the low velocity side , the distribution at high redshift is puzzling .
there is a strong peak in redshift at @xmath023000 , but then the histogram falls off precipitously to higher redshift , and between @xmath025000 and @xmath032000 there are very few galaxies found .
the histogram then rises to another peak between @xmath032000 and @xmath035000 , before dropping off once again to higher redshift .
we note that the spatial distribution of galaxies in the redshift peak at @xmath033000 show no indication of being concentrated into a background cluster .
moreover , as mentioned before the fraction of emission line galaxies in the @xmath033000 peak is 84% , indicative of a field , rather than a cluster , population .
the behavior of the redshift histogram over the full range out to cz = 65000is shown in fig .
[ hist200_all ] .
the velocity peaks at @xmath014000 and @xmath023000 are quite narrow , certainly less than 5000 in full width , indicating that they might represent galaxy clusters at these redshifts .
however , the galaxies in the peaks are distributed over the full 2 field of the 2df observations , and show no concentration into clusters .
rather , as was discussed in 3 , there is some evidence to indicate an overall ne - sw asymmetry between the distribution of low - velocity peak versus high - velocity peak galaxies , which may result from the clumping of these galaxies into groups .
thus , while it is perhaps problematic to associate the peaks at @xmath014000 and @xmath023000 with the resumption of normal hubble flow outside of a3128/a3125 , _ we conclude here that the low and high velocity peaks on either side of the double cluster are the working `` surfaces '' from which new galaxies are being recruited into the a3128/a3125 system .
_ the large , @xmath04000 , velocity gap between these surfaces and the central cluster velocity is far in excess of that expected from a cluster of normal mass , and can only be attributed to a much greater massive concentration ( kaiser 1987 ) .
as has been pointed out previously , a3128/a3125 is embedded in the horolgium - reticulum supercluster , which extends over an area at least 7 in extent ( lucey 1983 ) .
the h - r supercluster and shapley supercluster are the two largest known mass concentrations in the local universe , i.e. , out to 300 h@xmath37mpc ( zucca 1993 ; einasto 1994 ; hudson 1999 ) . in fact , zucca ( 1993 ) and einasto ( 1994 ) consider the h - r supercluster to encompass 18 and 32 clusters respectively .
thus we associate the largest structures in our data , i.e. , the @xmath04000 gaps to the low and high velocity of a3128/a3125 , with the huge gravitational potential well of the h - r supercluster .
the most important implication of this conclusion is that groups of galaxies which are detached fom the working `` surfaces '' are accelerated to unusually high infall velocity on their way toward a3128/a3125 , due to the acceleration produced by the deep potential well of the h - r supercluster .
specifically , the infall velocities will be supersonic with respect to the local sound speed in the a3128/a3125 complex , which , at an electron temperature of 4 kev , corresponds to @xmath01000 . to put the redshift histogram of the a3128/a3125 region into context
, there are strong similarities with the observed distribution in redshift of galaxies in the shapley supercluster found by , e.g. , quintana ( 2000 ) and drinkwater ( 1999 ) . in the case of the shapley supercluster , the area surveyed is very extensive , so that the complexity of the large gaps in redshift , similar in size to the 4000 gaps found by us in the h - r supercluster , is readily seen in , e.g. , figs . 6 and 7 of quintana ( 2000 ) .
in contrast , where a large body of redshift data has been collected in the vicinity of a cluster that is _ not _ associated with a mass concentration on the scale of the h - r supercluster , such as the coma cluster ( geller , diaferio , & kurtz 1999 ) , no large gaps are evident in the distribution of velocities .
thus the association of the velocity gaps in a3128/a3125 with the potential well of the h - r supercluster is further supported by comparison to other cases .
the next clearly defined scale below that of the h - r supercluster is the double nature of the a3128/a3125 cluster system .
much of this has been discussed in caldwell & rose ( 1997 ) , hence we repeat only a few key points .
the projected separation between the center of a3128 and the very poorly - defined center of a3125 is on the order of 1 , which corresponds to @xmath06 mpc .
surprisingly , a3125 and a3128 have nearly the same mean cluster velocity , i.e. , there is little relative motion between them along the line of sight .
thus one can not reconcile the large distance between the two clusters with the small velocity separation ( without appealing to a highly preferred viewing angle ) , _ unless a3125 is currently near turnaround after its passage by a3128 .
_ it is difficult to define reliable global velocity dispersions for a3125 and for a3128 , given the large number of groups and filaments present whose inclusion or exclusion greatly affects the derived @xmath24 ( and given the fact that a3125 is not a clearly defined cluster ) .
however , it is certainly clear that the velocity dispersion of a3125 is lower than that of a3128 , which is in accord with the fact that a3128 contains @xmath04 times more members than a3125 .
in addition , while a3128 has a well - defined central concentration in the galaxy distribution , a3125 presents a dispersed appearance .
based on n - body simulations , caldwell & rose ( 1997 ) attribute the scattered appearance of a3125 to a tidal passage of the latter past a3128 .
in addition , while the 20 ks chandra exposure of a3128 reveals extensive emission from a hot icm , the 10 ks exposure of a3125 reveals no extended emission .
what appears to be extended emission in an archival rosat 4 ks image is actually the merging of badly degraded point sources , due to the large off - axis location of a3125 in the fov of the rosat data . if there were any structures in a3125 as bright as the ne and sw components seen in a3128 , they would easily be detected in the 10 ks chandra image of a3125
consequently , the available evidence is consistent with a3125 being a relatively small cluster that has experienced a damaging encounter with a3128 .
we turn now to the intermediate - sized groups ( 10 - 30 observed members ) and filaments .
naturally , such groups are very difficult to locate in the crowded regions of a3128 and a3125 , if they do exist there , but are readily seen in the low density 4000 ` gaps ' between the clusters and the higher density regions outside the h - r supercluster .
while identification of galaxies with groups and filaments can be problematic on statistical grounds , especially in regard to filaments , nearly all galaxies present in the high - velocity ` gap ' region can be identified with intermediate - sized and small groups or filaments .
the two filaments are particularly interesting .
while their spatial extent of @xmath020corresponds to a length of 2 h@xmath2 mpc , their velocity distension of @xmath01000 corresponds to 20 h@xmath2 mpc in terms of hubble flow .
since it is highly unlikely that the filaments would be so nicely oriented along our line of sight with a3128/a3125 , we conclude that _ their distension in velocity is unlikely to be a reflection of hubble flow , but rather is due to a tidally - induced velocity gradient in an infalling group . _
furthermore , it was demonstrated in caldwell & rose ( 1997 ) that tidal distension of subclusters occurs only after the subcluster has passed through the main cluster , in either a head - on , or slightly off - center , encounter .
consequently , we expect that the tidally distended filements have already made a passage through the h - r supercluster and a3128/a3125 system and are now emerging from that passage . in contrast , the more circularly symmetric ( in both position and velocity ) groups are probably still in the infall phase of their passage through the cluster / supercluster . consequently , the filaments are more likely to be associated with visible signs of transient merging activity in the a3128/a3125 system , especially in regard to disturbances in the icm . to further assess this hypothesis regarding groups versus filaments , it would be of interest to compare the physical properties of the galaxies in groups versus filaments for indications of different recent evolutionary histories . over length scales
@xmath36100 kpc , and group membership less than a few galaxies , galaxies are of limited use as tracers of substructure in a3128/a3125 .
at this point , x - ray emission from the icm becomes our primary means to delineate small - scale structure .
as has been discussed in 3 , ample evidence is seen in the chandra image of a3128 for structures in the icm on scales below 1 h@xmath2 mpc .
we begin by briefly recapitulating those key observations related to the structure of the icm , namely : \(1 ) the x - ray emission shows two distinct regions with high surface brightness cores , and an additional lower surface brightness third component between the two main components .
\(2 ) the two bright components have small core radii .
\(3 ) the bright peaks , especially the ne one , are displaced with respect to their underlying lower brightness contours .
\(4 ) in addition to the overall displacement of brighter versus fainter contours , the ne x - ray component shows a strongly asymmetric morphology .
\(5 ) the heavy element abundance of the gas in the ne x - ray component ( @xmath00.13 solar ) is significantly lower than for the gas in the sw component ( @xmath00.6 ) .
\(6 ) a bright galaxy ( and member of a3128 ) is coincident with the sw x - ray peak ; no bright galaxy coincides with the ne peak .
\(7 ) radio sources are coincident with both x - ray peaks .
we now attempt to produce a coherent picture of the x - ray structures , and their relation to the optical galaxy and radio data .
we will first argue that the x - ray structures are transient , and likely associated with merger activity .
then we propose that the ne x - ray peak is the leading edge of highly supersonic gas , once associated with the f1 and g1 galaxy filament / group , which has now passed through the center of a3128 .
finally , we argue that the sw x - ray peak is most likely a gaseous component at the trailing end of the f1/g1 infall pattern which is still at least partially bound to the underlying potential well associated with a compact group of galaxies .
the most conspicuous aspect of the icm is the division of the x - ray emission into two bright peaks , separated by 1.2 h@xmath2 mpc along roughly the same ne - sw axis that joins a3128 with a3125 . a third fainter and less centrally concentrated component
is located between the ne and sw bright peaks , and very close to the center of the galaxy distribution .
the existence of multiple x - ray components on relatively small spatial scales implies transient disturbances in the icm .
in addition , the two bright components have core radii of @xmath030 h@xmath2 kpc .
while most clusters typically have core radii 10 times greater than this , there are two kinds of clusters with small core radii .
first are the cooling flow clusters , which are generally characterized by a cool peak of bright emission atop a relaxed icm with large core radius ( mohr , mathiesen , & evrard 1999 ) .
an example of such a cooling flow cluster with highly structured x - ray emission and a small core radius is a1795 ( fabian 2000 ) .
since the central temperature does not drop in either the ne or sw peak , nor is the underlying cluster relaxed ( given the double / triple nature of the x - ray emission and the extensive evidence for ongoing merging in a3128/a3125 ) , the identification of the bright x - ray components in a3128 with cooling flows is clearly inappropriate .
the second type of cluster with small core radii is represented by those , such as a548 , which are undergoing merging events ( davis 1995 ) .
the similarity between a3128 and a548 is striking , in that a548 also has multiple x - ray emission peaks , with small core radii . in this case the small core radii and low value of @xmath25 is considered a result of the truncation of the cluster profile ( mulchaey 2000 ; navarro , frenk , & white 1996 ; bartelmann & steinmetz 1996 ) .
finally , given that the sound travel time across the asymmetric 20 kpc ne region for a 4 kev gas is only @xmath02 x 10@xmath38 years , then unless the gas in the ne emission peak is pressure confined , we are witnessing a highly transient structure in the icm .
we argue in the appendix a.2 that the group gas has been stripped from the group and thus is no longer gravitationally bound , but is presently ram pressure confined .
furthermore , the ne component is elongated by a @xmath02:1 axis ratio along the same ne - sw axis that separates the two bright components .
thus we observe a preferred alignment along the ne - sw from scales below 20 h@xmath2 kpc to the 6 mpc distance separating a3128 from a3125 ; alignments over a large variety of scales is seen in other clusters as well ( west & blakeslee 2000 ) .
these large - scale alignments are believed to represent the preferred axis of infall and merging activity in the cluster .
having established the transient nature of the x - ray structures and its probable relation to merging events , we now attempt to link the major x - ray structures with specific spatial / kinematic substructures in the galaxies .
first , it is natural to associate the double x - ray structure with the a3128/a3125 double system .
however , the projected separation between a3128 and a3125 , at @xmath06 h@xmath2 mpc , is approximately 5 times larger than the observed separation between the two high surface brightness x - ray peaks .
furthermore , the small velocity difference between a3125 and a3128 implies a multi - gyr timescale for that passage , whereas the timescales for the two x - ray peaks ( e.g. , the cooling timescale , which is only @xmath01 gyr ) appear incompatible .
hence we look to smaller substructures in the galaxies for an explanation of the x - ray peaks . in examining the position - position and position-@xmath18 plots in figs .
[ groups2_xy ] and [ groups2_xyv ] , the two substructures that are at least projected close to the positions of the bright x - ray peaks are the group denoted by green annuli ( g1 in table [ groups_table ] ) and the filament denoted by the red squares ( f1 in table [ groups_table ] ) . if f1 has been correctly identified as a post - passage ( through a3128 ) tidally distended group , then it represents the most natural candidate for producing the transient structure seen in the chandra image of the icm . referring to fig .
[ groups2_xy ] it is evident that the projected axes of f1 and the ne - sw x - ray sources are approximately parallel .
moreover , since the distinction between f1 and g1 has been drawn somewhat artificially , we henceforth refer to them collectively as the f1/g1 group
. a closer look at fig .
[ groups2_xy ] reveals that the f1/g1 group extends @xmath030to the ne of the ne x - ray peak , and that this represents the high velocity , hence likely leading , edge of the group .
given our interpretation of the filaments as post - passage tidally distended groups , we argue that the ne end of f1 has passed through the center of a3128 , and that much of the group gas is continuing through , with some of it trailing behind , as indicated by the elongated isophotes . on the other hand , the bulk of g1 and the gas associated with the sw x - ray peak are located on the other side of the cluster center , and thus are considered to be still infalling .
support for the above scenario comes from the morphology of the ne peak in particular , its displacement by more than 2 , or 200 h@xmath2 kpc , to the ne of the lower surface brightness isophotes of that x - ray component .
given that the large acceleration imparted by the h - r supercluster should result in supersonic impacts of infalling groups into the main cluster , we can expect to observe signs of shock structures in the icm .
the large offset nature of the ne peak may well represent such activity , in which case the outward direction of its motion from the cluster center has been established , in agreement with our hypothesis about the outward flow of the ne end of f1 .
further , from the projected separation between the ne end of f1 and the outer edge of the ne x - ray core we can derive an estimate of the relative velocities of the f1 filament and the x - ray source .
this calculation is discussed in appendix a.1 .
here we simply report the main result , which is that the gas in the ne x - ray peak is inferred to be travelling at about half the velocity of the leading edge of f1 , and its velocity with respect to the a3128 cluster is @xmath03500 .
this speed corresponds to a mach number , m=6 , for a typical cluster gas temperature of 3 kev .
thus , as discussed in appendix a.2 , the gas in the ne peak is inferred to be highly supersonic , shocked gas which has been stripped from a group of galaxies ( now seen as f1 ) during a passage through a3128 at hypersonic velocity ( in excess of 4000 ) . as is discussed in appendix a.3 , at the high mach number ( m=6 ) inferred , the x - ray structure should be primarily determined by the effects of ram pressure ( which is of course axisymmetric ) as opposed to cluster thermal pressure ( which is spherically symmetric )
this is clearly consistent with the observed axial structure in the x - ray source .
one further piece of observational support for this scenario comes from the aforementioned radio arc , which lies slightly to the ne of the peak emission in the ne core , i.e. , where a bowshock is expected if the gas is moving hypersonically .
as mentioned previously , however , deeper radio imaging is required to verify whether or not the radio arc is indeed related to the cluster , rather than to a background source .
the most intriguing aspect of the x - ray structure is the existence of two bright peaks .
the question that naturally arises is whether both peaks can be attributed to a single merging event or whether two separate events must be invoked .
the answer to this question is of great significance to interpreting the events taking place in a3128 . on the one hand ,
our analysis has uncovered evidence for a variety of substructures in a3128 .
on the other hand , only the f1 and g1 groups appear to be in a region coincident with the main x - ray activity , and it is not even clear whether f1 and g1 are two independent systems .
earlier we interpeted the displacement of the core emission in the ne x - ray component from the lower surface brightness emission in terms of a ram pressure that delineates the direction of motion of the ne peak . in principle , the same argument can be applied to the sw core , to infer that its direction of motion , relative to the main cluster gas , is to the sw .
such an inference would present serious problems for the general scenario presented above , since we have argued that the _ entire _
f1/g1 is moving to the ne relative to a3128 , with the sw part of it at lower speed .
however , we propose instead that the displacement of the sw x - ray core is due to the effect of the local gravitational potential well of a compact group centered on the bright galaxy ( enacs#75 ) that is coincident with the sw x - ray core and the weak most radio source . a ctio 0.9-m r band image of the region immediately surrounding enacs#75
is shown in fig .
[ cgsw ] . there
it can be seen that enacs#75 itself , which appears as a single bright object ( at the contrast shown ) on the supercosmos grayscale image in fig .
[ xrscos ] , coincident with the sw x - ray core , is resolved into three objects , with several more galaxies to the east .
the @xmath39 sw x - ray core component coincides with the main enacs#75 object . while we have velocities only for enacs#75 , and for one other galaxy in the putative compact group , both of these ( v=19252 for enacs#75 and v=18380 for enacs#78 ) are consistent with the low velocity end of f1 and the mean velocity of g1 .
the main point is that enacs#75 is displaced to the west of the compact group , thus forming a local distortion to the potential well , and perhaps explaining the displacement of the x - ray core .
similarly asymmetric x - ray contours , which appear to follow the galaxy mass distribution , have been found in small groups of galaxies ( mulchaey 1996 ; davis 1996 ; mulchaey 2000 ) .
the compact group also provides a plausible explanation for the existence of the sw x - ray component itself . in this scenario
the sw component reflects the nature of the icm of an infalling group that may be associated with the f1/g1 system , but has not yet interacted with the a3128 icm .
since it has not yet passed though a3128 , its icm is likely still bound in the group , and reflects the local structure in that potential well . by way of contrast ,
the lack of association between the ne component and any prominent enhancement in the galaxy distribution , as well as its more disturbed morphology , is consistent with its identification as stripped gas from the leading group of galaxies in the f1 filament .
further insight regarding the merging event(s ) in a3128/a3125 can be gained by comparing the x - ray image of a3128 to high - resolution _ chandra _ imaging of other clusters with merging events . in several clusters ,
observed with _
chandra _ , signatures of ongoing merging events have been established . in particular , merging events have been studied in the clusters a2142 ( markevitch 2000 ) , a3667 ( vikhlinin , markevitch , & murray 2000a ) , and rxj1720.1 + 2638 ( mazzotta 2001 ) . in all three cases , the merging event appears to consist of a subcluster `` cold '' front moving into the icm of a main cluster . due to the lack of a pressure and temperature inversion across the front
, the inference is that the infalling subcluster is moving subsonically .
this represents a marked contrast to the situation in a3128 , where we see no evidence that the temperature drop sin the ne and sw components , and where we have inferred that the gravitational influence of the h - r supercluster has produced hypersonic infall velocities .
thus the a3128/a3125 system potentially represents a different regime of merger activity .
a more extensive temperature map of the cluster would be very useful in comparing with the `` cold '' front clusters .
thus far we have assumed that all of the diffuse x - ray emission seen in the @xmath39 acis - i image is due to gas associated with the merging events in a3128/a3125 .
the spatial coincidence between the gas in the sw component and the compact galaxy group in that area , coupled with the coincidence between the peak in the sw emission and the luminous galaxy enacs#75 provides circumstantial evidence that the gas in the sw component is , indeed , affiliated with a3128 .
the redshift of 0.065 found for the sw component gas ( see 4 ) provides conclusive proof of its association with the cluster complex .
as well , the redshift of 0.059 found for the diffuse component gas confirms its association with a3128 . in the case of the ne component , no such coincidence occurs with a particular galaxy group .
the strongest circumstantial evidence for a connection between the gas in the ne component with a3128 is the elongated morphology of the emission , that coincides well with the position angle of the a3128/a3125/f1 merger axis , as well as the asymmetric morphology of the outer isophotes , which taken together suggest a direction of motion of the ne component along the main cluster merger axis .
however , barring conclusive redshift information on the x - ray gas , the possibility that the gas could instead be a projection of a background cluster should be considered . in that regard
, we again consider the nature of the radio source(s ) in the ne component , and of the background galaxy that is coincident with the double - lobed source resolved in the atca 13 cm data .
as seen in the ctio 0.9-m optical r band image , in fig.[negal ] , the background galaxy is the brightest of a number of faint red objects in the field . if we assume that the faint fuzz surrounding that galaxy represents the extended envelope of a cd galaxy , i.e. , that it is the dominant galaxy of a background cluster , then by also assuming a typical absolute magnitude for a cd of -22 , we infer a redshift of [email protected] .
the curved blue feature to the southwest of this putative cd galaxy might then be seen as a gravitationally lensed arc .
then , if the x - ray emission in the ne is assumed to be associated with this background cluster , its inferred luminosity is @xmath03 x 10@xmath40 erg / sec in the 0.5 10 kev passband .
while this is a high luminosity , it is not inconsistent with that of the most x - ray luminous clusters known ( e.g. , ebeling 1996 ; wu , xue , & fan 1999 ) .
in addition , while the observed global temperature of the ne component , which is @xmath04 kev , is low compared to the typical @xmath08 kev temperature expected for such an x - ray luminous cluster , there is enough scatter in the l@xmath41 t relation that a @xmath04 kev temperature is not out of the question for the @xmath03 x 10@xmath40 erg / sec luminosity ( wu , xue , & fan 1999 ) .
the key to placing the x - ray gas in the ne component is to obtain a conclusive measure of the redshift using the x - ray emission lines present in the spectrum .
as was summarized in 4 , our best fit model produces a redshift of 0.10 , with a 90% confidence of 0.037 to 0.23 .
thus the redshift data is more consistent with a3128 , at z=0.06 , than with the proposed background cluster at [email protected] , but is not conclusive .
we have collected optical redshift data for 532 objects in the field of the merging double cluster galaxies a3128/a3125 .
the redshift information has been supplemented by both x - ray and radio imaging .
the goal of this program is to characterize and understand the variety of structures still present in the double cluster , and its relation to the larger environment of the horologium - reticulum supercluster .
our principal conclusion is that a large number of substructures are still present in the a3128/a3125 system .
this conclusion is based both on groups and filaments evident in the position - velocity space of the galaxy distribution and on the multi - component nature of the x - ray emitting icm .
the most striking large - scale feature in the galaxy distribution is the relatively low number density of galaxies for @xmath04000 on either side of the mean cluster velocity at @xmath017500 .
we interpret this feature as due to the large gravitational potential well of the h - r supercluster . within the 4000 `` depleted '' zone
, those galaxies that are present appear to be members of small groups or filaments .
the latter are extended features in position - velocity space whose statistical reality is somewhat uncertain .
we ascribe the filaments to the tidal distension of a group after it has fallen through the a3128/a3125 system and is emerging out the other side .
in fact , a3125 itself shows some characteristics of a tidally disturbed system .
we note that the filamentary structures appear to follow the main ne - sw axis along which most merging activity appears to be taking place .
this is the axis which connects a3125 to the larger a3128 ( at a projected separation of @xmath06 mpc , and also along which the x - ray emission is split into two bright components , at a projected separation of @xmath01 mpc .
we have proposed two ongoing merging events to explain the current state of the a3128/a3125 system .
the first is the merger event between a3125 and a3128 , which is most plausibly explained if a3125 has already passed through a3128 , and is highly dispersed as a result of the passage .
the second involves the double peaked nature of the x - ray emission .
here we have noticed the close spatial correspondence between one particular high velocity filament and group with the x - ray emitting gas . although the group / filament system is not massive , in comparison with the main body of a3128 , the high infall velocity generated by the potential well of the h - r supercluster produces a large energy deposition in the collision between the filament and a3128 , and allows for only a modest infalling system to produce a major impact on the cluster icm . specifically , we argue that the morphology of the ne x - ray peak , along with its coincidence with the higher velocity end of the galaxy filament , indicates that the ne x - ray component represents the surviving icm of the galaxy filament that has endured a hypersonic ( @xmath0mach 6 ) encounter with a3128 .
the sw x - ray component appears to be the still intact icm of a compact group that represents the still - infalling end of the galaxy filament - group .
this gas is believed to be responding to the potential well of a compact group of galaxies surrounding a bright galaxy . while the details of this picture are still quite uncertain , the key ingredient is the high encounter velocity produced by the h - r supercluster , which makes even the infall of a small group an energetic and interesting event in the life of a galaxy cluster .
further observations which could clarify this picture include deeper x - ray observations , to characterize the temperature structure in the icm , deep high resolution radio imaging to clarify the nature of the putative radio arc associated with the ne x - ray peak , redshifts of galaxies in the apparent compact group that is coincident with sw x - ray component , and further redshift surveys throughout the h - r supercluster .
we wish to thank the staff of the aat for acquiring the 2df data for us under the service observing program and for the significant help given to us in all phases of the observations .
the most is operated by the university of sydney and supported in part by grants from the australian research council .
the australia telescope is funded by the commonwealth of australia for operation as a national facility managed by csiro .
this research has been partially supported by nsf grant ast-9900720 to the university of north carolina .
mjh acknowledges the financial support of a physics & mathematical physics scholarship at the university of adelaide .
as has been discussed in 6 , the f1 galaxy filament ( and probably associated g1 group ) represents the strongest candidate for producing the transient structure seen in the chandra image of the icm .
if we assume that f1 is indeed responsible for the ne and sw x - ray components , then a comparison of the projected separation between the ne end of f1 and the outer edge of the ne x - ray core can yield an estimate of the relative velocities of the f1 filament and the x - ray sources .
if the x - ray structure was generated at the time when the f1 group first entered the a3128 cluster , then the angular separation is where @xmath43 is the time elapsed since the entry into a3128 , @xmath44 is the average velocity of f1 , @xmath45 is the average velocity of the leading edge of the ne x - ray core , and @xmath46 is the distance to the cluster . from figure 7 , the ne `` head '' of f1 is separated from the leading edge of the ne x - ray source by @xmath4730 .
the point at which f1 first entered the a3128 cluster is uncertain because a3128 does not have a sharp boundary , however , it appears that the sw edge of the a3128 galaxy distribution ( c.f .
figure 7 ) is about 15 beyond the sw x - ray lobe .
thus , the projected angular distance traveled by the `` head '' of f1 since entering a3128 is from the ratio of these two angular separations we can therefore estimate that the average velocity of the ne leading edge of the x - ray source is about 1/2 the average velocity of the leading edge of the f1 group .
the radial velocity of the `` head '' of f1 is about 4000 relative to the a3128 rest frame which implies that the radial velocity of the x - ray structure is about 2000 ( again , relative to the a3128 rest frame ) . assuming a random inclination with respect to the plane of the sky then implies that the observed x - ray structure has a velocity relative to the a3128 cluster of @xmath03500 .
this corresponds to a supersonic mach number , m=6 , for a typical cluster gas temperature of 3 kev .
this is obviously consistent with the observed axial structure in the x - ray source .
we therefore hypothesize that the ne component of the a3128 x - ray source(s ) consists of shocked gas which has been stripped from a group of galaxies ( now seen as the f1 filament ) that encountered and passed through a3128 at hypersonic velocity ( in excess of 4000 ) . in this scenario ,
as the f1 group approaches the a3128 cluster the gas held in the core of the f1 group is shocked by the encounter with the gas bound to a3128 . a bow shock forms and simultaneously a reverse shock propagates back through the f1 gas .
this shock system is responsible for decelerating and removing the gas from the f1 group and depositing it in the a3128 cluster . where @xmath50 is the ideal gas adiabatic index , @xmath51 is the velocity of the reverse shock relative to the f1 group , @xmath52 is the pressure in the reverse shock and @xmath53 is the local density of the unshocked gas in the f1 group at a particular point .
the region between the bow shock and the reverse shock , however , is nearly isobaric so the reverse shock pressure is where @xmath55 is the bow shock pressure , @xmath56 is the ambient density in abell 3128 , and @xmath44 is the incoming velocity of the f1 group ( relative to the barycenter of a3128 ) . as a result
there is a simple relation between the reverse shock velocity and the incoming f1 group velocity , i.e. if this reverse shock velocity is greater than escape velocity for the f1 precursor group , then the gas behind the shock front will be stripped from the f1 group and eventually merge with the gas in a3128 . in that case , the initial velocity of the stripped gas relative to a3128 is simply the difference between the f1 group velocity and the reverse shock velocity , i.e. note that the velocity of the stripped gas relative to a3128 depends critically on the density ratio , @xmath59 .
if @xmath60 it simply means that the gas in f1 will be stopped and deposited on the outer perimeter of a3128 as the f1 group begins to penetrate ( @xmath61 ) .
a negative stripping velocity indicates that the reverse shock gas actually has a net ( small ) velocity opposite to @xmath44 .
this mass of rapidly stripped lower density gas may contribute to the central x - ray component , or it may be trapped in the potential well associated with the g1 group . if , however , there is a high density core in the initial f1 gas distribution , as would be the case for an isothermal density distribution , then the @xmath59 ratio for this high density component would likely be less than 1 .
this high density gas would still be strongly decelerated ( and thus displaced ) from the f1 group , but would continue to penetrate into a3128 at a high ( supersonic ) velocity as given above for @xmath62 . clearly , in this case @xmath63 . in other words ,
as f1 enters and passes through a3128 , its gas content is essentially peeled away in layers corresponding to the quiescent density of the gas initially in the core of f1 , leaving a trail of stripped gas in its wake .
the above estimate that the initial velocity of the shocked , dense core gas represented by the ne x - ray component is about 1/2 of the velocity of f1 then implies that @xmath64 for the central ( highest ) density in the core . as the reverse shock passes through the core of f1 , the core gas decelerates and
is displaced from the group of galaxies .
in addition , the bow shock wraps around the slower moving shocked gas from the core center and establishes the standard condition for ram pressure confinement , i.e. where @xmath66 is the bulk velocity of the central core gas relative to a3128 , @xmath67 is the density of the shocked central gas , and @xmath68 is the sound speed of this core gas .
note that the initial value for @xmath66 is equal to @xmath69 for @xmath70 , i.e. , the central density in the unshocked gas .
a quantitative indication of ram pressure confinement is provided by the observed asymmetry in the ne x - ray core , with the highest surface brightness contours displaced toward the outer edge of the emission .
this is a standard signature of ram pressure confinement and deceleration .
furthermore , the axially elongated structure of the ne core can be explained in the context of a ram pressure confinement model by the following argument . ram pressure confinement controls the transverse expansion of the x - ray source , but it does not control the axial expansion as the core gas backfills its trailing wake cavity at the speed of sound . the ratio of these two expansion rates is where @xmath72 is simply the time derivative of the transverse radius , @xmath69 is the initial velocity of the ram pressure confined central core gas relative to a3128 , and @xmath73 is the density ratio . for the ne component ,
using the above estimate that @xmath74 and @xmath75 we conclude that @xmath76 . since the x - ray emitting cloud is slowing down , @xmath77 is increasing with time , so the observed value of @xmath78 in conjunction with our estimate that @xmath76 implies that the present velocity of the x - ray core has been reduced to about 40% of its initial value . from the central peaks of the two components we estimate that the central densities of the two x - ray components are @xmath88 and @xmath89 .
thus cooling times in these components are @xmath90 yr and @xmath91 yr .
although these cooling times are greater than the dynamical timescale , they are at most a factor of 2 - 4 larger , hence radiative losses may be an important dynamical factor in slowing the expansion of the x - ray sources as they decelerate ( c.f . ,
christiansen , rolison , & scott 1979 ) .
hambly , n. c. , macgillivray , h. t. , read , m. a. , tritton , s. b. , thomson , e. b. , kelly , b. d. , morgan , d. h. , smith , r. e. , driver , s. p. , williamson , j. , parker , q. a. , hawkins , m. r. s. , williams , p. m. , & lawrence , a. 2001 , , 326 , 1279 cccccr 1991 dec 03 & 843 & most & 151600 & 03 30 12.4 @xmath1052 33 48 & 12.0 + 1999 feb 28 & 1384/2496 & 6c & 1536000 & 03 31 15.0 @xmath1052 41 48 & 1.67 + 2000 mar 23 & 1384/2496 & 6d & 775878 & 03 31 15.0 @xmath1052 41 48 & 1.15 + 2000 nov 04 & 1384/2496 & 6d & 775878 & 03 31 15.0 @xmath1052 41 48 & 0.65 + [ radio_obs ] rrrrrrrrrrrrc 1 & & 03 & 22 & 53.67 & -52 & 39 & 51.1 & 16.8 & 18365 & a & 103.5 & n + 3 & & 03 & 23 & 01.99 & -52 & 58 & 42.2 & 18.3 & 26219 & e & 98.1 & y + 5 & & 03 & 23 & 10.56 & -53 & 05 & 08.1 & 17.7 & 17682 & a & 32.4 & n + 6 & & 03 & 23 & 15.56 & -52 & 53 & 44.1 & 17.3 & 17573 & a & 37.1 & n + 7 & & 03 & 23 & 20.91 & -52 & 26 & 20.5 & 18.4 & 34195 & e & 68.9 & y + 8 & & 03 & 23 & 23.45 & -53 & 08 & 09.2 & 17.8 & 9877 & e & 189.7 & y + 9 & & 03 & 23 & 26.54 & -52 & 29 & 03.5 & 16.9 & 21977 & ae & 63.9 & y + 10 & & 03 & 23 & 38.52 & -53 & 13 & 48.7 & 16.4 & 17951 & a & 42.4 & y + 11 & & 03 & 23 & 38.72 & -53 & 18 & 32.7 & 18.4 & 32721 & ae & 45.9 & y + 13 & & 03 & 23 & 41.41 & -53 & 10 & 00.5 & 18.1 & 322 & a & 48.5 & n + 15 & & 03 & 23 & 42.75 & -52 & 42 & 52.6 & 17.0 & 23583 & a & 263.6 & n + 16 & & 03 & 23 & 43.41 & -53 & 19 & 16.9 & 17.8 & 17859 & e & 161.1 & y + 17 & & 03 & 23 & 44.20 & -53 & 15 & 25.1 & 17.7 & 19591 & a & 71.2 & y + 19 & & 03 & 23 & 44.57 & -53 & 17 & 32.7 & 16.6 & 17134 & ae & 239.9 & y + 21 & & 03 & 23 & 44.59 & -53 & 20 & 46.8 & 16.7 & 23350 & ae & 39.9 & y + 25 & & 03 & 23 & 45.33 & -52 & 50 & 27.4 & 17.2 & 23026 & ae & 58.1 & y + 29 & & 03 & 23 & 51.31 & -53 & 19 & 04.9 & 17.2 & 17790 & a & 132.6 & y + 32 & & 03 & 23 & 59.84 & -52 & 43 & 26.7 & 16.4 & 21935 & ae & 63.9 & y + 33 & & 03 & 24 & 00.13 & -52 & 34 & 23.6 & 16.9 & 119 & a & 35.6 & n + 35 & & 03 & 24 & 01.67 & -53 & 08 & 18.5 & 16.9 & 17604 & ae & 98.5 & y + 36 & & 03 & 24 & 04.33 & -52 & 22 & 51.4 & 17.9 & 19437 & e & 89.8 & y + 38 & & 03 & 24 & 04.67 & -53 & 18 & 53.9 & 17.7 & 23122 & ae & 104.6 & y + 40 & & 03 & 24 & 07.26 & -52 & 49 & 43.2 & 17.4 & 24009 & ae & 104.0 & y + 42 & & 03 & 24 & 12.05 & -53 & 20 & 32.9 & 15.9 & 17808 & ae & 46.7 & y + 47 & & 03 & 24 & 20.70 & -52 & 57 & 35.5 & 18.4 & 17719 & a & 95.5 & n + 48 & & 03 & 24 & 24.35 & -52 & 42 & 48.4 & 16.3 & 18211 & ae & 82.3 & y + 50 & & 03 & 24 & 26.47 & -52 & 38 & 04.6 & 17.4 & 21971 & ae & 97.6 & y + 53 & & 03 & 24 & 30.52 & -52 & 49 & 35.0 & 18.2 & -130 & a & 229.8 & n + 54 & & 03 & 24 & 33.76 & -53 & 07 & 42.5 & 18.3 & 32638 & ae & 130.0 & y + 56 & & 03 & 24 & 35.16 & -52 & 37 & 24.5 & 17.6 & 23206 & ae & 102.2 & y + 58 & & 03 & 24 & 38.33 & -52 & 49 & 48.9 & 18.2 & 18306 & a & 62.1 & n + 59 & & 03 & 24 & 42.40 & -52 & 33 & 53.8 & 17.3 & 14007 & e & 84.5 & y + 60 & & 03 & 24 & 45.31 & -53 & 25 & 45.2 & 16.2 & 17716 & ae & 44.0 & y + 62 & & 03 & 24 & 46.45 & -53 & 02 & 09.9 & 15.6 & 18 & a & 13.0 & n + 64 & & 03 & 24 & 47.90 & -53 & 22 & 33.0 & 17.9 & 17894 & a & 42.2 & y + 66 & & 03 & 24 & 49.04 & -52 & 39 & 10.8 & 18.2 & 52673 & e & 109.9 & y + 68 & & 03 & 24 & 57.48 & -53 & 23 & 10.8 & 17.9 & 16701 & ae & 68.5 & y + 69 & & 03 & 24 & 58.47 & -53 & 24 & 58.9 & 16.3 & 17956 & ae & 44.8 & y + 70 & & 03 & 25 & 02.21 & -52 & 08 & 40.1 & 16.8 & 21324 & a & 75.4 & n + 71 & & 03 & 25 & 02.94 & -53 & 14 & 38.3 & 17.5 & 24084 & a & 62.6 & y + 73 & & 03 & 25 & 03.22 & -52 & 43 & 33.2 & 18.2 & 22072 & a & 58.7 & y + 74 & & 03 & 25 & 03.81 & -52 & 48 & 44.6 & 17.9 & 18351 & a & 126.7 & n + 75 & & 03 & 25 & 07.82 & -53 & 19 & 10.8 & 18.3 & 17966 & ae & 122.6 & y + 76 & & 03 & 25 & 11.28 & -52 & 33 & 12.6 & 17.7 & 18227 & ae & 48.0 & y + 77 & & 03 & 25 & 12.75 & -52 & 17 & 11.5 & 17.3 & 23119 & ae & 57.5 & y + 78 & & 03 & 25 & 16.36 & -53 & 35 & 38.5 & 16.9 & 17825 & ae & 69.1 & y + 79 & & 03 & 25 & 17.04 & -52 & 51 & 44.8 & 16.3 & 13813 & ae & 68.4 & y + 80 & & 03 & 25 & 18.49 & -53 & 21 & 28.7 & 17.0 & 17996 & ae & 182.1 & y + 81 & & 03 & 25 & 18.99 & -52 & 43 & 42.4 & 17.5 & 21835 & a & 76.9 & y + 82 & & 03 & 25 & 20.10 & -52 & 32 & 05.8 & 16.9 & 18181 & ae & 47.1 & y + 83 & & 03 & 25 & 20.86 & -52 & 06 & 29.3 & 17.9 & 18175 & a & 46.7 & n + 84 & & 03 & 25 & 22.32 & -53 & 24 & 40.9 & 17.7 & 17651 & ae & 62.8 & y + 85 & & 03 & 25 & 22.45 & -52 & 12 & 09.2 & 17.2 & 20750 & ae & 89.7 & y + 87 & & 03 & 25 & 27.03 & -52 & 57 & 54.4 & 17.4 & 31826 & e & 108.3 & y + 88 & & 03 & 25 & 33.01 & -52 & 39 & 27.3 & 18.5 & 39719 & ae & 219.1 & y + 89 & & 03 & 25 & 35.66 & -52 & 34 & 54.1 & 18.2 & 18343 & ae & 225.0 & y + 94 & & 03 & 25 & 41.72 & -52 & 55 & 04.3 & 18.5 & 176 & a & 109.9 & n + 95 & & 03 & 25 & 42.50 & -52 & 31 & 00.0 & 16.0 & 18901 & a & 37.7 & y + 96 & & 03 & 25 & 44.42 & -52 & 32 & 58.2 & 16.1 & 18360 & ae & 52.1 & y + 97 & & 03 & 25 & 45.24 & -52 & 51 & 45.4 & 18.3 & 17347 & e & 58.6 & y + 99 & & 03 & 25 & 46.62 & -53 & 28 & 53.7 & 17.4 & 32695 & ae & 148.6 & y + 100 & & 03 & 25 & 47.75 & -52 & 53 & 32.8 & 17.5 & 18078 & e & 76.1 & y + 101 & & 03 & 25 & 49.68 & -53 & 23 & 03.6 & 17.0 & 19117 & ae & 111.3 & y + 102 & & 03 & 25 & 51.06 & -52 & 16 & 22.4 & 17.5 & 21152 & a & 68.0 & n + 104 & & 03 & 25 & 58.52 & -53 & 10 & 19.4 & 18.0 & 22853 & e & 197.2 & y + 105 & & 03 & 26 & 00.95 & -52 & 10 & 37.7 & 17.8 & 16436 & e & 93.5 & y + 106 & & 03 & 26 & 01.07 & -53 & 26 & 06.3 & 16.2 & 18354 & ae & 23.0 & y + 109 & & 03 & 26 & 02.42 & -52 & 08 & 04.5 & 18.2 & 39381 & a & 177.4 & n + 110 & & 03 & 26 & 03.59 & -52 & 35 & 55.5 & 17.7 & 18017 & e & 217.4 & y + 113 & & 03 & 26 & 09.69 & -53 & 10 & 51.9 & 17.7 & 23206 & ae & 52.9 & y + 114 & & 03 & 26 & 10.22 & -52 & 54 & 49.1 & 18.1 & 31781 & a & 16.7 & n + 115 & & 03 & 26 & 10.24 & -52 & 34 & 56.9 & 16.8 & 25113 & a & 206.1 & y + 117 & & 03 & 26 & 13.02 & -52 & 28 & 37.9 & 16.9 & 18194 & a & 43.0 & n + 118 & & 03 & 26 & 15.69 & -53 & 20 & 45.1 & 17.0 & 19370 & a & 52.5 & n + 119 & & 03 & 26 & 18.02 & -52 & 49 & 09.1 & 17.1 & 17390 & e & 86.4 & y + 120 & & 03 & 26 & 18.50 & -53 & 32 & 47.3 & 15.8 & 18870 & ae & 70.3 & y + 121 & & 03 & 26 & 21.60 & -52 & 55 & 01.3 & 18.2 & 31792 & e & 56.5 & y + 122 & & 03 & 26 & 23.13 & -52 & 39 & 43.3 & 16.5 & 18049 & ae & 74.5 & y + 123 & & 03 & 26 & 26.16 & -52 & 12 & 29.4 & 17.7 & 17807 & e & 164.7 & y + 124 & & 03 & 26 & 26.41 & -52 & 30 & 24.8 & 17.7 & 17759 & a & 86.3 & n + 126 & & 03 & 26 & 27.05 & -53 & 28 & 58.6 & 17.8 & 18330 & e & 59.2 & y + 127 & & 03 & 26 & 30.91 & -53 & 37 & 34.9 & 17.3 & 18792 & a & 245.2 & y + 128 & & 03 & 26 & 31.94 & -53 & 39 & 04.6 & 16.5 & 18563 & ae & 69.8 & y + 129 & & 03 & 26 & 32.10 & -52 & 59 & 31.7 & 16.9 & 31842 & a & 76.3 & y + 130 & & 03 & 26 & 35.35 & -53 & 20 & 07.0 & 16.9 & 19552 & ae & 111.5 & y + 131 & & 03 & 26 & 38.53 & -52 & 52 & 46.2 & 16.8 & 18259 & ae & 119.4 & y + 133 & & 03 & 26 & 40.18 & -52 & 39 & 35.7 & 18.5 & 19110 & e & 243.2 & y + 134 & & 03 & 26 & 44.39 & -52 & 00 & 50.6 & 17.9 & 26506 & e & 224.9 & y + 135 & & 03 & 26 & 45.09 & -53 & 05 & 40.8 & 18.3 & 23327 & e & 103.7 & y + 136 & & 03 & 26 & 47.16 & -52 & 57 & 45.6 & 17.3 & -41 & a & 90.8 & n + 137 & & 03 & 26 & 50.23 & -53 & 21 & 39.1 & 17.5 & 32354 & ae & 96.3 & y + 138 & & 03 & 26 & 50.68 & -53 & 31 & 05.2 & 17.1 & 18729 & a & 46.6 & n + 141 & & 03 & 26 & 56.83 & -52 & 53 & 19.4 & 17.3 & 19606 & a & 46.7 & y + 142 & & 03 & 26 & 57.99 & -52 & 55 & 51.5 & 18.0 & 22709 & ae & 48.1 & y + 143 & & 03 & 27 & 01.66 & -53 & 09 & 15.6 & 16.9 & 19987 & a & 58.8 & y + 145 & & 03 & 27 & 03.80 & -53 & 28 & 23.2 & 18.0 & 18825 & e & 52.0 & y + 146 & & 03 & 27 & 05.04 & -53 & 21 & 41.7 & 16.9 & 18693 & e & 116.2 & y + 149 & & 03 & 27 & 08.30 & -52 & 10 & 15.3 & 17.6 & 13441 & ae & 87.2 & y + 150 & & 03 & 27 & 08.44 & -53 & 37 & 26.3 & 17.5 & 17565 & ae & 263.8 & y + 151 & & 03 & 27 & 10.31 & -53 & 00 & 36.4 & 15.6 & 9083 & ae & 187.9 & y + 152 & & 03 & 27 & 10.53 & -51 & 55 & 51.5 & 18.3 & 60556 & e & 143.5 & y + 153 & & 03 & 27 & 10.85 & -53 & 39 & 46.9 & 14.1 & 17799 & ae & 72.7 & y + 155 & & 03 & 27 & 12.96 & -53 & 02 & 35.1 & 17.5 & 17955 & ae & 92.9 & y + 156 & & 03 & 27 & 16.88 & -52 & 35 & 03.5 & 18.2 & 18267 & a & 52.7 & n + 157 & & 03 & 27 & 19.06 & -52 & 47 & 35.3 & 18.1 & 24478 & a & 397.8 & n + 158 & & 03 & 27 & 20.20 & -53 & 28 & 34.5 & 17.3 & 18476 & ae & 145.0 & y + 159 & & 03 & 27 & 20.35 & -52 & 34 & 14.2 & 18.4 & 19512 & e & 109.8 & y + 165 & & 03 & 27 & 27.70 & -52 & 29 & 14.9 & 17.9 & 19723 & e & 0.0 & y + 168 & & 03 & 27 & 29.36 & -52 & 52 & 48.7 & 17.6 & 22293 & ae & 135.0 & y + 170 & & 03 & 27 & 30.99 & -52 & 38 & 18.4 & 18.1 & 18489 & a & 162.8 & n + 172 & & 03 & 27 & 31.09 & -53 & 23 & 40.4 & 17.6 & 19359 & a & 49.8 & n + 174 & & 03 & 27 & 32.98 & -53 & 20 & 22.3 & 16.5 & 17866 & a & 69.8 & y + 175 & & 03 & 27 & 33.98 & -53 & 40 & 07.3 & 17.6 & 19005 & a & 32.8 & n + 178 & & 03 & 27 & 40.10 & -51 & 59 & 22.4 & 18.0 & 17 & a & 36.8 & n + 180 & & 03 & 27 & 40.38 & -53 & 33 & 31.6 & 17.7 & 17834 & e & 93.7 & y + 181 & & 03 & 27 & 41.83 & -52 & 32 & 49.5 & 17.6 & 19035 & e & 232.1 & y + 186 & & 03 & 27 & 42.76 & -53 & 38 & 23.8 & 17.4 & 17801 & ae & 118.7 & y + 188 & & 03 & 27 & 42.90 & -53 & 34 & 23.6 & 16.3 & 18430 & a & 63.3 & n + 191 & & 03 & 27 & 45.63 & -52 & 10 & 41.8 & 18.0 & -20 & a & 60.5 & n + 193 & & 03 & 27 & 47.26 & -52 & 30 & 04.1 & 17.8 & 9 & a & 13.1 & n + 195 & & 03 & 27 & 47.91 & -52 & 31 & 21.2 & 18.0 & 18525 & a & 54.1 & n + 197 & & 03 & 27 & 49.07 & -52 & 24 & 05.9 & 17.0 & 17924 & ae & 40.7 & y + 203 & & 03 & 27 & 52.55 & -53 & 24 & 08.5 & 16.2 & 18405 & ae & 40.9 & y + 204 & & 03 & 27 & 53.12 & -51 & 53 & 48.2 & 17.3 & -32 & a & 60.1 & n + 206 & & 03 & 27 & 53.93 & -52 & 23 & 03.8 & 17.8 & 17935 & a & 34.6 & n + 207 & & 03 & 27 & 53.98 & -53 & 35 & 10.3 & 17.8 & 17974 & ae & 559.8 & y + 209 & & 03 & 27 & 54.18 & -53 & 46 & 57.3 & 16.7 & 16034 & a & 27.2 & n + 217 & & 03 & 28 & 03.56 & -52 & 06 & 53.7 & 18.1 & 17264 & e & 16.9 & y + 218 & & 03 & 28 & 03.94 & -53 & 33 & 02.2 & 18.0 & 28250 & ae & 109.9 & y + 220 & & 03 & 28 & 04.41 & -52 & 44 & 42.2 & 18.4 & 18106 & e & 201.6 & y + 223 & & 03 & 28 & 07.49 & -52 & 36 & 50.9 & 18.4 & 16869 & ae & 62.9 & y + 224 & & 03 & 28 & 08.56 & -52 & 30 & 31.7 & 18.1 & 11808 & ae & 250.6 & y + 225 & e24 & 03 & 28 & 10.28 & -52 & 48 & 59.7 & 18.3 & 17939 & e & 126.8 & y + 226 & & 03 & 28 & 11.45 & -53 & 38 & 18.7 & 17.7 & 33310 & a & 131.1 & n + 228 & & 03 & 28 & 12.82 & -52 & 15 & 59.3 & 18.3 & 10 & a & 137.5 & n + 229 & & 03 & 28 & 14.03 & -53 & 38 & 34.5 & 17.4 & 33218 & ae & 201.6 & y + 231 & & 03 & 28 & 15.31 & -52 & 42 & 01.1 & 18.3 & 18031 & ae & 53.7 & y + 233 & & 03 & 28 & 16.51 & -52 & 13 & 57.6 & 18.4 & 47184 & a & 29.0 & n + 235 & & 03 & 28 & 17.33 & -51 & 50 & 53.3 & 17.7 & 18572 & e & 137.2 & y + 236 & & 03 & 28 & 17.86 & -53 & 34 & 59.3 & 18.3 & 17084 & a & 65.0 & n + 237 & & 03 & 28 & 21.35 & -52 & 35 & 19.6 & 16.6 & -43 & a & 60.7 & n + 240 & & 03 & 28 & 23.20 & -53 & 42 & 32.6 & 18.1 & 21534 & ae & 145.3 & y + 241 & & 03 & 28 & 23.38 & -53 & 34 & 36.0 & 16.2 & 18880 & ae & 188.7 & y + 242 & & 03 & 28 & 24.53 & -51 & 58 & 52.1 & 17.4 & 18834 & e & 73.1 & y + 245 & & 03 & 28 & 27.46 & -51 & 55 & 30.8 & 17.1 & 17183 & e & 67.2 & y + 247 & & 03 & 28 & 28.17 & -53 & 27 & 28.0 & 17.6 & 79 & a & 11.6 & n + 248 & & 03 & 28 & 28.18 & -53 & 47 & 12.7 & 17.4 & 17896 & ae & 61.1 & y + 251 & & 03 & 28 & 30.56 & -53 & 33 & 48.5 & 17.9 & 17730 & a & 176.6 & y + 253 & e31,cr27a & 03 & 28 & 32.40 & -52 & 57 & 09.7 & 17.2 & 18047 & a & 174.2 & n + 254 & e30 & 03 & 28 & 32.60 & -52 & 15 & 56.3 & 18.1 & 23130 & e & 109.1 & y + 255 & & 03 & 28 & 33.17 & -53 & 11 & 40.5 & 17.6 & 22844 & e & 91.2 & y + 256 & & 03 & 28 & 34.44 & -53 & 24 & 35.1 & 17.0 & 18125 & ae & 64.1 & y + 257 & & 03 & 28 & 35.08 & -53 & 48 & 43.1 & 18.1 & 33559 & ae & 149.6 & y + 261 & & 03 & 28 & 39.75 & -53 & 19 & 29.2 & 18.3 & 22874 & e & 241.4 & y + 263 & & 03 & 28 & 42.33 & -53 & 29 & 13.6 & 17.7 & 18092 & a & 95.4 & n + 264 & e33 & 03 & 28 & 44.99 & -53 & 03 & 27.4 & 17.4 & 23038 & ae & 56.6 & y + 267 & e34,cr20a & 03 & 28 & 45.99 & -52 & 58 & 50.5 & 16.8 & 18284 & a & 265.1 & n + 268 & & 03 & 28 & 46.69 & -52 & 38 & 23.5 & 18.1 & 17084 & a & 76.2 & n + 270 & & 03 & 28 & 47.45 & -52 & 33 & 59.6 & 18.2 & 18002 & e & 129.5 & y + 271 & e37 & 03 & 28 & 48.73 & -52 & 27 & 18.8 & 17.6 & 17692 & ae & 68.3 & y + 272 & & 03 & 28 & 49.09 & -53 & 27 & 52.0 & 18.1 & 17564 & e & 206.0 & y + 274 & e38 & 03 & 28 & 51.90 & -52 & 13 & 52.1 & 16.7 & 16664 & ae & 41.9 & y + 276 & & 03 & 28 & 53.64 & -52 & 48 & 12.5 & 18.3 & 18235 & a & 56.9 & n + 277 & e40 & 03 & 28 & 55.01 & -52 & 12 & 58.6 & 17.3 & 18116 & a & 145.6 & n + 281 & & 03 & 28 & 59.76 & -53 & 22 & 20.4 & 17.8 & 23078 & e & 406.4 & y + 285 & e46 & 03 & 29 & 05.94 & -52 & 46 & 32.1 & 17.9 & 16772 & a & 39.8 & n + 287 & & 03 & 29 & 08.09 & -53 & 42 & 15.4 & 18.1 & 17872 & e & 108.6 & y + 288 & & 03 & 29 & 09.45 & -52 & 48 & 52.9 & 18.3 & 17427 & e & 436.8 & y + 290 & e51 & 03 & 29 & 16.51 & -52 & 37 & 38.1 & 15.8 & 18378 & a & 104.6 & n + 292 & & 03 & 29 & 18.95 & -52 & 12 & 38.7 & 18.5 & 18054 & a & 68.3 & n + 293 & & 03 & 29 & 20.75 & -52 & 47 & 58.5 & 17.8 & 24720 & e & 114.3 & y + 294 & e54 & 03 & 29 & 22.12 & -52 & 37 & 10.3 & 15.0 & 11664 & a & 42.1 & n + 295 & & 03 & 29 & 23.59 & -51 & 56 & 36.4 & 18.3 & 50551 & a & 228.4 & n + 296 & & 03 & 29 & 24.32 & -53 & 16 & 08.5 & 18.5 & 12005 & ae & 160.6 & y + 297 & & 03 & 29 & 24.73 & -52 & 33 & 02.4 & 18.1 & 89 & a & 34.2 & n + 300 & e58,cr118a & 03 & 29 & 26.47 & -52 & 30 & 27.6 & 17.0 & 19555 & a & 39.4 & n + 301 & & 03 & 29 & 27.11 & -52 & 40 & 54.5 & 17.3 & 31884 & ae & 49.5 & y + 305 & & 03 & 29 & 28.56 & -52 & 42 & 36.4 & 17.8 & 16378 & ae & 92.7 & y + 306 & e61,cr167a & 03 & 29 & 29.96 & -52 & 17 & 58.2 & 17.0 & 18076 & a & 45.1 & n + 310 & e66 & 03 & 29 & 33.38 & -52 & 46 & 33.9 & 17.5 & 16747 & a & 62.4 & n + 311 & & 03 & 29 & 33.82 & -53 & 45 & 50.5 & 16.8 & 24763 & a & 55.5 & y + 312 & e68 & 03 & 29 & 34.47 & -52 & 51 & 00.2 & 17.5 & 17104 & a & 65.9 & n + 313 & & 03 & 29 & 34.92 & -53 & 33 & 03.9 & 17.9 & 17486 & ae & 38.5 & y + 315 & e67 & 03 & 29 & 35.39 & -52 & 12 & 52.0 & 17.9 & 17848 & a & 27.2 & y + 316 & & 03 & 29 & 36.01 & -52 & 34 & 16.5 & 18.5 & 19163 & a & 59.6 & n + 317 & & 03 & 29 & 36.66 & -52 & 29 & 35.0 & 18.1 & 16657 & a & 194.9 & n + 319 & & 03 & 29 & 37.01 & -53 & 34 & 55.1 & 18.1 & 34900 & e & 108.9 & y + 320 & e70 & 03 & 29 & 37.80 & -52 & 41 & 55.5 & 16.7 & 17152 & ae & 89.7 & y + 321 & & 03 & 29 & 37.97 & -52 & 49 & 12.8 & 18.3 & 18472 & a & 303.2 & n + 326 & & 03 & 29 & 41.46 & -52 & 29 & 35.4 & 17.6 & 17511 & ae & 48.3 & y + 327 & e73 & 03 & 29 & 41.57 & -52 & 31 & 16.6 & 18.1 & 18825 & a & 72.1 & n + 328 & & 03 & 29 & 43.49 & -53 & 15 & 34.3 & 18.5 & 22916 & e & 380.5 & y + 330 & & 03 & 29 & 45.76 & -52 & 00 & 22.0 & 18.3 & 21270 & a & 55.5 & n + 331 & & 03 & 29 & 46.62 & -52 & 43 & 51.9 & 18.1 & -16 & a & 60.9 & n + 332 & & 03 & 29 & 46.70 & -52 & 37 & 05.7 & 18.4 & 18712 & a & 96.1 & n + 335 & & 03 & 29 & 50.98 & -53 & 27 & 58.9 & 17.2 & 23004 & ae & 46.4 & y + 337 & & 03 & 29 & 51.92 & -52 & 40 & 43.4 & 17.6 & 18383 & a & 68.7 & y + 341 & e77 & 03 & 29 & 52.85 & -52 & 51 & 55.1 & 17.3 & 17131 & a & 45.1 & n + 343 & & 03 & 29 & 53.54 & -53 & 35 & 33.4 & 17.7 & 27973 & a & 99.6 & n + 348 & & 03 & 29 & 56.24 & -52 & 37 & 28.2 & 18.3 & 19454 & a & 331.5 & n + 349 & e80 & 03 & 29 & 56.83 & -53 & 09 & 03.3 & 16.7 & 19920 & e & 201.8 & y + 351 & & 03 & 29 & 59.78 & -53 & 26 & 47.0 & 17.5 & 17904 & ae & 62.8 & y + 352 & & 03 & 29 & 59.88 & -52 & 38 & 13.1 & 18.3 & 19309 & a & 98.4 & n + 353 & e81 & 03 & 30 & 00.33 & -52 & 18 & 31.8 & 18.1 & 45606 & a & 140.2 & n + 355 & & 03 & 30 & 03.00 & -52 & 55 & 30.6 & 18.3 & 16761 & a & 98.2 & n + 361 & e84 & 03 & 30 & 05.04 & -52 & 15 & 25.1 & 17.9 & -29 & a & 64.8 & n + 365 & & 03 & 30 & 06.11 & -53 & 42 & 37.5 & 15.4 & 15940 & ae & 43.9 & y + 367 & & 03 & 30 & 08.47 & -53 & 23 & 51.8 & 17.3 & 17826 & e & 122.6 & y + 368 & & 03 & 30 & 09.11 & -53 & 18 & 47.0 & 18.2 & 22630 & e & 100.3 & y + 370 & & 03 & 30 & 10.39 & -53 & 45 & 18.6 & 17.4 & 17684 & ae & 105.5 & y + 371 & & 03 & 30 & 10.52 & -53 & 34 & 06.6 & 17.3 & 18419 & e & 65.3 & y + 372 & & 03 & 30 & 10.58 & -52 & 27 & 07.6 & 17.7 & 17386 & a & 38.5 & n + 373 & e89 & 03 & 30 & 11.13 & -52 & 14 & 51.6 & 17.7 & 44935 & e & 235.6 & y + 374 & e90 & 03 & 30 & 11.60 & -53 & 09 & 03.5 & 18.0 & 19884 & ae & 63.4 & y + 376 & & 03 & 30 & 12.81 & -52 & 45 & 04.9 & 18.2 & 17817 & a & 56.8 & n + 377 & & 03 & 30 & 12.95 & -51 & 52 & 31.3 & 17.4 & 25686 & a & 236.6 & n + 378 & & 03 & 30 & 13.18 & -53 & 47 & 03.2 & 16.5 & 15833 & e & 179.1 & y + 380 & e94 & 03 & 30 & 13.60 & -53 & 14 & 33.1 & 17.5 & 17506 & a & 46.0 & n + 388 & & 03 & 30 & 19.46 & -53 & 37 & 03.4 & 17.8 & 17726 & a & 159.0 & n + 390 & e99,cr130a & 03 & 30 & 22.10 & -52 & 26 & 07.6 & 17.6 & 17739 & a & 48.5 & n + 393 & e103 & 03 & 30 & 23.11 & -53 & 02 & 58.9 & 18.0 & 22274 & ae & 115.8 & y + 396 & & 03 & 30 & 24.09 & -52 & 44 & 37.2 & 18.2 & 19884 & ae & 63.8 & y + 397 & & 03 & 30 & 25.00 & -53 & 36 & 50.7 & 17.7 & 17791 & a & 111.0 & n + 402 & & 03 & 30 & 32.24 & -52 & 38 & 49.7 & 17.6 & 19651 & a & 80.9 & n + 403 & & 03 & 30 & 33.78 & -53 & 18 & 40.9 & 17.5 & 14208 & a & 32.4 & y + 404 & & 03 & 30 & 33.85 & -52 & 23 & 52.6 & 18.4 & 18229 & e & 16.0 & y + 414 & cr128a & 03 & 30 & 39.71 & -52 & 26 & 28.4 & 18.3 & 17405 & a & 127.6 & n + 415 & & 03 & 30 & 40.99 & -53 & 15 & 34.0 & 17.7 & 17652 & ae & 189.6 & y + 416 & & 03 & 30 & 41.11 & -52 & 24 & 46.3 & 17.9 & 18689 & a & 36.6 & n + 422 & & 03 & 30 & 46.35 & -53 & 40 & 16.8 & 17.1 & 18152 & a & 127.2 & n + 430 & e120 & 03 & 30 & 50.26 & -52 & 14 & 29.9 & 18.2 & 20887 & a & 48.4 & n + 431 & & 03 & 30 & 50.47 & -52 & 18 & 56.5 & 18.4 & 18542 & ae & 72.7 & y + 432 & e124 & 03 & 30 & 50.67 & -52 & 58 & 31.0 & 17.9 & 17968 & a & 64.4 & n + 434 & & 03 & 30 & 51.28 & -52 & 31 & 15.1 & 17.1 & -62 & a & 83.6 & n + 438 & & 03 & 30 & 54.01 & -53 & 40 & 36.7 & 17.8 & 15921 & a & 336.7 & n + 440 & & 03 & 30 & 54.88 & -53 & 37 & 31.0 & 17.9 & 44269 & e & 76.0 & y + 442 & & 03 & 30 & 55.49 & -53 & 35 & 09.7 & 18.4 & 27840 & a & 114.3 & n + 443 & & 03 & 30 & 55.62 & -51 & 51 & 50.4 & 18.5 & 18134 & e & 265.6 & y + 445 & e131 & 03 & 30 & 56.30 & -53 & 07 & 34.5 & 16.8 & 18538 & ae & 374.9 & y + 447 & & 03 & 30 & 56.49 & -53 & 45 & 49.8 & 16.8 & 15704 & a & 159.0 & n + 448 & e130 & 03 & 30 & 56.52 & -52 & 51 & 20.0 & 16.8 & 18710 & ae & 48.8 & y + 449 & e128 & 03 & 30 & 57.02 & -52 & 24 & 32.7 & 18.4 & 18221 & a & 99.9 & n + 451 & & 03 & 30 & 59.28 & -53 & 41 & 52.8 & 17.0 & 18147 & ae & 43.0 & y + 456 & & 03 & 31 & 01.65 & -52 & 40 & 50.4 & 18.5 & 19036 & e & 130.2 & y + 458 & e138 & 03 & 31 & 02.14 & -53 & 06 & 06.8 & 18.0 & 18378 & ae & 111.4 & y + 464 & & 03 & 31 & 06.67 & -52 & 30 & 39.0 & 17.8 & 17503 & a & 43.1 & n + 467 & e143,cr144a & 03 & 31 & 08.18 & -52 & 25 & 03.2 & 17.4 & 18502 & a & 41.7 & n + 470 & & 03 & 31 & 08.89 & -53 & 19 & 12.3 & 18.0 & 22882 & e & 168.1 & y + 472 & & 03 & 31 & 10.47 & -52 & 55 & 41.2 & 16.9 & 21488 & ae & 81.5 & y + 473 & e146 & 03 & 31 & 11.06 & -52 & 10 & 33.8 & 17.8 & 18385 & a & 41.3 & n + 475 & e147 & 03 & 31 & 11.81 & -52 & 26 & 13.8 & 17.8 & 65 & a & 41.9 & n + 477 & e148,cr80a & 03 & 31 & 12.26 & -52 & 35 & 56.8 & 15.6 & 16946 & ae & 399.1 & y + 478 & & 03 & 31 & 12.99 & -52 & 13 & 14.0 & 18.4 & 21332 & a & 64.9 & n + 479 & & 03 & 31 & 13.72 & -52 & 04 & 40.2 & 16.4 & 9421 & e & 113.6 & y + 481 & e151,cr53a & 03 & 31 & 14.96 & -52 & 41 & 48.1 & 16.0 & 19936 & ae & 63.7 & y + 482 & e150,cr156a & 03 & 31 & 15.36 & -52 & 22 & 27.4 & 16.4 & 18070 & a & 209.0 & n + 483 & cr105a & 03 & 31 & 15.86 & -52 & 30 & 09.9 & 18.1 & 17656 & a & 24.2 & n + 484 & e152 & 03 & 31 & 18.27 & -52 & 38 & 17.5 & 18.0 & 24327 & e & 72.9 & y + 485 & & 03 & 31 & 18.77 & -52 & 12 & 13.1 & 18.4 & 49771 & ae & 192.4 & y + 486 & & 03 & 31 & 19.97 & -53 & 30 & 30.6 & 18.3 & 18038 & a & 90.9 & n + 487 & e153 & 03 & 31 & 21.45 & -53 & 01 & 43.6 & 18.0 & 23185 & a & 68.0 & n + 489 & & 03 & 31 & 23.49 & -53 & 27 & 50.6 & 18.5 & 23247 & a & 65.9 & n + 490 & & 03 & 31 & 23.73 & -53 & 05 & 13.5 & 18.4 & 19651 & e & 268.5 & y + 491 & & 03 & 31 & 23.85 & -53 & 14 & 36.4 & 17.4 & 22978 & ae & 210.6 & y + 492 & e157 & 03 & 31 & 24.80 & -52 & 50 & 30.7 & 18.1 & 18963 & a & 122.5 & n + 493 & e156,cr126a & 03 & 31 & 25.44 & -52 & 28 & 10.0 & 15.9 & 16755 & a & 213.0 & n + 494 & & 03 & 31 & 25.44 & -52 & 36 & 16.1 & 17.9 & 16757 & a & 31.5 & n + 497 & & 03 & 31 & 27.26 & -53 & 01 & 45.7 & 17.7 & 22563 & e & 85.2 & y + 498 & & 03 & 31 & 27.33 & -52 & 30 & 05.3 & 18.0 & 17490 & a & 43.8 & n + 501 & & 03 & 31 & 30.75 & -52 & 03 & 42.6 & 18.2 & 54493 & e & 122.9 & y + 502 & e159 & 03 & 31 & 31.16 & -52 & 20 & 44.1 & 16.3 & 18499 & ae & 41.5 & y + 509 & e166 & 03 & 31 & 37.77 & -53 & 00 & 17.8 & 17.0 & 23042 & ae & 74.1 & y + 510 & & 03 & 31 & 38.99 & -52 & 06 & 47.3 & 17.6 & 24991 & e & 61.7 & y + 512 & & 03 & 31 & 40.42 & -52 & 22 & 23.5 & 18.3 & 17718 & a & 45.9 & n + 513 & & 03 & 31 & 41.23 & -53 & 07 & 58.8 & 18.3 & 18403 & ae & 93.2 & y + 514 & e168 & 03 & 31 & 42.30 & -52 & 25 & 11.9 & 16.9 & 19946 & a & 37.0 & n + 516 & & 03 & 31 & 47.75 & -52 & 09 & 25.7 & 18.4 & 57996 & a & 104.0 & n + 517 & & 03 & 31 & 50.81 & -52 & 01 & 26.1 & 18.1 & 22997 & ae & 65.4 & y + 519 & cr161a & 03 & 31 & 52.56 & -52 & 18 & 26.0 & 17.3 & 18140 & a & 40.4 & n + 520 & & 03 & 31 & 52.95 & -52 & 07 & 24.1 & 16.0 & 11765 & ae & 249.8 & y + 522 & & 03 & 31 & 53.72 & -52 & 08 & 08.9 & 18.5 & 18624 & e & 105.0 & y + 527 & & 03 & 31 & 57.79 & -52 & 41 & 13.0 & 18.1 & 16926 & ae & 86.5 & y + 528 & e173 & 03 & 31 & 58.02 & -52 & 58 & 16.5 & 17.9 & 50096 & a & 599.6 & n + 530 & & 03 & 32 & 04.30 & -53 & 13 & 14.1 & 17.4 & 17908 & e & 70.2 & y + 532 & e174 & 03 & 32 & 07.51 & -52 & 15 & 39.0 & 16.7 & 18257 & e & 76.2 & y + 533 & e175 & 03 & 32 & 07.80 & -52 & 33 & 29.8 & 17.5 & 11849 & e & 270.8 & y + 534 & & 03 & 32 & 08.66 & -53 & 05 & 27.9 & 18.3 & 13705 & e & 62.1 & y + 535 & & 03 & 32 & 09.70 & -53 & 29 & 58.1 & 17.9 & 32652 & a & 111.4 & n + 537 & e181 & 03 & 32 & 12.52 & -52 & 56 & 42.4 & 17.0 & 18126 & e & 118.2 & y + 538 & e178 & 03 & 32 & 12.57 & -52 & 43 & 49.7 & 17.2 & 17320 & a & 62.2 & n + 540 & e182 & 03 & 32 & 14.40 & -52 & 35 & 51.1 & 16.8 & 16926 & a & 52.9 & y + 541 & & 03 & 32 & 15.58 & -51 & 56 & 08.2 & 18.4 & -16 & a & 47.3 & n + 544 & & 03 & 32 & 21.03 & -52 & 19 & 23.1 & 18.1 & 17987 & ae & 70.3 & y + 547 & e185 & 03 & 32 & 26.74 & -52 & 22 & 53.6 & 16.7 & 16760 & a & 46.3 & y + 548 & & 03 & 32 & 28.02 & -52 & 13 & 54.9 & 17.4 & 12479 & a & 103.9 & y + 549 & & 03 & 32 & 28.24 & -53 & 09 & 20.2 & 17.8 & 33999 & e & 198.7 & y + 550 & & 03 & 32 & 28.70 & -52 & 44 & 30.1 & 18.2 & 18125 & a & 29.1 & n + 551 & & 03 & 32 & 29.48 & -52 & 30 & 18.4 & 17.4 & 16744 & ae & 27.5 & y + 552 & & 03 & 32 & 29.56 & -53 & 30 & 07.4 & 17.5 & 33018 & ae & 152.4 & y + 553 & & 03 & 32 & 30.40 & -53 & 24 & 46.7 & 18.3 & 23307 & a & 69.3 & y + 555 & & 03 & 32 & 30.70 & -52 & 01 & 20.0 & 17.2 & 18379 & e & 249.4 & y + 558 & & 03 & 32 & 33.55 & -52 & 32 & 24.6 & 18.0 & 18692 & ae & 102.3 & y + 559 & e188,cr153a & 03 & 32 & 34.66 & -52 & 21 & 16.9 & 17.2 & 18280 & a & 153.7 & n + 560 & & 03 & 32 & 39.73 & -52 & 59 & 44.7 & 18.2 & 27763 & a & 94.4 & n + 563 & & 03 & 32 & 42.41 & -52 & 17 & 55.4 & 18.5 & 18664 & e & 110.9 & y + 565 & & 03 & 32 & 45.32 & -52 & 07 & 38.5 & 17.2 & 18395 & ae & 117.4 & y + 566 & & 03 & 32 & 45.40 & -52 & 10 & 57.7 & 18.1 & 20243 & e & 117.8 & y + 567 & & 03 & 32 & 45.92 & -52 & 33 & 31.0 & 16.7 & -9 & a & 17.9 & n + 569 & & 03 & 32 & 47.45 & -53 & 25 & 50.7 & 16.1 & 22775 & a & 81.5 & n + 570 & & 03 & 32 & 47.49 & -53 & 28 & 24.6 & 17.9 & 23822 & ae & 145.9 & y + 572 & & 03 & 32 & 48.41 & -52 & 51 & 50.3 & 18.4 & 33 & a & 18.1 & n + 574 & & 03 & 32 & 50.67 & -53 & 27 & 37.3 & 17.6 & 22548 & a & 59.5 & n + 576 & & 03 & 32 & 55.39 & -52 & 47 & 56.8 & 17.2 & 12752 & ae & 150.8 & y + 578 & & 03 & 32 & 57.35 & -52 & 16 & 25.2 & 18.0 & -33 & a & 278.2 & n + 579 & & 03 & 32 & 57.65 & -52 & 54 & 05.2 & 18.2 & 27779 & ae & 152.6 & y + 581 & & 03 & 32 & 58.46 & -52 & 15 & 16.2 & 18.5 & 38932 & ae & 110.6 & y + 582 & & 03 & 32 & 59.93 & -53 & 03 & 31.7 & 16.1 & 11796 & ae & 131.8 & y + 583 & & 03 & 33 & 10.22 & -53 & 29 & 27.9 & 17.7 & 22533 & a & 221.6 & n + 587 & & 03 & 33 & 17.61 & -52 & 12 & 41.2 & 17.2 & -49 & a & 285.7 & n + 588 & e191 & 03 & 33 & 18.61 & -52 & 27 & 59.7 & 17.2 & 17676 & a & 31.7 & n + 589 & & 03 & 33 & 18.94 & -53 & 28 & 44.9 & 17.6 & 22761 & a & 46.6 & n + 590 & & 03 & 33 & 19.22 & -53 & 31 & 36.4 & 17.8 & 23651 & a & 63.6 & n + 591 & e192 & 03 & 33 & 20.12 & -52 & 28 & 43.6 & 17.5 & 17569 & ae & 38.6 & y + 592 & & 03 & 33 & 20.40 & -52 & 01 & 59.2 & 17.5 & 25216 & ae & 60.5 & y + 593 & & 03 & 33 & 21.83 & -52 & 33 & 09.2 & 16.2 & 13405 & ae & 412.7 & y + 595 & e193 & 03 & 33 & 24.92 & -52 & 30 & 24.9 & 17.5 & 17572 & ae & 98.7 & y + 596 & & 03 & 33 & 25.94 & -52 & 50 & 12.7 & 18.4 & 23014 & a & 67.5 & n + 597 & & 03 & 33 & 28.27 & -53 & 13 & 17.5 & 18.2 & 17446 & e & 248.3 & y + 598 & & 03 & 33 & 30.06 & -53 & 10 & 23.9 & 16.2 & 17052 & ae & 273.0 & y + 599 & & 03 & 33 & 34.79 & -53 & 12 & 24.5 & 18.3 & 18453 & e & 102.9 & y + 602 & & 03 & 33 & 37.63 & -52 & 16 & 25.0 & 18.5 & 53569 & e & 161.5 & y + 603 & & 03 & 33 & 38.88 & -52 & 12 & 02.7 & 16.6 & 20306 & a & 158.7 & n + 604 & & 03 & 33 & 45.13 & -52 & 19 & 31.1 & 18.4 & 20333 & ae & 117.7 & y + 605 & & 03 & 33 & 46.58 & -52 & 26 & 42.6 & 17.6 & 31943 & ae & 101.4 & y + 606 & & 03 & 33 & 49.97 & -53 & 01 & 34.4 & 18.3 & 17703 & e & 185.4 & y + 607 & & 03 & 33 & 52.74 & -52 & 14 & 00.4 & 17.8 & 19721 & e & 354.1 & y + 608 & & 03 & 33 & 53.05 & -52 & 11 & 07.5 & 17.2 & 19983 & ae & 70.2 & y + 611 & & 03 & 33 & 59.66 & -52 & 10 & 10.0 & 17.4 & 44705 & a & 83.6 & n + 612 & & 03 & 34 & 03.19 & -52 & 10 & 58.2 & 17.6 & 20734 & a & 91.6 & y + 614 & & 03 & 34 & 11.99 & -53 & 23 & 42.1 & 17.9 & 18067 & a & 68.2 & n + 615 & & 03 & 34 & 15.31 & -52 & 43 & 38.9 & 18.4 & 31348 & ae & 58.8 & y + 616 & & 03 & 34 & 18.43 & -53 & 07 & 32.6 & 17.7 & 20429 & a & 57.6 & y + 617 & & 03 & 34 & 18.65 & -53 & 11 & 44.9 & 18.3 & 16953 & e & 46.5 & y + 618 & & 03 & 34 & 21.89 & -53 & 02 & 49.8 & 17.9 & 21710 & a & 54.0 & n + 619 & & 03 & 34 & 24.84 & -52 & 25 & 48.3 & 18.2 & 19409 & e & 85.0 & y + 620 & & 03 & 34 & 28.25 & -52 & 33 & 41.7 & 17.1 & 13530 & e & 223.9 & y + 622 & & 03 & 34 & 29.79 & -53 & 00 & 16.1 & 18.4 & 44485 & e & 184.7 & y + 623 & & 03 & 34 & 33.19 & -52 & 41 & 00.8 & 17.6 & 11553 & ae & 63.9 & y + 625 & & 03 & 34 & 37.17 & -53 & 26 & 52.4 & 17.8 & 18479 & a & 78.6 & n + 626 & & 03 & 34 & 39.77 & -52 & 29 & 40.9 & 17.8 & 22790 & a & 76.0 & n + 627 & & 03 & 34 & 43.14 & -52 & 50 & 00.5 & 18.4 & 44417 & e & 241.1 & y + 628 & & 03 & 34 & 47.64 & -52 & 58 & 30.5 & 17.5 & 20339 & a & 78.7 & y + 629 & & 03 & 34 & 49.29 & -53 & 14 & 13.7 & 16.7 & 9245 & ae & 298.8 & y + 635 & & 03 & 35 & 07.67 & -53 & 00 & 54.1 & 16.1 & 17371 & a & 91.0 & y + 636 & & 03 & 35 & 11.31 & -52 & 50 & 59.5 & 18.1 & 17231 & ae & 129.6 & y + 637 & & 03 & 35 & 17.70 & -52 & 59 & 25.2 & 17.5 & 20927 & ae & 47.6 & y + 638 & & 03 & 35 & 26.80 & -53 & 08 & 51.0 & 16.4 & 16939 & ae & 80.5 & y + 640 & & 03 & 35 & 34.68 & -52 & 53 & 23.1 & 16.6 & 22990 & a & 66.5 & n + [ velocities_table ] g1 & 25 & 18600 19400 & -18.0 10.0 & & 4.4 x 10@xmath92 & 0.54 & 0.40 & 6.6 & 9.7 + ( control ) & 24 & 18600 19400 & & & & 0.91 & 0.92 & 17.1 & 31.7 + g2 & 14 & 18200 19100 & & -80 . -50 . & 7.9 x 10@xmath93 & 0.78 & 0.14 & 9.5 & 5.5 + ( control ) & 83 & 18200 19100 & & & & 0.41 & 0.97 & 20.1 & 17.1 + f1 & 19 & 19400 20500 & & -10.0 22.0 & 0.20 & 0.018 & 1.13 x 10@xmath92 & 26.0 & 9.5 + ( control ) & 14 & 19400 20500 & & & & 1.69 x 10@xmath92 & 0.59 & 27.7 & 12.9 + f2 & 10 & 19100 20000 & & -53.0
-22.0 & 0.097 & 0.19 & 1.75 x 10@xmath92 & 20.7 & 6.9 + ( control ) & 25 & 19100 20000 & & & & 0.024 & 0.13 & 22.2 & 14.4 + + c1 & 24 & 17500 18300 & -30.0 -22.0 & & 0.025 & 0.94 & 0.91 & 21.5 & 1.9 + ( control ) & 118 & 17500 18300 & & & & 1.00 & 1.00 & 24.2 & 28.4 + [ groups_table ] g3 & 4 & 19700 20100 & & -18.0 -7.0 & 4.0 & 2.2 + ( control ) & 9 & 19700 20100 & & & 24.8 & 26.5 + g4 & 5 & 21500 22400 & & -20.0 5.0 & 7.1 & 6.3 + ( control ) & 4 & 21500 22400 & & & 27.5 & 22.0 + g5 & 5 & 20300 21800 & -2.0 7.0 & & 1.5 & 18.7 + & 4 & 20300 21800 & -2.0 7.0 & & 1.6 & 1.9 + ( control ) & 12 & 20300 21800 & & & 27.0 & 32.5 + g6 & 5 & 15000 16300 & & -80 . -60 . &
11.0 & 2.9 + ( control ) & 4 & 15000 16300 & & & 7.0 & 4.4 + g7 & 8 & 22000 24100 & 12.0 30.0 & & 3.0 & 13.5 + & 7 & 22000 24100 & 12.0 30.0 & & 2.8 & 2.3 + ( control ) & 41 & 22000 24100 & & & 27.4 & 21.0 + [ groups3_table ] ne ( global ) & 0 - 180 & 3.9 @xmath94 & 0.13 @xmath95 & 1.08 + ne ( small core ) & 0 - 22.5 & 5.1 @xmath96 & 0.93 @xmath97 & 0.99 + ne ( large core ) & 0 - 45 & 5.2 @xmath98 & 0.18 @xmath99 & 0.72 + ne ( outer region ) & 45 - 90 & 3.9 @xmath100 & 0.17 @xmath101 & 1.15 + ne ( outer region ) & 90 - 180 & 3.6 @xmath102 & 0.13 @xmath103 & 1.29 + + sw ( large core ) & 0 - 90 & 3.3 @xmath104 & 0.61 @xmath105 & 1.29 + sw ( outer region ) & 90 - 180 & 3.7 @xmath106 & 0.34 @xmath107 & 0.99 + [ temps_table ]
|
multi - fiber spectroscopy has been obtained for 335 galaxies in the field of the double cluster a3128/a3125 , using the 2df multi - fiber positioner on the anglo - australian telescope . when combined with previously published results , a total of 532 objects in the double cluster now have known redshifts .
we have also obtained a 20 ks _
chandra _
acis - i image of the central 16x 16of a3128 and radio imaging of the cluster with the molonglo observatory synthesis telescope and the australia telescope compact array .
the spatial - kinematic distribution of redshifts in the field of a3128/a3125 , when combined with the _ chandra _ acis - i image of a3128 , reveals a variety of substructures present in the galaxy distribution and in the hot intracluster medium ( icm ) . the most striking large - scale feature in the galaxy distribution is a relatively underpopulated redshift zone @xmath04000 on either side of the mean cluster velocity at @xmath017500 .
we attribute this depletion zone to the effect of the extensive horologium - reticulum ( h - r ) supercluster , within which a3128/a3125 is embedded .
in addition to this large - scale feature , numerous smaller groups of galaxies can be identified , particularly within the underpopulated region within @xmath14000 of the mean cluster redshift . due to the large gravitational influence of the h - r supercluster ,
these groups arrive at a3128 with a high infall velocity , well in excess of the local sound speed .
two of these groups appear as elongated filaments in position - velocity diagrams , indicating that they are tidally distended groups which have been disrupted after a close passage through a3128 .
in fact , a3125 itself appears to be in such a post passage condition .
we have identified a primary ne - sw merger axis connecting a3128 with a3125 , along which the filaments are also oriented .
in addition , the _ chandra _ image reveals that the x - ray emission is split into two components , each with very small core radii , that are separated by @xmath01 mpc along the ne - sw axis . we have combined the redshift , x - ray , and radio data to propose that the complex x - ray morphology revealed in the _ chandra _ image is likely the result of a hypersonic infall of a relatively small group into a3128 .
the group produces a major disruption in the icm due to its high infall velocity .
|
public opinion data can be used to determine public awareness , to predict outcomes of events , and to infer characteristics of human behaviors .
indeed , readily available public opinion data is valuable to researchers , policymakers , marketers , and many other groups , but is difficult to generate .
solicited polls can be expensive , prohibitively time consuming , and may only reach a limited number of people on a limited number of days .
polling accuracy evidently relies on accessing representative populations and high response rates .
poor temporal sampling will weaken any poll s value as individual opinions vary in time and in response to social influence@xcite . with the continued rise of social media as a communication platform
, the ability to construct unsolicited public opinion polls has become a possibility for researchers though parsing of massive text - based datasets .
social media provides extraordinary access to public expressions in real time , and has been shown to play a role in human behavior @xcite . with its open platform , twitter has proved to be a boon for many research enterprises @xcite , having been used to explore a variety of social and linguistic phenomena @xcite ; harnessed as a data source to create an earthquake reporting system in japan @xcite ; made possible detection of influenza outbreaks @xcite ; and used to analyze overall public health @xcite .
predictions made using twitter have focused on elections @xcite , the spread of disease @xcite , crime @xcite , and the stock market @xcite .
these studies demonstrate a proof - of - concept , avoiding the more difficult task of building operational systems for continued forecasting
. we must be clear that for all its promise , prediction via social media is difficult .
indeed , we have seen a number of high profile failures such as google flu trends @xcite and various attempts to predict election outcomes @xcite .
opinionfinder was used in @xcite to evaluate tweets containing direct expressions of emotion as in ` i feel ' , or ` i am feeling ' , and shown not to have predictive power for the stock market despite limitations which we address later in sec .
iv , twitter data reveals an unprecedented view of human behavior and opinion related to major issues of global importance @xcite . in a previous study @xcite
, we analyzed the sentiment surrounding climate change conversation on twitter .
we discovered that sentiment varies in response to climate change news and events , and that the conversation is dominated by activists .
another study by helmuth et al .
analyzed tweets by united states senators to determine which research oriented science organizations and which senators are best at getting science - related findings into the hands of the general public @xcite .
twitter is also often used to analyze public opinion of political issues @xcite , and in several previous works as an opinion polling resource . in an application using neural networks called truehappiness ,
users enter one of 300,000 words to obtain a sentiment estimation based on this word s usage in a massive wikipedia data set , and on previously collected sentiment scores for 10,222 words on amazon s mechanical turk @xcite , hereafter referred to as the labmt word set in another application called raccoon , users are invited to enter a query term and a rough sketch to obtain words or phrases on twitter that correlate well with the inputs @xcite .
google correlate is a similar tool that discovers google searches for terms or phrases that match well with real - world time series @xcite .
financial term searches from google trends was shown by preis et al .
@xcite to correlate with dow jones economic indices .
we argue that twitter is a better source for opinion mining than wikipedia , used in truehappiness , due to the personal nature of each post .
raccoon uses only user estimates for correlations and not actual survey data .
google correlate uses only frequencies of google searches to compare time series and may only be useful in specific situations . in many studies that use text data from twitter ,
the results are not compared to actual polling data , leaving the conclusions open to interpretation .
one example work which does make a direct comparison is @xcite , where the authors use a twitter data set from 2008 and 2009 to compare sentiments on twitter , calculated with opinionfinder , with daily and monthly public opinion polls .
our approach here is analogous to that of @xcite , but we use the sentiment analysis techniques developed in @xcite to investigate public opinion regarding over 10,000 search terms . specifically , for each of 10,222 of the most frequently used english words , we calculate daily , weekly , monthly , and yearly sentiment time series of co - occurring words from september 2008 to november 2015 .
we compare many of these happiness time series to actual polling data , which is not typically available at such a high resolution in time .
we investigate a wide range of topics , including politics , the economy , and several commercial organizations .
given the size of the dataset , we are unable to exhaustively compare all search terms to solicited opinion polls , and have released the data publicly along with this paper ( see http://compstorylab.org/share/papers/cody2016a/index.html ) .
overall , out aim is to determine the extent to which twitter can be used to complement traditional public opinion surveys , ideally as a dashboard indicator accompanied by solicited feedback .
we implement the hedonometer " , an instrument designed to calculate a happiness score for a large - scale text , based on the happiness of individual words in the text . the hedonometer uses previously assessed happiness scores for the labmt word set , which contains the most frequently used english words in four disparate corpora @xcite .
we choose the hedonometer to obtain lexical coverage of twitter text and produce meaningful word shift graphs @xcite .
the words were scored in isolation by human subjects in previous work on a scale from 1 ( least happy ) to 9 ( most happy ) .
we remove neutral and ambiguous words ( scores between 4 and 6 ) from the analysis . for details regarding stop words ,
see dodds et al .
@xcite .
we use the hedonometer to calculate what we refer to as _ ambient happiness _
scores for each of the labmt words ( first defined in @xcite ) .
we determine ambient happiness of a given word , @xmath0 , by calculating the average happiness of the words that appear in tweets with that word , i.e. , @xmath1 here , @xmath2 is a word that appears in a tweet with word @xmath0 , @xmath3 is the surveyed happiness score of word @xmath4 , @xmath5 is the frequency of word @xmath4 , and @xmath6 is the number of words in labmt ( with stop words removed ) appearing in tweets containing word @xmath7 .
note that we do not include the frequencies or happiness scores of the given word ( @xmath0 ) in the calculation of ambient happiness .
for example , @xmath8 for love " and the ambient happiness for tweets containing love " , averaged using eqn [ amb ] over the 7 year period , is 6.17 . for hate " , @xmath9 and we find the ambient happiness of hate " is 5.75 . as seen in the appendix in fig .
[ scatter ] , we find that due to averaging , ambient happiness covers a smaller range of scores than labmt happiness for individual words .
we use the ambient happiness scores to create time series for each of the words in the labmt word set , and we correlate the happiness time series with polling data at various temporal resolutions .
we collected tweets from twitter s gardenhose api from september 2008 to november 2015 . during this time period ,
the volume of tweets grew by three orders of magnitude , but the random percentage of all public tweets fell from 50% to 10% .
for each word in the labmt dictionary , e.g. obama " , we subsample the gardenhose for all tweets matching the word .
we then tabulate the daily frequencies of labmt words in this term - defined collection of tweets , resulting in temporal counts of the words co - occuring with obama " .
for example , the resulting collection of counts for obama " is a 2,628 ( days ) by 10,222 ( words ) matrix with entry @xmath10 representing the frequency of labmt word @xmath7 appearing in a tweet containing the term obama " on day @xmath4 .
this collection of counts is posted on the http://compstorylab.org/share/papers/cody2016a/index.html[online appendix ] for this paper .
[ obamaday ] gives the average daily ambient happiness of obama " , along with the average daily happiness of all tweets during the same time period .
along with a general slow decline , we see spikes in happiness each year on august 4th , the president s birthday , with the largest spike occurring on october 9 , 2009 when president obama was awarded the nobel peace prize .
we see a strong dip shortly after on october 26 , 2009 when president obama declares a state of emergency for the h1n1 virus .
we see spikes in relative frequency of obama " on both election days in 2008 and in 2012 . to compare our findings with solicited opinions , we collected yearly and quarterly polling data from gallup @xcite .
we focus on quarterly data , as it is the highest resolution we were able to obtain .
the yearly analysis provides us with only 7 data points , and results are in the appendix .
we compare president obama s job approval rating on gallup and on pollster @xcite , which allows for daily data collection through their api .
finally , we use the university of michigan s index of consumer sentiment data , which is collected monthly @xcite .
here we present happiness time series for several words for which we find interesting patterns .
figure [ amb_figs ] gives examples of ambient happiness and relative frequency time series for a few selected words .
happiness associated with certain religious words , e.g. church " and muslim " has decreased in recent years , with dips corresponding to several mass shootings including the charleston shooting in june 2015 and the chapel hill shooting in february 2015 .
the relative frequency of church " peaks each year on easter sunday .
we see that ambient happiness of `` snow '' is seasonal , with the highest happiness during the northern hemisphere summer and lowest during the winter , while the relative frequency is highest during the winter and lowest during the summer .
the saddest day for `` snow '' was june 14 , 2015 when a popular game of thrones character is presumed dead .
the ambient happiness scores of democrat " and republican " are on a slow decline , with relative frequencies peaking during presidential and midterm elections .
president obama s press conference after the sandy hook shooting is the saddest day for `` democrat '' , while the saddest day for `` republican '' coincides with an incident involving the egyptian republican guard .
ambient happiness of love " peaks around the holidays each year , and the relative frequency was increasing until recently . while ambient happiness of `` love '' peaks on christmas each year , the relative frequency of love " peaks on valentine s day each year , which could be due to the difference in labmt scores for `` christmas '' and `` valentines '' ( 7.96 and 7.30 respectively ) . in traditional polls
, there may be large differences in public opinion from one time period to the next . in a yes / no or multiple choice survey question it is impossible to use that data to determine why differences occur .
here we use word shift graphs to determine the cause of a shift in ambient happiness . a word shift graph ranks words by their contributing factor to the change in happiness between two pieces of text .
for example , in fig .
[ snowshift ] we investigate why the ambient happiness of snow " is higher in the northern hemisphere summer ( when its relative frequency is lowest ) and lower in the winter ( when its relative frequency is highest ) . the word shift graph in fig .
[ snowshift ] compares snow " tweets during the winter months ( december , january , february ) to snow " tweets in the summer months ( june , july , august ) .
english speaking countries like australia and new zealand will necessarily be included in the wrong season , however their contribution is small .
we find that twitter users loathe the snow during the winter , and miss the snow during the summer , as indicated by the increase in the word hate " , negatives , and profanity during the winter months and the decrease in the word love " .
the influence of the disney classic snow white " is also visible , appearing to be referenced more often in summer months due to its motion picture release on june 1 , 2012 .
we next investigate the relationship between president obama s job approval rating from two public opinion polling resources and the ambient happiness of obama " tweets .
president obama s quarterly job approval rating is freely available on gallup.com @xcite , and president obama s daily job approval rating is freely available on pollster.com @xcite .
we correlate the average quarterly happiness of tweets containing the word obama " with president obama s quarterly job approval rating and find a strong positive correlation ( see appendix fig .
[ obamaquartera ] ) .
however , we find the correlation is much stronger in fig .
[ obamaquarter ] , which gives the happiness time series at a one quarter lag .
similarly , we find a strong positive correlation between the daily approval rating available on pollster and the daily ambient happiness of obama " ( see appendix fig .
[ obamadailya]a ) with an improvement in the correlation when the tweets are lagged by 30 days in appendix fig .
[ obamadailya]b .
this indicates that real time twitter data has the potential to predict solicited public opinion polls .
figure [ obamaquarter ] shows that president obama s highest approval rating in all three sources was during his first quarter ( january march , 2009 ) .
his lowest approval rating was during his 23rd quarter ( july september , 2014 ) .
[ obamashift ] shows which words contributed most to this shift in ambient happiness .
tweets containing the word obama " discuss war and terrorism more often during his 23rd quarter than his first quarter .
next , we investigate a monthly poll on consumer sentiment designed by the university of michigan @xcite .
this poll asks participants five questions about their current and future financial well being and calculates an index of consumer sentiment ( ics ) based on responses . in fig .
[ ics ] we correlate this monthly time series with the ambient happiness of the word job " .
we find that the correlation is much stronger starting in 2011 ( fig .
[ ics]b ) , and slightly stronger still when the ambient happiness is lagged one month ( fig . [ ics]c ) . in fig .
[ ics]d we correlate the ics with the relative frequency of the word job " on twitter .
we find a strong negative correlation , indicating that it is more likely that a user will tweet the word job " when they are searching for one .
in this section , we investigate the changes in twitter sentiment surrounding two businesses , walmart and mcdonalds .
we examine the ambient happiness time series to determine how sentiment changes in response to events that took place at specific stores .
[ business ] gives the ambient happiness and relative frequency of the words
walmart " and mcdonalds " .
for the paper .
] many of the spikes in the walmart " ambient happiness time series correspond to free giveaways to which twitter users are responding .
a dip in november 2008 corresponds to the trampling to death of a walmart employee on black friday ( the day after thanksgiving , notorious in the u.s . for shopping ) .
shootings that took place in walmart stores in 2014 are shown with orange dots in fig .
[ business]a . in june 2014
the jerad and amanda miller las vegas shootings ended with 5 deaths ( including themselves ) in a nevada walmart . in september 2014
, the police officer who shot john crawford in an ohio walmart was indicted . in december 2014 , a 2 year old accidentally shot and killed his mother in an idaho walmart . we also see a dip in happiness on the day tracy morgan sues walmart over a fatal crash with one of their tractor trailers in july 2014 . the happiest day in the
mcdonalds " ambient happiness time series is valentine s day in 2015 . upon reading some example tweets from this day , we find that mcdonalds was a popular ironic destination for valentine s day dinner that year among twitter users .
a second spike corresponds to a prestigious award given to the mcdonalds enterprise in february 2013 .
mcdonalds was given the
top toilet award " for the cleanliness of its restrooms .
the saddest day for mcdonalds on twitter was august 18 , 2014 , the day that ferguson protesters broke into a mcdonalds to steal milk to relieve tear gas victims . in fig .
[ bus_by_month ] we explore the monthly ambient happiness of walmart " and mcdonalds " .
we find that the ambient happiness of walmart " reaches its maximum in march 2011 , and its minimum in october 2015 , and the ambient happiness of mcdonalds " reaches its maximum in february 2015 and its minimum shortly after in may 2015 . to investigate the texture behind these observations , we use word shift graphs to compare the happiest and saddest months for each business in fig .
[ bus_shifts ] . in november 2015 ( comparison text fig .
[ bus_shifts]a ) , there were black friday altercations at many walmarts throughout the country , often caught on camera , leading to an increase in negative words such as caught " , fail " , ridiculous " , and captured " . twitter users were happier about walmart in march 2011 ( reference text fig .
[ bus_shifts]a ) due in part to a free gift card giveaway .
happier tweets included the words lol " , love " , haha " , and super " . surprisingly , we actually see more curse words during the happiest month than the saddest month . the happiest month for mcdonalds was february 2015 ( reference text fig .
[ bus_shifts]b ) when a surprising number of twitter users were spending valentine s day there , hence the decrease in the words valentines " and love " .
the decrease in happiness in may 2015 is in large part in an increase in the word disappear " . during this time
, a video of a michigan mcdonalds employee performing a practical joke , in which he claims he s going to make a penny disappear in a bottle of water , went viral . using word shifts , we are thus able to determine `` disappear '' was not a true indicator for negativity . the next step would be to add `` disappear '' to our stop word list and reevaluate the time series .
in general , word shifts are very sensitive diagnostics that allow us to make sense of apparent sentiment patterns , and to adjust the hedonometer as needed .
here we present positive correlations between ambient happiness on twitter and solicited public opinion surveys .
the methods used in this work present several limitations that we address here .
first , we are unable to compare ambient happiness to public opinion surveys for the majority of subjects discussed on twitter .
ambient happiness surrounding commercial businesses can be compared to stock prices , however the relationship is unlikely to be linear .
stock prices take into account much more than the success of one business , and thus reflect a different signal than public opinion .
we acknowledge that our dataset can be influenced by small events that are retweeted many times , and cause a transient spike or a dip in ambient happiness .
tweets reflect an broad spectrum of conversation and reactions to major events , including bots and sarcasm to name a few difficulties .
the stochastic nature of twitter timeseries data also makes it difficult to compare high resolution ambient happiness to low resolution opinion surveys .
appendix fig .
[ good ] shows our attempt to compare ambient happiness to six important topics available on gallup .
gallup surveys take place over several days , once a year , leading to a natural comparison with ambient happiness averaged for a full 365 days .
we are therefore unable to report which method is a more accurate indicator of public opinion . instead , we conclude that public opinion polling with twitter has the potential to complement traditional public opinion surveys . finally , without knowing user demographics in detail , we are unable to adjust for the known sample bias of twitter users relative to the general population @xcite .
our conclusions pertain directly only to the twitterverse and as suggestive results for the broader public .
twitter users are a mixture of individuals , news outlets , corporations , and , perhaps most problematically , are not even always human .
previous work shows that twitter contains many bots , which send tweets automatically , often to advertise a product or a political campaign @xcite .
we do not eliminate these tweets in this work , however many methods for uncovering them have been suggested @xcite .
the objective of this research was to determine the extent to which ambient happiness on twitter can be used as a reliable public opinion polling device .
a central motivation is that solicited public opinion polling data is difficult to obtain at a high temporal resolution , if at all @xcite .
with data from twitter we can investigate topics other than political or global issues , which are the focus of a large majority of solicited surveys .
we can use ambient happiness to determine how people feel about seemly neutral topics like snow " , or how they are using the words love " , and feel " .
we also have shown that twitter users respond to various kinds of events taking place at commercial businesses , and thus ambient happiness could be used in market analysis to predict or improve a company s sales . of the available polling data we were able to obtain , we find that ambient happiness of selected words correlates well with solicited public opinions . in some cases ,
the correlation increases when the tweets are lagged , indicating that real time twitter data has the potential to predict solicited public opinion polls .
not only can tweets anticipate survey responses , but we can use individual words within tweets to determine why one time period is happier than another , something that is not possible in traditional polls due to the multiple choice aspect of most surveys .
several other advantages of using tweets for public opinion polling include the ability to track movement @xcite and make maps @xcite using geolocated tweets .
data collection itself is largely algorithmic , and does not rely on the responses of participants .
we find that for many topics , twitter is a valuable resource for mining public opinions without solicited surveys .
we encourage readers to explore the data online at http://compstorylab.org/share/papers/cody2016a/data.html .
social media may be the future of public opinion polling , revealing important signals complementary to traditional surveys .
the authors are grateful for the computational resources provided by the vermont advanced computing core and the vermont complex systems center .
psd and cmd acknowledge support from nsf grant # 1447634 .
emc acknowledges support from the national science foundation under project dge-1144388 .
10 the demographics of social media users .
http://www.pewinternet.org/2015/08/19/the - demographics - of - social - media - users/. accessed : 2016 - 07 - 18 .
gallup trends .
accessed : 2016 - 03 - 08 .
google correlate .
https://www.google.com/trends/correlate , .
pollster api .
accessed : 2016 - 03 - 08 .
university of michigan index of consumer sentiment .
accessed : 2016 - 03 - 08 .
d. antenucci , m. r. andwerson , p. zhao , and m. cafaerlla .
a query system for social media signals .
e. aramaki , s. maskawa , and m. morita .
witter catches the flu : detecting influenza epidemics using twitter . in _ proceedings of the conference on empirical methods in natural language processing _ , pages 15681576 .
association for computational linguistics , 2011 .
p. barber .
less is more ?
how demographic sample weights can improve public opinion estimates based on twitter data .
j. bollen , h. mao , and x. zeng .
twitter mood predicts the stock market .
, 2(1):18 , 2011 .
n. cao , y .- r .
lin , x. sun , d. lazer , s. liu , and h. qu . whisper : tracing the spatiotemporal process of information diffusion in real time .
, 18(12):26492658 , 2012 .
z. chu , s. gianvecchio , h. wang , and s. jajodia . detecting automation of twitter accounts : are you a human , bot , or cyborg ?
, 9(6):811824 , 2012 .
r. b. cialdini and n. garde . ,
volume 3 .
a. michel , 1987 .
e. m. clark , j. r. williams , c. a. jones , r. a. galbraith , c. m. danforth , and p. s. dodds .
sifting robotic from organic text : a natural language approach for detecting automation on twitter . , 2015 .
e. m. cody , a. j. reagan , l. mitchell , p. s. dodds , and c. m. danforth .
climate change sentiment on twitter : an unsolicited public opinion poll .
, 10(8):e0136092 , 2015 .
j. p. dickerson , v. kagan , and v. subrahmanian .
using sentiment to detect bots on twitter : are humans more opinionated than bots ? in _ advances in social networks
analysis and mining ( asonam ) , 2014 ieee / acm international conference on _ , pages 620627 .
ieee , 2014 .
p. u. diehl , b. u. pedroni , a. cassidy , p. merolla , e. neftci , and g. zarrella .
truehappiness : neuromorphic emotion recognition on truenorth . , 2016 .
j. digrazia , k. mckelvey , j. bollen , and f. rojas .
more tweets , more votes : social media as a quantitative indicator of political behavior . , 8(11):e79449 , 2013 .
p. s. dodds , k. d. harris , i. m. kloumann , c. a. bliss , and c. m. danforth .
temporal patterns of happiness and information in a global social network : hedonometrics and twitter .
, 6(12):e26752 , 2011 .
e. ferrara , o. varol , c. davis , f. menczer , and a. flammini .
the rise of social bots .
, 2014 .
m. r. frank ,
l. mitchell , p. s. dodds , and c. m. danforth .
happiness and the patterns of life : a study of geolocated tweets .
. d. gayo - avello .
`` i wanted to predict elections with twitter and all i got was this lousy paper''a balanced survey on election prediction using twitter data . , 2012 .
d. gayo - avello . a meta - analysis of state - of - the - art electoral prediction from twitter data .
, page 0894439313493979 , 2013 .
m. glenski and t. weninger .
rating effects on social news posts and comments . , 2016 .
b. helmuth , t. c. gouhier , s. scyphers , and j. mocarski .
trust , tribalism and tweets : has political polarization made science a `` wedge issue '' ?
, 3(1):1 , 2016 .
p. n. howard and b. kollanyi .
bots , # strongerin , and # brexit : computational propaganda during the uk - eu referendum . , 2016 .
s. kamvar and j. harris . .
simon and schuster , 2009 .
i. m. kloumann , c. m. danforth , k. d. harris , c. a. bliss , and p. s. dodds .
positivity of the english language . , 7(1):e29484 , 2012 .
y. kryvasheyeu , h. chen , n. obradovich , e. moro , p. van hentenryck , j. fowler , and m. cebrian . rapid assessment of disaster damage using social media activity . , 2(3):e1500779 , 2016 .
d. lazer , r. kennedy , g. king , and a. vespignani .
the parable of google flu : traps in big data analysis .
, 343:12031205 , 2014 .
lin , b. keegan , d. margolin , and d. lazer .
rising tides or rising stars ? :
dynamics of shared attention on twitter during media events .
, 9(5):e94093 , 2014 .
lin , d. margolin , b. keegan , and d. lazer .
voices of victory : a computational focus group framework for tracking opinion shift in real time . in _ proceedings of the 22nd international conference on world wide web _ , pages 737748 .
international world wide web conferences steering committee , 2013 .
y. mejova , i. weber , and m. w. macy . .
cambridge university press , 2015 .
g. miller .
social scientists wade into the tweet stream .
, 333(6051):18141815 , 2011 .
l. mitchell , m. r. frank , k. d. harris , p. s. dodds , and c. m. danforth .
the geography of happiness : connecting twitter sentiment and expression , demographics , and objective characteristics of place .
, 8(5):e64417 , 2013 .
a. mittal and a. goel .
stock prediction using twitter sentiment analysis . , 2012 .
b. oconnor , r. balasubramanyan , b. r. routledge , and n. a. smith . from tweets to polls : linking text sentiment to public opinion time series .
, 11(122 - 129):12 , 2010 .
a. pak and p. paroubek .
twitter as a corpus for sentiment analysis and opinion mining . in _ lrec _ ,
volume 10 , pages 13201326 , 2010 .
m. j. paul and m. dredze .
you are what you tweet : analyzing twitter for public health .
, 20:265272 , 2011 .
t. preis , h. s. moat , and h. e. stanley .
quantifying trading behavior in financial markets using google trends .
, 3 , 2013 .
a. reagan , b. tivnan , j. r. williams , c. m. danforth , and p. s. dodds .
benchmarking sentiment analysis methods for large - scale texts : a case for using continuum - scored words and word shift graphs , 2015 .
j. ritterman , m. osborne , and e. klein .
using prediction markets and twitter to predict a swine flu pandemic . in _
1st international workshop on mining social media _ , volume 9 , pages 917 .
uk / miles / papers / swine09 .
pdf ( accessed 26 august 2015 ) , 2009 .
t. sakaki , m. okazaki , and y. matsuo .
earthquake shakes twitter users : real - time event detection by social sensors . in _ proceedings of the 19th international conference on world wide web _ , pages 851860 .
acm , 2010 .
m. j. salganik , p. s. dodds , and d. j. watts . experimental study of inequality and unpredictability in an artificial cultural market . , 311(5762):854856 , 2006 .
a. tumasjan , t. o. sprenger , p. g. sandner , and i. m. welpe .
predicting elections with twitter : what 140 characters reveal about political sentiment .
, 10:178185 , 2010 . c. vaccari , a. valeriani , p. barber , r. bonneau , j. t. jost , j. nagler , and j. tucker
. social media and political communication : a survey of twitter users during the 2013 italian general election .
, 43(3):381410 , 2013 .
x. wang , m. s. gerber , and d. e. brown .
automatic crime prediction using events extracted from twitter posts . in _ social computing , behavioral - cultural modeling and prediction _ , pages 231238 .
springer , 2012 .
we use anomaly correlation ( pearson correlation ) to determine the relationship between the twitter happiness time series and the polling data .
when doing so , we subtract the mean of the series , @xmath11 , from each data point , @xmath12 , to determine anomalies , and then calculate the cosine of the angle between the series of anomalies , i.e. , @xmath13 @xmath14 the variables @xmath15 and @xmath16 represent happiness time series and polling time series respectively , and @xmath17 is the length of the time series .
each word in our data set was previously assigned a happiness score through amazon s mechanical turk ( labmt scores ) .
we investigate the relationship between surveyed scores and ambient happiness scores in fig .
[ scatter ] .
we find a positive slope , indicating that ambient happiness rises with surveyed happiness , however we see a much smaller range of scores , which can be attributed to averaging a large amount of words .
we give the top 10 and bottom 10 words sorted by ambient happiness in table [ sort_amb ] .
top words included birthday wishes and prize giveaways , and bottom words suggest legal news stories .
gallup trends provide yearly polling data on many topics without a subscription .
the gallup survey questions can be found in table [ questions ] .
these polls , however , take place only once a year in the same month over several days .
this presents a challenge as to the amount of twitter data we should include in our correlations , as opinions may change . for each gallup datapoint
, we use the current year s worth of tweets from 2009 through 2015 for various subjects of national or global interest .
[ good ] shows several topics that correlate quite well with ambient happiness on twitter .
we find that the favorability of two major countries , iran and iraq , has a positive correlation with the ambient happiness of iran " and iraq " .
we also find that the united states opinion on religion has a strong positive correlation with yearly ambient happiness of religion " .
other topics , including the united states opinion on afghanistan and immigration show no significant correlation to twitter data .
there is a strong negative correlation between the satisfaction of the position of the united states in the world , indicating there may be some sarcasm associated with usa " tweets .
|
solicited public opinion surveys reach a limited subpopulation of willing participants and are expensive to conduct , leading to poor time resolution and a restricted pool of expert - chosen survey topics . in this study , we demonstrate that unsolicited public opinion polling through sentiment analysis applied to twitter correlates well with a range of traditional measures , and has predictive power for issues of global importance .
we also examine twitter s potential to canvas topics seldom surveyed , including ideas , personal feelings , and perceptions of commercial enterprises .
two of our major observations are that appropriately filtered twitter sentiment ( 1 ) predicts president obama s job approval three months in advance , and ( 2 ) correlates well with surveyed consumer sentiment . to make possible a full examination of our work and to enable others research ,
we make public over 10,000 data sets , each a seven - year series of daily word counts for tweets containing a frequently used search term .
|
in our paper @xcite we described an upgrade of the monte carlo generator tauola using the results of the resonance chiral lagrangian ( @xmath3 ) for the @xmath2 lepton decay into the most important two and three meson channels .
the necessary theoretical concepts were collected , numerical tests of the implementations were completed and documented . finally , we presented strategy for fitting experimental data and the systematic uncertainties associated with the experimental measurement .
however , there was and remain until now , an obvious limitation due to the fact that we are using one - dimensional projections of the invariant masses for a multi - dimensional distribution .
the first comparison @xcite of the @xmath3 results for the @xmath4 mode with the babar data @xcite , did not demonstrated a satisfactory agreement for the two pion invariant mass distributions . with the recent availability of the unfolded distributions for all invariant masses constructed from observable decay products for this channel @xcite , we found ourselves in an excellent position to work on model improvement for the @xmath4 mode .
we would like to stress here that the choice of the three pion mode is not accidental .
the kinematical configuration is complex and the three pion mode has the largest branching ratio among the three meson decay modes .
moreover , this decay mode together with the decays into two pions , which are much easier to model , are a useful tool for spin - parity analysis of the recently discovered higgs boson @xcite through its di-@xmath2 decays @xcite .
our paper is organized as follows . in section 2
we present modifications to the currents previously prepared by us @xcite .
we give a brief motivation for the choice of selected extensions . in section 3 ,
we present numerical results and we discuss available options .
section 4 is devoted to documentation of our fitting approach , which could be substantially simplified thanks to availability of unfolded invariant mass distributions . in section 5
we discuss systematic uncertainties for the fit resulting from statistical and systematic uncertainties of the experimental data .
some technical details of our set - up use are collected in this section as well .
summary , section 6 , closes the paper .
for the final state of three pions @xmath5 , @xmath5 , @xmath6 with the momenta respectively @xmath7 , @xmath8 and @xmath9 , lorentz invariance determines the decomposition of the hadronic current to be @xcite @xmath10\nonumber\\ & & + q^\mu f_4 - { i \over 4 \pi^2 f^2 } c_5 \epsilon^\mu_{.\ \nu\rho\sigma } p_1^\nu p_2^\rho p_3^\sigma f_5 \bigr\ } , \label{fivef}\end{aligned}\ ] ] where : @xmath11 denotes the transverse projector , and @xmath12 is the momentum of the hadronic system .
here @xmath13 stands for the pion decay constant in the chiral limit . in the isospin symmetry limit ,
the @xmath14 form factor for the three pion mode is zero due to @xmath15-parity conservation @xcite and thus we will neglect it .
functions @xmath16 , the hadronic form factors , depend in general on three independent invariant masses that can be constructed from the three meson four - vectors .
we chose @xmath17 and two invariant masses @xmath18 , @xmath19 built from pairs of momenta
. then @xmath20 can be calculated . ] from the other three invariants and is @xmath21 .
the form of the hadronic current is the most general one and constrained only by lorentz invariance .
the normalization factor is @xmath22 .
it is convenient to write down the hadronic form factors as @xmath23 where @xmath24 is the chiral contribution , @xmath25 is the one resonance contribution and @xmath26 is the double - resonance part .
the @xmath27 constant equals 1 for @xmath4 ( and -1 for @xmath28 ) . the exact form of the function @xmath16 is presented in @xcite eqs .
( 4)-(11 ) .
the comparison to data @xcite hints that the lack of the @xmath29 ( or @xmath1 ) meson contribution in our parameterization may be responsible for this discrepancy decays @xcite ( see also the focus article @xcite ) , where the @xmath1 contribution was modeled relying on a breit - wigner factor . in this case , the inclusion of the @xmath1 contribution is mandatory to describe the data , since it produces basically half of the decay width .
the good agreement with data shown in @xcite can be taken as a support of modeling the @xmath1 contribution by means of a breit - wigner , as it is done in this paper , see eq .
( [ eq : bw_sig ] ) , and can be considered as phenomenologically sufficiently sound solution till we have better based parameterization . ] .
the @xmath1 meson is , predominantly , a tetraquark state @xcite , and it can not be included in the @xmath3 formalism decays , the @xmath1 meson has been observed to play an important role in @xmath30 scattering @xcite , and it has also been relevant for the understanding of the @xmath31 and @xmath32 decays @xcite .
however further investigation of its effect in processes with the same hadronic final states , as the one considered in this paper , is required .
see our discussion on the fit to the experimental data including @xmath1 and other alternatives , as coulomb interaction , in section [ sec : numeric ] and the paper summary .
] , which is devised for ordinary @xmath33 resonances assignment for the lightest axial - vector mesons is favored within the @xmath3 approach @xcite , see however refs .
@xcite . ] .
the inclusion of rescattering effects of this kind is an involved task from the computational point of view , even in the case of scalar amplitudes @xcite . in view of this
we have decided to incorporate the @xmath1 meson following a more phenomenological approach , more specifically , a simple extension of the one used by cleo @xcite .
this effect is included into the @xmath34 and @xmath35 form factors in the following way @xmath36\ , , { \nonumber}\\ f_1^{\mbox{\tiny rr } } & & \rightarrow f_1^{\mbox{\tiny rr}}+ \frac{4f_ag_v}{3f^2 } \frac{q^2}{q^2-m_{a_1}^2-im_{a_1}\gamma_{a_1}(q^2 ) } \\ & & \left[\gamma_\sigma bw_\sigma(s_1)f_\sigma(q^2,s_1 ) + \delta_\sigma bw_\sigma(s_2)f_\sigma(q^2,s_2)\right ] , \nonumber\end{aligned}\ ] ] where @xmath37 , \label{eq : f_sig}\end{aligned}\ ] ] and @xmath38 and @xmath39 .
bose symmetry implies that the form factors @xmath40 and @xmath41 are related @xmath42 . as a consequence the hadronic current ( [ fivef ] ) is symmetrical under exchange of @xmath43 and @xmath44 .
the main differences between the @xmath1 meson parameterization , eqs .
( [ eq : ff_sig ] ) to ( [ eq : f_sig ] ) , and the one used by cleo can be summarized as : * there is one resonance contribution ( @xmath45 ) in ( [ eq : ff_sig ] ) whereas the cleo parameterization contains only the double resonance contributions ( @xmath46 ) .
the presence of the one resonance diagrams is a feature of @xmath3 ( and other lagrangian approaches ) , for discussion see section 7 of @xcite ; * @xmath47 and @xmath48 , no symmetry requirement enforces these equalities in equation ( [ eq : ff_sig ] ) . in contrary , cleo collaboration @xcite uses the simplified case where vertices @xmath49 and
@xmath50 are assumed to be constant and @xmath51 and @xmath52 ; * the cleo collaboration fixed both the @xmath1 mass and its width to the values predicted by the tornqvist unitarized quark model @xcite whereas we fit these parameters .
the width for the @xmath53 decay can be written down in the terms of the hadronic form factors in eqs .
( [ eq : ff_gen ] ) and ( [ eq : ff_sig ] ) @xmath54 , { \nonumber}\label{q2spec}\end{aligned}\ ] ] where @xmath55 the following phase space integration limits have to be used @xmath56 where @xmath57 ^ 2\bigg\ } \ , . { \nonumber}\end{aligned}\ ] ] the coulomb interaction can be important near the production threshold .
we use the far - field approximation ; the final - state pions are treated as stable point - like objects and the three pion interaction is treated as a superposition of the two pion ones . the corresponding invariant masses ( we have attractive interaction for the pion pairs of the invariant mases @xmath43 and @xmath44 and the repulsive interaction for the @xmath58 ) define the strength of the pair interactions .
the two pion system can be in s or p - wave state . to estimate the coulomb interaction in s - wave
we can apply the results of section 94 of ref .
@xcite which changes the differential decay width as follows @xmath59}\ ; \ ; \ ; \ ; \ ; \ ; \ ; \ ; \ ; \\ & & \frac{2\alpha \pi / v_0(s_2)}{1-\mathrm{exp}\left[-2\alpha \pi / v_0(s_2)\right]}\ ; \frac{2\alpha \pi / v_0(s_3)}{\mathrm{exp}\left[2\alpha \pi / v_0(s_3)\right]-1 } { \nonumber}\,,\end{aligned}\ ] ] where in r.h.s .
@xmath60 is the s - wave part of the differential decay width ( see eq . [ eq : wid3pi ] ) neglecting the coulomb interactions , @xmath61 is the relative velocity of two pions in the two pion system with invariant mass @xmath62 , i.e. @xmath63 , where @xmath64 is defined below eq .
( [ eq : bw_sig ] ) . precision studies of the two - pion vector form factor ( see e.g. @xcite and references therein ) , more specifically , a p - wave two pion state , did not require to include the coulomb interaction between the pion pair to describe the data accurately . we will consequently neglect p - wave coulomb interaction among the final - state pions and stick to formula ( [ eq : coul ] ) to evaluate the effect of electromagnetic interactions among the pions and set ( [ eq : wid3pi ] ) on r.h.s . of ( [ eq : coul ] ) . the @xmath65 width can be written down as the imaginary part of the two - loop axial - vector axial - vector correlator with suitable flavour indices @xcite @xmath66 where @xmath67 stands for the contribution from the individual three - pion @xcite and two kaons - one pion @xcite absorptive cuts . as similar integrands
are present in eq .
( [ eq : wid3pi ] ) for the partial decay width of @xmath2 to three pions and in eq .
( [ eq : a1width ] ) for the @xmath68-dependent @xmath65 width , we could use this property to simplify the code for calculating the invariant mass spectra and profit in full from the unfolded invariant mass distribution of babar which recently became public @xcite .
the monte carlo simulations could be avoided and we could use semi - analytical functions in fits .
we would like to stress that we neglect , in our numerical analysis , the @xmath1 contribution on r.h.s of eq .
( [ eq : a1part ] ) , which should be suppressed in the large-@xmath69 counting .
we will come to discussion of additional ambiguities related to the @xmath1 contribution in the summary .
[ cols="<,<,<,<,<,<,<,<,<",options="header " , ] this kind of analysis is mandatory for the @xmath70 and @xmath71 resonance , for which the determination of its pole parameters has become an area of precision calculations . in particular
there is very nice agreement in the literature for the @xmath71 pole parameters : @xmath72 mev and @xmath73 mev using different types of data and methods ( see @xcite and references therein for details ) . our value in table [ tab : poleparameters ] ,
@xmath74 mev@xmath75 is compatible both for the mass and width , although the last value appears slightly large .
we believe that this may be the result of the degeneracies and instabilities of the fit caused by the model used for the @xmath1 . in the case of the @xmath76 resonance ,
the comparison of our results in table [ tab : poleparameters ] with those in ref .
@xcite show agreement for the width but some tension on the mass that is @xmath77 mev smaller in table [ tab : poleparameters ] .
it is important to note here , that there are observed resonances , @xmath78 and @xmath79 , which were reported cleo @xcite and are suggested by the fit to the babar data .
these additional resonances and the correlations between the @xmath80 and @xmath81 parameters in the fits may also have affected the pole parameters of these resonances . therefore , until these problems have been solved , the pole values for the masses and widths for the @xmath1 , @xmath80 and @xmath81 may be of limited use .
after a long - standing debate about the existence of the @xmath1 , it is well - established now and the state - of - the - art dispersive analyses of @xmath30 scattering data @xcite agree within a few mev for the pole position of the @xmath1 around @xmath82 mev .
although with less precision , the results obtained using @xmath83 data , @xmath84 gev@xmath75 @xcite , are in reasonable agreement with the previous values , although with a smaller width .
our figures in table [ tab : poleparameters ] , @xmath85 gev@xmath75 , are in accord with the more precise determinations obtained analyzing @xmath30 scattering data . despite this result
supports the consistency of our picture where the role of the @xmath1 in the examined decays is manifest ( specially in the decay distribution at low values of the @xmath86 invariant mass ) we judge that it is still too early to claim this as a confirmed fact and we would like to see if this trend is kept analyzing the complete multi - dimensional data set and/or improving the model of the @xmath1 contribution and possibly adding the missing , for example the @xmath87 resonance exchange , as it is discussed in the conclusions .
we are thus confident that in future stages of our work , when multi - dimensional distributions will be available for fits we shall be able to improve the determination of the poles associated to the resonances whose exchange dominates the @xmath88 decays . for the above applications to be possible we had to adapt tauola library .
let us document the necessary changes . the following fortran functions with one , two and three - dimensional distributions have been prepared .
note that all of these functions have been also defined in + demo - fit / [email protected] for use in c / c++ environment .
all of the parameters and return values of these functions are double precision .
* function ffwid3pi(qq , s1,s3 ) + input : qq = @xmath90 , s1 = @xmath91 , s3 = @xmath92 .
returns @xmath93 .
if qq s1 s3 are outside of the phase space , this function returns zero . *
function dgams3qq(qq ) + function dgams1qq(qq ) + calculates @xmath94 width as function of f(qq , s3 ) and f(qq , s1 ) .
the second parameter of the calculation is hidden in common block external .
the definition in demo - fit / [email protected] also includes the wrappers for these functions that handle the hidden parameter . *
function dgams1(xs1b ) + function dgams3(xs3b ) + function dgamqq(xqqb ) + these functions calculate @xmath95 width as function of s1 , s3 or qq using formula ( [ eq : ff_gen ] ) , ( [ eq : ff_sig ] ) , limits of integration are in ( [ eq : lim_int ] ) . to run the example program ,
it is required that root is installed and its configuration is available through root - config .
* compile tauola - fortran library by executing make in tauola - fortran / tauola . *
compile additional libraries by executing make in tauola - fortran / demo - standalone .
* compile example by executing make in tauola - fortran / demo - fit . *
execute ./[email protected] in tauola - fortran / demo - fit .
the program should return output : + + last integration variable qq : total width of tau= ... + last integration variable s1:total width of tau= ... + last integration variable s3:total width of tau= ... + calculating h12 ( dgams3 ) ... + calculating h13@xmath8923 ( dgams1 ) ... + calculating h123 ( dgamqq ) ... + + where the total widths depend on the options and parameters used for computation .
additionally , output file out.root should be present with the histograms h12 , h13@xmath8923 and h123 containing the distributions s3 , s1 ( combined with s2 ) and qq correspondingly .
this is a basic example used mostly as a technical test of numerical integration .
it shows how the semi - analytic distributions can be accessed from the c++ program , and how they can be used to produce the histograms for comparison with the data .
it also shows how fit parameters can be modified .
it can be used as a starting point for further analysis or for writing new fitting algorithm .
the main fitting algorithm is located in subdirectory tauola / tauola / fitting . as the algorithm itself is not a focus of this paper
, its detailed description will be omitted . for users interested in more details
we refer to readme inside this directory for instructions on how to run the main program and how to use the algorithms provided .
we refer to header files for documentation of the details of their use .
we would like to point out that , as reiterated in tauola / fitting / readme , this code is not standalone and requires properly formatted data files which are not part of the distribution . finally let us point that the tar - ball of software distribution which includes code for the present work would also include code for unfinished work on other @xmath2 decay modes such as @xmath96 .
that is why , at present , we plan to distribute the tar - ball upon request only .
in the present paper we have documented modification of the resonance chiral theory currents for @xmath97 decay of reference @xcite , necessary for agreement with the experimental data .
the choice of this channel was motivated by its relatively large branching ratio , availability of unfolded experimental distribution and already non - trivial dynamics of three - pion final state .
in addition , this channel is important for higgs spin - parity studies through the associated di-@xmath2 decays .
previously missing contribution from the @xmath1 resonance was added to our currents and final state coulomb interactions were taken into account . as a result , we improved agreement with the data by a factor of about eight .
remaining differences are well below uncertainties expected for the @xmath3 currents approach .
this is the first case when agreement for a non - trivial @xmath2 decay channel was obtained between the babar data @xcite and the theoretical model .
these comparisons allow for future precision tau decays physics studies @xcite .
the agreement with data has improved , mainly , due to the inclusion of the @xmath1 meson , which is outside the @xmath3 approach . in tables
[ tab : fit ] and [ tab : fit - results ] , it can be seen that the value of @xmath98 is at the boundary .
this appears to be the result of a numerical instability in the phenomenological description of the @xmath1 caused by the correlation between the @xmath99 in the exponential decay , equation 6 , and with width of the breit - wigner , equation 4 .
this problem can only be resolved with a better parametrization of the @xmath1 which is currently not available to us . without the @xmath1 contribution
the @xmath3 demonstrates only slightly better agreement with data than the cleo model @xcite , fig.1 .
therefore , we can not exclude that fits of similar good quality could be obtained if further work was done on the cleo model approach . ] .
however , the @xmath3 approach ensures that the results for all hadronic currents reproduce the chiral limit of qcd at least up to next - to - next - to - leading order decay modes can be found in refs .
@xcite . ] .
in contrast , the phenomenological approximation , as it was done at cleo , do not do that @xcite .
we will come back to this point in our future analysis of the @xmath100 modes .
even though this work is based on one dimensional unfolded experimental distributions , we think that it is an important step forward .
we have enriched our fitting arrangements , to profit from the availability of invariant mass distributions .
use of unfolded distributions and our technical arrangements substantially improved the speed of our fitting .
we have implemented solutions based on parallel calculations .
the @xmath3 currents are ready for comparison with data for other @xmath2 decay channels and for work when unfolded multi - dimensional distributions are used .
this may constrain the set of parameters obtained from fit to the @xmath0 channel , which features strong correlations .
there exist other solutions ( local minima ) for the parameters which give somewhat worse agreement with the experimental data .
it is worth mentioning that for the results of final fit the minuit hesse method evaluates some elements of the correlation matrix to surpass 0.9 . an alternative approach , using linear approximation of the dependence on fitted parameters ( described in ref .
@xcite ) , is pursued independently @xcite .
it may become useful in later phases of the project .
it has an advantage that it allows for experimental cuts to be introduced into the fitted distributions .
it is , however , much slower and features instabilities if the choice of the fitted function is poor .
it provides , however , independent test of our numerical methods used in the paper . in the present paper a simpler fitting method has been used as the unfolded distributions from experimental data are now available @xcite .
discrepancies in the high mass region of the @xmath101 invariant mass indicate the possibility of missing resonances in our @xmath3 approach .
this is consistent with the observation of additional resonances , more specifically the @xmath78 and @xmath79 , by cleo in @xcite . around @xmath102 mev
, there is another discrepancy between the model and the data that suggests the presence of the @xmath103 resonance , the significance of the difference between the model and the data has been computed to be around @xmath104 @xmath1 in the region from @xmath105 mev to @xmath106 mev . the @xmath103 resonance , as opposed to the @xmath1 , is included in the @xmath3 framework @xcite and its ( small ) contribution to the @xmath107 decays can with some time be evaluated . ] .
although we could add phenomenologically the contribution of these resonances to the amplitude , we prefer not to do it at the moment to keep a compromise between the number of parameters , the stability of the fit and the amount of experimental data
. we also comment below on the need to include the contribution of the @xmath87 resonance .
it should be pointed out that the necessity of the @xmath1 in our model may be an artifact caused either by neglecting the dynamics associated with these additional resonances parameters are unconstrained in the fit , they usually leave the expected physical range hints to such a possibility .
this is especially highlighted by the fact that even in our results , parameter @xmath108 is at its upper limit ( see table [ tab : fit ] ) .
fit results for @xmath1 parameters seem to depend on minuscule changes in its setup , this may also support conjecture of unphysical nature of @xmath1 contribution to the currents or to its inaccurate modeling . ] or by the crudeness of our @xmath1 parameterization .
the largest discrepancies between data and the fitted distribution , which are responsible for a significant part of the total @xmath109 , are observed in the 3@xmath110 invariant mass distributions . the slope and shape of the disagreement in the 3@xmath110 invariant mass spectra , in particular around @xmath111 gev in fig .
[ fig : pipipi ] , indicates the possibility of interference between @xmath112 and its excited state @xmath87 .
the disagreement in the low mass regions around @xmath113 gev and below could be the result of a bias in the fit caused by leaving out the additional resonance .
although we currently restrict ourselves to the lowest axial - vector resonance , @xmath112 , in future analysis of multi - dimensional distributions we are going to include the @xmath87 in the same way as it was done for the @xmath114 to test our hypothesis the constant width approximation can be used for this resonance . ] .
we prefer not to do this at the moment because the improvement of the fit on the peak and tail regions in the three - meson invariant mass distribution will come again together with an increase of the fit instability and a growth of the correlations between the ( larger number of ) fit parameters . lacking the complete multi - dimensional distributions
it is advisable to proceed this way at present .
cleo has also suggested a hint of a possible contribution of the pseudo - scalar resonance @xmath115 , which they have excluded at the @xmath116 level with a 90% cl@xcite .
this is below the current sensitivity of our fit .
finally , we note that previous dedicated studies within the @xmath3 framework @xcite did not obtain an improved description of the data when including the chiral logarithms @xcite .
this question could also be readdressed when more exclusive data will become available .
this paper represents a significant improvement in the agreement between our model and the data .
the resulting fit parameters are substantially improved and are reasonable from the point of view of internal consistency of @xmath3 .
the obtained value for the @xmath65 mass can be used for the future theoretical study .
predictions for the @xmath28 are obtained from the ones for @xmath4 directly with the help of isospin symmetry without any modifications except the part of @xmath1 contribution which was adapted and constraints from cleo were used . then , our current results with the partial branching ratio , which is @xmath117 higher than the pdg result , is reasonable for our somewhat simplistic assumption . on technical side ,
we have prepared fitting environment which can be after minor adaptation used for multi - dimensional experimental distributions or other @xmath2 decay modes .
we acknowledge help and discussion with jakub zaremba and sergey alekhin on numerical details of fitting arrangements .
we are also thankful to alexander korchin and pedro ruiz femena for a discussion about the coloumb interaction and to rafel escribano , bastian kubis and jorge portols for their comments about the @xmath1 meson .
useful discussions with john michael roney , simon eidelman , hisaki hayashii , denis epifanov and swagato banerjee are acknowledged .
we thank members of babar collaboration for discussions , in particular william dunwoodie for discussion of the @xmath1 meson .
we benefited from discussions with daniel gmez dumm on the determination of the pole parameters of the resonances exchanged in @xmath107 decays .
this project is financed in part from funds of polish national science centre under decisions dec-2011/03/b / st2/00107 and dec-2011/03/b / st2/00220 .
we acknowledge funding from the alexander von humboldt - stiftung / foundation and the spanish grant fpa2011 - 25948 as well . p.r .
acknowledges dgapa for funding his contract and the support of project papiit in106913 .
o.s . acknowledges that the final step of this project is financed in part from funds of foundation of polish science grant pomost/2013 - 7/12 .
pomost programme is cofinanced from european union , regional development fund .
this project was supported in part by pl - grid infrastructure .
|
in this paper we document the modifications introduced to the previous version of the resonance chiral lagrangian current ( _ phys.rev . _ * d86 * ( 2012 ) 113008 ) of the @xmath0 decay which enable the one dimensional distributions measured by the babar collaboration to be well modeled . the main change required to model
the data is the addition of the @xmath1 resonance .
systematic errors , theoretical and experimental ones , limitations due to fits of one dimensional distributions only , and resulting difficulties and statistical / systematic errors for fitted parameters are addressed .
the current and fitting environment is ready for comparisons with the fully exclusive experimental data .
the present result for @xmath0 is encouraging for work on other @xmath2 decay modes and resonance chiral lagrangian based currents .
|
supersolids with both diagonal and off - diagonal long - range order are observed in various model , either with or without hard - core constraint.@xcite one obvious necessary condition for the occurrence of supersolids ( sss ) is the presence of a small but finite kinetic energy in competition with a relatively large repulsive interactions .
previous results suggest that the hard - core constraint may destroy supersolid phases by the formation of domain walls , which leads to phase separation instead.@xcite stable supersolid phases in hard - core models are found , however , in systems contain frustrated interactions.@xcite the extended bose - hubbard model is a typical example that has been proposed to demonstrate a supersold phase within the mean - field approximation @xcite , and is confirmed by quantum monte carlo simulations.@xcite as long as the next - nearest - neighbor ( nnn ) interaction @xmath2 is dominant , hard - core bosons lining up as stripes to reduce potential energy near half filling @xcite ( see fig .
[ ordering ] ) . at quarter filling , frustrations induced by competing nearest neighbor ( nn ) @xmath3 can lead to a stable star order phase that characterized by finite structure factors at wave vector @xmath4 and @xmath5 [ @xmath6 .
the existence of the corresponding supersolid is also numerically confirmed recently.@xcite the melting of stripe supersolids has been investigated in ref .
9 where the finite - temperature phase diagrams are determined . at half filling ,
the stripes melt at a first order phase transition , while in the doped system , the melting transitions of the solid and supersolid are either very weakly first order or of second order .
both the broken rotational and translational symmetries of the stripe phases are found to be restored at a single continuous transition and no intermediate nematic ( liquid crystal ) phase has been observed .
nevertheless , ref .
9 focuses solely on the case with no nn interaction @xmath3 and so only the melting of stripe phases are discussed .
the thermal melting behavior of the star supersolid phases near the quarter filling is then largely unknown .
since the star order is characterized by two structure factors , the two broken translation symmetries can , in principle , be restored simultaneously or at two different temperatures .
the latter scenario means the star phase melts into a stripe phase before changes to a normal fluid or superfluid .
it is also worth to examine whether the missing nematic phase in ref .
9 can be stabilized by the introducing @xmath3 in the system .
it is the purpose of this work to address these issues and determine the finite - temperature phase diagram for the competing nn and nnn interactions .
to start with , the extended bose - hubbard model on a two - dimensional square lattice with the hamiltonian is defined as @xmath7 with hard - core constraint @xmath8 . in the above formula
, @xmath9 is the boson destruction ( creation ) operator and @xmath10 sums over the nearest ( next - nearest ) neighboring sites .
the energy scale is fixed by @xmath11 throughout this paper .
in the case of @xmath12 , it is well known that the ground state at half filling is a checkerbroad state , which becomes a superfluid as the system is doped away from half filling . on the other hand , in the case of @xmath13 ,
the half - filled ground state has a stripe order that characterized by the finite structure factors at wave vector * q*@xmath14 or * q*@xmath15 ( fig . [ ordering ] ) .
more interesting is the coexistence of the stripe order and superfluidity , i.e. , the stripe supersolid phase , in the doped system .
most condensed bosons in this exotic states move in the direction parallel to the stripes but superflow in the perpendicular direction is also observed.@xcite the stripe quantum solid proceeds to the superfluid via two second - order transitions , in contrast to the case of vanishing @xmath2 , where transitions from quantum solid to superfluid are discontinuous .
-@xmath1 rotational symmetry is broken in the stripe phase but not in the star phase.,width=283 ] the phase diagram becomes even richer when both @xmath3 and @xmath2 are finite .
an extra quarter - filled solid ( see fig . [ ordering ] ) and its corresponding supersolid are predicted within the mean - field theory @xcite and are confirmed by numerically simulations .
@xcite quarter - filled solids with two distinct crystal structures are possible depending on the competition of @xmath3 and @xmath2 , while only one structure is stable for the supersolid phase.@xcite this so - called star supersolid has been discussed in detail but the finite - temperature phase diagram has not been examined so far and therefore serves as the subject of this work .
one of the most efficient approach to study the bosonic models is the standard stochastic series expansion ( sse ) monte carlo method implemented with a directed loop algorithm.@xcite in sse , the superfluidity , given by @xmath16 , is computed by measuring the winding number fluctuation within the simulations .
hereafter we present the average superfluidity @xmath17 .
the broken translational symmetries in the @xmath0 and @xmath1 directions of the striped phases are characterized by the structure factors @xmath18 at the wave vectors @xmath19 and @xmath20 , respectively . for convenience
, we measure the average value of @xmath21/2,\ ] ] in order to avoid the difficulty to determine the direction of broken translational symmetry in each simulation .
moreover , we also study the rotational symmetry breaking in the local - density correlation by measuring the order parameter @xmath22 , @xmath23 with @xmath24 the boson number operator at the site @xmath25 .
we stress that @xmath22 signals the local broken rotational symmetry but not the global one .
the star order phase , on the other hand , is characterized by the finite values of both @xmath26 and @xmath27 , which breaks the translational symmetry in both directions .
it contains a half - filled stripe quantum solid and a quarter - filled star quantum solid and their corresponding sss .
finite - temperature phase diagrams scan along the lines a and b are presented in figs .
[ qf ] and [ ftp ] , respectively .
the dotted line indicates a first - order transition from the star supersolid to stripe supersolid .
the system size used here is @xmath28 with @xmath29.,width=283 ] the ground - state phase diagrams @xmath3 vs @xmath2 at quarter filling have been presented before @xcite .
the star supersolid is found to be stable for a wide region of parameters @xmath3 and @xmath2 even at commensurate filling @xmath30 .
for the following discussion , we choose a representative nn interaction @xmath31 of which the phase diagram @xmath32 vs. @xmath33 is shown in fig .
[ gs ] . because of the particle - hole symmetry , the symmetric upper half of the diagram is not shown .
the main lobe in the diagram is the striped quantum solid phase at half filling and from which decreasing the boson number leads to a striped supersolid phase .
as mentioned before , the striped supersolid is characterized by a finite @xmath27 that breaks one translational symmetry but preserves the translational symmetry in another direction ( fig .
[ ordering ] ) . around quarter filling , the @xmath26 becomes finite as the system enters the star supersolid phase .
the translational symmetry is now broken in both @xmath0 and @xmath1 directions . at exactly quarter filling the superfluidity @xmath34 vanishes for @xmath35 ( see fig . [ qf ] ) but remains finite otherwise . as shown in the fig . 1 of ref . 8 ,
the general features of the ground - state phase diagram can also be described by the mean - field theory .
the quarter - filled star quantum solid phase is enclosed by a narrow region of star supersolid phase before enters the stripe supersolid phase .
however , the mean - field approximation tends to overestimate the extent of the stripe and star supersolid phases as presented in the fig .
this indicates that the neglected quantum fluctuations in the mean - field calculations destabilize the supersolid phases and reduce the region of the supersolid phases in the ground - state phase diagram . in the following , we present the finite - temperature phase diagrams at quarter filling and away from quarter filling along the lines a and b in fig .
[ gs ] , respectively .
the black solid lines are guides to the eyes .
data points are obtained from the finite - size analysis as described in figs .
[ fsa ] and [ bc1 ] . in the following figures ,
error bars are smaller the symbol size unless otherwise stated.,width=283 ] figure [ qf ] shows the finite - temperature phase diagram with fixed boson density of 1/4 ( line a in fig . [ gs ] ) . in our grand - canonical simulation
we adjust the chemical potential @xmath32 in order to obtain the average particle density @xmath36 equals to quarter filling . at zero temperature ,
the off - diagonal long - range order is depressed when @xmath2 increases , while the diagonal long - range order , the star order in this case , is enhanced . in the intermediate @xmath2 ,
both orderings coexist and gives rise to a star supersolid phase .
the presence of the supersolid phase is in contrast to @xmath36=1/2 case where the superfluid proceeds to the stripe quantum solid via a first - order transition with no intermediate supersolid phase . in fig . [ qf ]
, the superfluid changes to the star quantum solid via two second - order transitions instead . for the study of thermal transitions ,
we carry out a finite - size analysis on the order parameters @xmath34 , @xmath27 , and @xmath26 . in a two - dimensional system ,
the superfluid to normal - fluid phase transition can be well described by the kosterlitz - thouless ( kt ) transition.@xcite plotting the superfluidity @xmath34 as a function of @xmath37 , as shown in fig .
[ fsa ] for a representative @xmath2 , the intersection values of @xmath38 for different sizes @xmath39 follow the logarithmic correction @xmath40\}$ ] . fitting to this relation gives the value of critical temperature @xmath41 for infinite system .
indeed , the values of @xmath42 shown in the inset of fig .
[ fsa ] follow very well the logarithmic correction and confirms the superfluid to normal fluid transition is of the kt type .
the critical temperature t@xmath43 , as shown in fig .
[ qf ] , reduces smoothly as @xmath2 increases and disappears for @xmath44 . for @xmath45 .
the inset shows the logarithmic correction @xmath40\}$].,width=283 ] next , we analyze the temperature dependence of the structure factors @xmath27 and @xmath26 . as mentioned before , the two broken translational symmetries can be restored at two different temperatures or restored at the same temperature . to determine the transition temperatures
, we recall that for second - order transitions , the fourth order binder cumulant ratios @xmath46 should intersect at the same critical point for different sizes @xmath39.@xcite figure [ bc1 ] shows the cumulant ratios of @xmath27 and @xmath26 , respectively . indeed ,
@xmath47 meet at the same transition points independent of the system size . particularly , the resulted transition points from both cumulant ratios are very close to each other and are indistinguishable within the statistical uncertainties of our data .
our data thus suggest a single second - order transition for the melting of the star quantum solid to the normal - fluid phase , or a very narrow intermediate stripe quantum solid in between the phases . and @xmath26 as a function of temperature for different system sizes ( @xmath31 , @xmath48).,width=283 ] to further clarify these two scenarios , we suppose there are two second - order phase transitions that restores the translation symmetries sequentially .
we expect both transitions are in the three - dimensions ising universality class as each transition breaks a z@xmath49 symmetry .
therefore the structure factors should follow the finite size scaling law @xmath50 $ ] with @xmath51 and @xmath52 as expected for the ising class.@xcite the data of @xmath53 for different @xmath39 should intersect at the same temperature that determines the value of @xmath54 . by rescaling the temperature to @xmath55
, all data will collapse into a single line , provided the data are close enough to the critical point where the scaling law is valid . as shown in figs .
[ fs1 ] and [ fs2 ] , the data of @xmath27 and @xmath26 intersect at a size - independent critical temperature , respectively .
the critical temperature of @xmath26 is found significantly lower than that of the @xmath27 and therefore suggests an intermediate stripe quantum solid phase .
however , the rescaled data collapse well only for @xmath56 but a clear derivation from the scaling law is observed for temperatures below @xmath54 .
this failure of data collapse invalidates the assumption of two transitions of ising type .
therefore it rules out the possibility of a very narrow intermediate stripe quantum solid phase lies between the star quantum solid and normal - fluid phases .
the melting of the star quantum solid to normal fluid is then likely to be a single second - order transition .
the system happens not to choose to restore the translational symmetry in one particular direction first and the other direction later , but it restores both broken translatonal symmetries simultaneously .
we will comment on this in the next section .
( @xmath45 ) by assuming the critical exponents @xmath51 and @xmath52 of the ising universality class . at low temperatures , significant derivations from the scaling law
indicate the transition is not of the ising type.,width=283 ] .,width=283 ]
, @xmath48 ) as a function of @xmath32 along the line b in fig . [ gs].,width=264 ]
when the system is doped away from the quarter filling , the finite - temperature phase diagram is even richer .
figure [ ftp ] presents the finite - temperature phase diagram , determined by the binder cumulants as discussed previously , as a function of chemical potential @xmath32 with fixed @xmath57 ( line b in fig .
the zero - temperature ground states are the superfluid , star supersolid , and stripe supersolid phases , respectively , as the boson number is increased .
the quantum phase transition of the superfluid to star supersolid is second order , whereas the transition from the star supersolid to the stripe supersolid is first order .
figure [ hist ] plots the ground - state order parameters as a function of @xmath32 from which an abrupt change is observed at the transition between two supersolids .
the histogram of @xmath22 ( see the inset ) shows the double peaks feature as a clear signature for a first - order transition .
this discontinuous transition reflects the fact that the star order is more than simply breaking another translational symmetry of the stripe order .
it implies the two broken translational symmetries in the star order are not independent and therefore can not be constructed by breaking the translational symmetries sequentially via two second - order transitions .
this also explains the direct melting of the star quantum solid observed in fig .
[ qf ] .
( @xmath31 , @xmath48 , @xmath58 , @xmath59 ) .
the parameters @xmath22 are plotted for sizes @xmath58 ( filled circles ) and @xmath28 ( open circles ) to demonstrate the finite - size effect .
inset : histogram of the parameter @xmath22 at @xmath60 shows a double peak feature of the first - order transition between the star and the stripe supersolid phases.,width=283 ] as the temperature is increased , again , the broken gauge symmetry is restored via a kt transition .
however , in contrast to the fig .
[ qf ] where the critical temperature @xmath61 reduced smoothly without being affected by the emergence of broken translational symmetries , @xmath61 in fig .
[ ftp ] exhibits a sudden drop in the star order phase .
the @xmath61 remains rather flat throughout the whole star supersolid phase .
this sudden suppress of superfluidity is thus unlikely due to the increase in particle number alone , but may be resulted by the coupling of the broken translational symmetries .
this is also in contrast to the supersolid melting in most systems @xcite where the restoration of superfluid symmetry and translational symmetry are largely independent .
this interesting behavior may be related to the intrinsic structure of the star supersolid and further study is needed to gain more insight of this particular phase . for larger chemical potential , and particle density
as well , we found that the stripe supersolid melts into a stripe quantum solid and then further into the normal fluid through a second - order transition . from the plots of binder cumulants for both @xmath27 and @xmath22 in fig .
[ bc2 ] , we observe that the transition temperatures for both broken translation and rotational symmetries are within our data uncertainties and again suggests a single second phase transition . therefore no intermediate nematic phase is observed .
while this finding is consistent to the case of vanishing nn interaction @xmath3 , whether larger @xmath3 will be able to stabilize the nematic phase is worth to study . and @xmath22 as a function of temperature for different system sizes ( @xmath31 , @xmath48).,width=283 ]
we present the finite - temperature phase diagrams of the extended bose - hubbard model in the vicinity of the quarter - filling .
the transitions from the super to normal phases that restore the gauge symmetry are of the kt type in nature .
interestingly , the transition temperature exhibits a sudden drop when the star order emerges .
it differs from the usual supersolid that the superfluid @xmath54 behaves largely independent of the crystal order of the system .
futher exploration of the possible coupling between the broken gauge and translational symmetries should provide more insight of this behavior . for the melting of star order , our result support the scenario of a single second - order transition instead of two ising - type second - order transitions .
translational symmetries are restored simultaneously and no intermediate phase of stripe order is found .
our result indicates that the two order parameters of the star order couple together and the state can not be viewed simply as two interwined stripe orders
. it will be interesting to investigate the effect of a small anisotropy in one dimension that lifts the degeneracy of the @xmath0 , and @xmath1 symmetry .
as one translational symmetry is more robust , the critical temperatures of the broken translational symmetries should be different and the melting of star quantum solid might occur as two second - order transitions
. the effect of anisotropy on the stripe order melting will also be worth to study to see if the nematic state can be stabilized near the stripe order .
we thank y. c. chen and m. f. yang for stimulating discussions .
the numerical computations are performed in the center for high performance computing of the thu . k.k.n .
acknowledges the support by the national center for the theoretical science and the finanical support from the nsc ( r.o.c . ) , grant no .
nsc 97 - 2112-m-029 - 003-my3 .
|
we investigate numerically the finite - temperature phase diagrams of the extended bose - hubbard model in a two - dimensional square lattice . in particular , we focus on the melting of supersolid phases of two different crystal orderings , stripe and star orders , arising from the competition of the nearest- and next - nearest -neighbor interactions in the vicinity of quarter filling .
the two crystal orders are the result of broken translational symmetry in either one or in both @xmath0 , and @xmath1 directions .
the broken gauge symmetry of the supersolids are found to be restored via a kosterlitz - thouless transition while the broken translational symmetries are restored via a single second - order phase transition , instead of two second - order transitions in the ising universality class . on the other hand ,
the phase transitions between the star and stripe orders are first order in nature .
|
preferential attachment random graphs are randomgraphs that evolve by sequentially adding vertices and edges in a random way so that connections to vertices with high degree are favored .
particular versions of these models were proposed by @xcite as a mechanism to explain the appearance of the so - called power law behavior observed in some real world networks ; for example , the graph derived from the world wide web by considering webpages as vertices and hyperlinks between them as edges . following the publication of @xcite , there has been an explosion of research surrounding these ( and other ) random growth models .
this work is largely motivated by the idea that many real world data structures can be captured in the language of networks [ see @xcite for a wide survey from this point of view ] .
however , much of this work is experimental or empirical and , by comparison , the rigorous mathematical literature on these models is less developed [ see @xcite for a recent review ] . for preferential attachment models ,
the seminal reference in the mathematics literature is @xcite , in which one of the main results is a rigorous proof that the degree of a randomly chosen vertex in a particular family of preferential attachment random graph models converges to the yule
simon distribution .
corresponding approximation results in total variation for this and related preferential attachment models can be found in @xcite and @xcite .
here we study the distribution of the degree of a fixed vertex in two preferential attachment models . in model
1 we start with a graph @xmath0 with two vertices labeled one and two with an edge directed from vertex two to vertex one .
given graph @xmath1 , graph @xmath2 is obtained by adding a vertex labeled @xmath3 and adding a single directed edge from this new vertex to a vertex labeled from the set @xmath4 , where the chance that @xmath3 connects to vertex @xmath5 is proportional to the degree of vertex @xmath5 in @xmath1 ( here and below degree means in - degree plus out - degree ) .
model 2 is one studied in @xcite and allows for self - connecting edges .
there , we start with a graph @xmath6 with a single vertex labeled one and with an edge directed from vertex one to itself . given graph @xmath1 , graph @xmath2
is obtained by adding a vertex labeled @xmath3 and adding a single directed edge from this new vertex to a vertex labeled from the set @xmath7 , where the chance that @xmath3 connects to vertex @xmath8 is proportional to the degree of vertex @xmath5 in @xmath1 ( a loop at a vertex contributes two to its degree ) and the chance that vertex @xmath3 connects to itself is @xmath9 .
let @xmath10 be the degree of vertex @xmath5 in @xmath1 under either of the models above .
our main result is a rate of convergence in the kolmogorov metric ( defined below ) of @xmath11 to its distributional limit as @xmath12 .
although the literature on these models is large , there is surprisingly little known about these distributions .
the fact that these limits exist for the first model has been shown by @xcite and @xcite and the same result for both models can be read from @xcite by relation to a generalized plya urn , although the existing descriptions of the limits are not very explicit . a further related result from @xcite ( and
the only main result there having bearing on our work here ) is that for large @xmath5 , the distribution of @xmath10 is approximately geometric with parameter @xmath13 , with the error in the approximation going to zero as @xmath14 . thus it can be seen that if @xmath15 and @xmath14 , then the distribution of @xmath16 converges to a rate one exponential ; cf . proposition [ prop3](iii ) below . the primary tool we use here to characterize the limits and
obtain rates of convergence is a new distributional transformation for which the limit distributions are the unique fixed points .
this transformation allows us to develop a new variation of stein s method ; we refer to @xcite , @xcite and @xcite for introductions to stein s method . to formulate our main result we first define the family of densities @xmath17 where @xmath18 denotes the gamma function and @xmath19 denotes the confluent hypergeometric function of the second kind ( also known as the kummer @xmath20 function ) [ see @xcite , chapter 13 ] .
propositions [ prop1 ] and [ prop3 ] below imply that @xmath21 is indeed a density for @xmath22 and we denote by @xmath23 the distribution function defined by the density @xmath21 .
define the kolmogorov distance between two cumulative distribution functions @xmath24 and @xmath25 as=-1 @xmath26=0 [ thm1 ] let @xmath10 be the degree of vertex @xmath5 in a preferential attachment graph on @xmath27 vertices defined above and let @xmath28 . for model @xmath29 with @xmath30 and some constants @xmath31 independent of @xmath27 , @xmath32 for model @xmath33 with @xmath34 and some constants @xmath31 independent of @xmath27 , @xmath35 using proposition [ prop3 ] below we see an interesting difference in the behavior of the two models . in model 1 the limit distribution for the degree of the first vertex ( which by symmetry is the same as that for the second vertex ) is @xmath36 , the absolute value of a standard normal random variable , whereas in model 2 the limit distribution for the first vertex is @xmath37 , the square root of an exponential random variable . to ease exposition we present our upper bounds as rates , but the constants are recoverable ( although probably not practical especially for large @xmath5 )
theorem [ thm1 ] will follow from a more general result derived by developing stein s method for the distribution @xmath23 .
the key ingredient to our framework follows from observing that @xmath23 is a fixed point of a certain distributional transformation which we will refer to as the `` @xmath38-transformed double size bias '' ( @xmath38-tdsb ) transformation , which we now describe .
recall for a nonnegative random variable @xmath39 having finite mean we say @xmath40 has the _ size bias distribution _ of @xmath39 if @xmath41 for all @xmath42 such that @xmath43 [ see @xcite and @xcite for surveys and applications of size biasing ] . if in addition @xmath39 has finite second moment , then we will write @xmath44 to denote a random variable having the size bias distribution of @xmath40 .
alternatively we say @xmath44 has the _
double size bias distribution _ of @xmath39 and it is straightforward to check that @xmath45 although not used below , it is also appropriate to say that @xmath44 has the _ square bias distribution _ of @xmath39 since ( [ ] ) 2 implies that we are biasing @xmath39 against its square . this terminology is used in @xcite and @xcite albeit under a different notation .
now , we have the following key definition . [ def1 ] for fixed @xmath22 let @xmath46 and @xmath47 be two independent random variables uniformly distributed on the interval @xmath48 $ ] , and let @xmath49 be a bernoulli random variable with parameter @xmath50 independent of @xmath46 and @xmath47 .
define the random variable @xmath51 we say that @xmath52 has the _
@xmath38-transformed double size biased _
( @xmath38-tdsb ) distribution of @xmath39 , if @xmath53 where @xmath44 , the double size bias of @xmath39 , is assumed to be independent of @xmath54 .
our next result implies that the closer a distribution is to its @xmath38-tdsb transform , the closer it is to the @xmath23 distribution . besides the kolmogorov metric we also consider the wasserstein metric between two probability distribution functions @xmath24 and @xmath25 , defined as @xmath55 [ thm2 ] let @xmath39 be a nonnegative random variable with @xmath56 and let @xmath57 or @xmath58 .
let @xmath52 have the @xmath38-tdsb distribution of @xmath39 and be defined on the same probability space as @xmath39 .
then if @xmath57 , @xmath59 and , for any @xmath60 , @xmath61.\ ] ] if @xmath58 then @xmath62 and , for any @xmath60 , @xmath63.\ ] ] as can easily be read from the work of section [ sec1 ] ( in particular , propositions [ prop1 ] and [ prop3 ] ) , the distributions @xmath23 can roughly be partitioned into three regions where similar behavior within the range can be expected : @xmath58 , @xmath64 and @xmath57 .
the theorem only covers the first and last cases as this is what is needed to prove theorem [ thm1 ] .
analogs of the results of the theorem hold in the region @xmath64 , but we have omitted them for simplicity and brevity . from lemma [ lem12 ] below and the fact that for @xmath65 with bounded derivative @xmath66 we see that for @xmath57 or @xmath58 and all @xmath67 , @xmath68 choosing @xmath69 yields @xmath70 thus we can obtain bounds in the kolmogorov metric if @xmath71 is appropriately bounded with high probability or in expectation . [ rem1 ] it follows from lemmas [ lem5 ] and [ lem15 ] below that @xmath72 if and only if @xmath73 .
in the case that @xmath74 , @xmath54 is uniform on @xmath75 and proposition [ prop3 ] below implies that @xmath36 is distributed as the absolute value of a standard normal random variable .
thus we obtain the interesting fact that @xmath76 for @xmath20 uniform @xmath75 and independent of @xmath44 if and only if @xmath39 is distributed as the absolute value of a standard normal variable .
this fact can also be read from its analog for the standard normal distribution [ @xcite , proposition 2.3 ] : @xmath77 for @xmath78 uniform @xmath79 and independent of @xmath80 if and only if @xmath39 has the standard normal distribution [ see also @xcite ] .
although there are general formulations for developing stein s method machinery for a given distribution [ see @xcite ] , our framework below does not adhere to any of these directly since the characterizing operator we use is a second order differential operator [ see ( [ ] ) 23 and ( [ ] ) 25 below ] . for the distribution @xmath23 , the usual first order stein operator derived from the density approach of @xcite [ following @xcite ] is a complicated expression involving special functions .
however , by composing this more canonical operator with an appropriate first order operator , we are able to derive a second order stein operator ( see lemma [ lem7 ] below ) which has a form that is amenable to our analysis
. this strategy may be useful for other distributions which have first order operators that are difficult to handle .
the usual approach to developing stein s method is to decide on the distribution of interest , find a corresponding stein operator and then derive couplings from it .
the operator we use here was suggested by the @xmath38-tdsb transform which in turn arose from the discovery of a close coupling in the preferential attachment application .
we believe this approach of using couplings to suggest a stein operator is a potentially fruitful new strategy for extending stein s method to new distributions and applications . there have been several previous developments of stein s method using fixed points of distributional transformations .
@xcite develop stein s method using the zero - bias transformation for which the normal distribution is a fixed point .
letting @xmath20 be a uniform @xmath75 random variable independent of all else , @xcite and @xcite develop stein s method for the exponential distribution using the fact that @xmath39 and @xmath81 have the same distribution if and only if @xmath39 has an exponential distribution [ @xcite and @xcite also use this property ] .
we will show below that @xmath39 and @xmath82 have the same distribution if and only if @xmath39 is distributed as the absolute value of a standard normal random variable ( see also remark [ rem1 ] above ) . in this light
this paper can be viewed as extending the use of these types of distributional transformations in stein s method .
the layout of the remainder of the article is as follows . in section [ sec1 ]
we discuss various properties and alternative representations of @xmath23 , in section [ sec2 ] we develop stein s method for @xmath23 and prove theorem [ thm2 ] and in section [ sec3 ] we prove theorem [ thm1 ] by constructing the coupling needed to apply theorem [ thm2 ] and bounding the appropriate terms .
in this section we collect some facts about @xmath23 . recall the notation and definitions associated to the formula ( [ ] ) 1 for the density @xmath83 . from @xcite [ ( @xcite ) , chapter 13 ] , the kummer @xmath20 function ,
denoted @xmath19 , is the unique solution of the differential equation @xmath84 which satisfies ( [ ] ) 11 below .
the following lemma collects some facts about @xmath19 ; the right italic labeling of the formulas corresponds to the equation numbers from @xcite [ ( @xcite ) , chapter 13 ] , and the notation @xmath85 refers to the derivative with respect to @xmath86 .
[ lem1 ] let @xmath87 ; @xmath88 & & \hspace*{6pt}\hspace*{293pt}\textit{\aseq{2.5 } } \\[-8pt ] \nonumber & & \hspace*{6pt}\qquad u(a , b , z)=\frac{1}{\gamma ( a ) } \int _ 0^\infty e^{-zt}t^{a-1}(1+t)^{b - a-1 } \,dt , \\ & & \hspace*{6pt } u'(a , b , z ) = -au(a+1,b+1,z),\hspace*{104pt}\qquad\textit{\aseq{4.21}}\label{7 } \\ & & \hspace*{6pt}(1+a - b)u(a , b-1,z ) \nonumber \\[-8pt ] & & \hspace*{6pt}\hspace*{287pt}\textit{\aseq{4.24 } } \\[-8pt ] \nonumber & & \hspace*{6pt}\qquad=(1-b)u(a , b , z)-zu'(a , b , z ) , \label{8 } \\ & & \hspace*{6pt } u(a , b , z)-u'(a , b , z)=u(a , b+1,z ) , \hspace*{91pt}\quad\textit{\aseq{4.25}}\label{9 } \\ & & \hspace*{6pt}\qquad u(a-1,b-1,z ) = ( 1-b+z)u(a , b , z)-zu'(a , b , z),\hspace*{6.5pt}\textit{\aseq{4.27}}\label{10 } \\ & & \hspace*{6pt } u(a , b , z ) \sim z^{-a},\qquad(z{\to\infty } ) , \hspace*{112pt}\quad\qquad\textit{\aseq{5.2}}\label{11 } \\ & & \hspace*{6pt}\mbox{for $ a>- { \frac{1}{2}}$ } , \qquad u \bigl(a , { \tfrac{1}{2}},0 \bigr)=\gamma\bigl ( { \tfrac{1}{2 } } \bigr)/\gamma\bigl(a+ { \tfrac{1}{2 } } \bigr ) .
\hspace*{68pt}\quad\textit{\aseq{5.10}}\label{12}\end{aligned}\ ] ] as a direct consequence of ( [ ] ) 9 we have @xmath89 combining ( [ ] ) 7 and ( [ ] ) 11 with @xmath90 we find @xmath91 and using ( [ ] ) 5 with @xmath92 , @xmath93 and ( [ ] ) 14 implies that for @xmath94 , @xmath95 by comparing integrands in ( [ ] ) 6 , we also find the following fact .
[ lem2 ] let @xmath96 , @xmath97 and @xmath94 .
then @xmath98 the next results provide simpler representations for @xmath23 .
[ prop1 ] if @xmath99 and @xmath49 are two independent random variables having distributions @xmath100 where @xmath101 denotes the beta distribution , and @xmath102 where @xmath103 denotes the gamma distribution and @xmath104 the exponential distribution , then @xmath105 let @xmath106 and observe that by first conditioning on @xmath99 , we can express the density of @xmath107 as @xmath108 after making the change of variable @xmath109 in ( [ ] ) 16 , we find @xmath110 and now using ( [ ] ) 6 with @xmath111 and @xmath93 in the definition ( [ ] ) 1 of @xmath21 implies that @xmath112 . similarly ,
if @xmath113 , then we can express the density of @xmath114 as @xmath115 and after making the change of variable @xmath109 in ( [ ] ) 17 , we find @xmath116 where we have used ( [ ] ) 6 with @xmath117 and @xmath118 in the second equality .
applying ( [ ] ) 5 with @xmath111 and @xmath93 to this last expression implies @xmath119 .
the previous representations easily yield useful formulas for mellin transforms .
[ prop2 ] if @xmath120 with @xmath22 , then for all @xmath121 , @xmath122 for @xmath123 , we use proposition [ prop1 ] and well - known formulas for the mellin transforms of the beta and gamma distributions to find @xmath124 an application of the gamma duplication formula yields @xmath125 which combined with ( [ ] ) 19 implies ( [ ] ) 18 for the case @xmath123 .
the case @xmath58 follows from proposition [ prop3](i ) below which implies that if @xmath126 , then @xmath127 .
now ( [ ] ) 19 easily follows from well - known mellin transform formulas and thus ( [ ] ) 18 also follows . in a few special cases we can simplify and extend proposition [ prop1 ] .
below @xmath128 denotes the distribution function of @xmath23 .
[ prop3 ] we have the following special cases of @xmath23 : @xmath129 the identities ( i ) and ( ii ) are immediate from ( [ ] ) 15 and ( [ ] ) 14 , respectively . using stirling s formula for the gamma function to take the limit as @xmath130 for fixed @xmath131 in ( [ ] ) 18 yields the moments of @xmath132 which proves ( iii ) . as discussed below
, the preferential attachment model we study is a special case of a generalized plya triangular urn scheme as studied by @xcite .
the limiting distributions in his theorem 1.3@xmath133 with @xmath134 and @xmath135 include @xmath23 .
in fact , @xcite , example 3.1 , discusses these limits , but , with the exception of the case @xmath74 , it does not appear that the decomposition of proposition [ prop1 ] has previously been exposed . on the other hand , up to a scaling factor , the moment formula of @xcite , theorem 1.7 , simplifies to that of proposition [ prop2 ] for @xmath23 .
the distribution @xmath23 also appears in this urn context in section 9 of the survey article @xcite . additionally , if @xmath120 , then @xmath136 for @xmath22 , where @xmath137 is a _
dufresne law _ as defined in @xcite .
dufresne laws are essentially a generalization of products of independent beta and gamma random variables .
we now collect one more fact about @xmath23 , which will also prove useful in developing the stein s method framework below .
[ lem3 ] for every @xmath138 and @xmath139 , @xmath140 using the definition ( [ ] ) 1 of @xmath21 , making the change of variable @xmath141 and then applying ( [ ] ) 5 with @xmath111 and @xmath93 , ( [ ] ) 13 with @xmath117 and @xmath93 and then ( [ ] ) 11 with @xmath117 , we find @xmath142 so that @xmath143 first note that by applying ( [ ] ) 5 with @xmath111 and @xmath93 in the denominator of the final expression of ( [ ] ) 20 we have @xmath144 where the inequality follows by lemma [ lem2 ] .
now applying ( [ ] ) 5 to ( [ ] ) 20 both in the denominator as before and in the numerator with @xmath117 and @xmath93 , we find @xmath145 where again the inequality follows by lemma [ lem2 ] .
now applying lemma [ lem4 ] below to this last expression and combining with ( [ ] ) 21 yields the lemma .
[ lem4 ] if @xmath57 , then @xmath146 @xcite , theorem 1 , implies that @xmath147 is a decreasing function on @xmath148 , so that for @xmath57 , ( [ ] ) 22 is bounded above by @xmath149 .
moreover , stirling s formula implies @xmath150
in this section we develop stein s method for @xmath23 and prove theorem [ thm2 ] .
[ lem5 ] if @xmath151 for @xmath22 , then for every twice differentiable function @xmath42 with @xmath152 and such that @xmath153 , @xmath154 and @xmath155 are finite , we have @xmath156 let @xmath157 .
first note that @xmath158 using ( [ ] ) 11 and ( [ ] ) 7 with @xmath111 and @xmath93 we find that ( [ ] ) 24 equals @xmath159 where in the first equality we have used fubini s theorem [ justified by@xmath160 and the fact that @xmath161 .
we also have @xmath162 where in the second equality we have used fubini s theorem [ justified by @xmath163 and the fact that @xmath164 , and in the third we have used ( [ ] ) 10 with @xmath165 and @xmath118 .
hence , @xmath166 which proves the claim . for the sake of brevity , let @xmath167 . [ lem6 ] for all functions @xmath65 such that @xmath168 exists , the second order differential equation @xmath169 with initial conditions @xmath152 has solution @xmath170 \\[-8pt ] \nonumber & = & -\frac{1}{s}v_s(x ) \int_{0}^x \frac{1}{v_s(y ) \kappa_s(y)}\int_{y}^\infty\tilde{h}(z ) \kappa_s(z)\,dz\,dy,\end{aligned}\ ] ] where @xmath171 .
in order to prove lemma [ lem6 ] , we use the following intermediate result .
[ lem7 ] if @xmath172 and @xmath42 are functions such that @xmath173 and for @xmath174 , @xmath175 where @xmath176 then @xmath42 solves ( [ ] ) 25 and @xmath161 .
conversely , if @xmath42 is a solution to ( [ ] ) 25 with @xmath152 and @xmath177 , then @xmath178 and @xmath42 and @xmath172 satisfy @xmath179 .
assume @xmath42 and @xmath172 satisfy ( [ ] ) 27 and @xmath180 .
the fact that @xmath161 follows easily from the second equation of ( [ ] ) 27 . to show that ( [ ] ) 27 yields a solution to ( [ ] ) 25 , differentiate the second equality in ( [ ] ) 27 and combine the resulting equations to obtain @xmath181 hence , we only need to show that @xmath182 in order to simplify the calculations , let us introduce @xmath183 note that @xmath184 . with this and @xmath185 , ( [ ] ) 29
becomes @xmath186 the left - hand side of ( [ ] ) 30 is equal to @xmath187 where we have used ( [ ] ) 7 with @xmath111 and @xmath93 to handle the derivatives in the first equality and then ( [ ] ) 10 with @xmath165 and @xmath118 in the second .
hence , ( [ ] ) 29 holds , as desired . if @xmath42 is a solution to ( [ ] ) 25 with @xmath152 and @xmath177 , then obviously @xmath178 and the second assertion of the lemma follows from the previous calculations .
proof of lemma [ lem6 ] lemma [ lem7 ] implies that we only need to solve ( [ ] ) 27 .
note first that the general differential equation @xmath188 has solution @xmath189 hence , noticing that @xmath190 the solution to the first equation in ( [ ] ) 27 is @xmath191 whereas the solution to the second equation in ( [ ] ) 27 is @xmath192 which is the first identity of ( [ 26 ] ) ; the second follows by observing that @xmath193 . before developing the stein s method machinery
further we need two more lemmas , the first of which is well known and easily read from @xcite .
[ lem8 ] for @xmath194 , @xmath195 [ lem9 ] if @xmath196 is defined by ( [ ] ) 28 , then for @xmath57 and @xmath174 @xmath197 to prove the first assertion note that ( [ ] ) 7 with @xmath111 and @xmath93 followed by ( [ ] ) 5 with @xmath117 and @xmath93 and lemma [ lem2 ] imply @xmath198 the claimed upper bound now follows from lemma [ lem4 ] .
the lower bound follows from the final expression of ( [ ] ) 32 , since for @xmath106 , the integral representation ( [ ] ) 6 implies all terms in the quotient are nonnegative , and for @xmath74 , ( [ ] ) 14 implies @xmath199 . for the second assertion , we use ( [ ] ) 8 with @xmath111 and @xmath93 in the second equality below to find @xmath200 \\[-8pt ]
\nonumber & = & 2\biggl(s- { \frac{1}{2}}\biggr)\frac{u ( s-1,- { { 1}/{2 } } , { { x^2}/{(2s ) } } ) } { u ( s-1 , { { 1}/{2 } } , { { x^2}/{(2s ) } } ) } -1.\end{aligned}\ ] ] applying lemma [ lem2 ] to ( [ ] ) 33 proves the remaining upper bound .
the second lower bound follows from the first .
[ lem10 ] if @xmath172 satisfies the first equation of ( [ ] ) 27 with @xmath178 , then @xmath201 * if @xmath65 is nonnegative and bounded , then for all @xmath174 and @xmath57 , @xmath202 * if @xmath65 is absolutely continuous with bounded derivative , then for all @xmath57 @xmath203 the first assertion is a restatement of ( [ ] ) 31 , recorded in this lemma for convenient future reference .
if @xmath204 for @xmath138 with @xmath205 , then for all @xmath57 and @xmath174 , @xmath206 where we have used lemma [ lem3 ] ; this shows ( [ ] ) 34 . let @xmath65 be absolutely continuous with @xmath207 , and without loss of generality assume that @xmath208 so that for @xmath138 , @xmath209 . in particular ,
if @xmath210 , then @xmath211 and noting that @xmath212 ( using proposition [ prop2 ] ) , we can apply lemma [ lem3 ] to find that for @xmath174 , @xmath213 to bound the integral in this last expression , we make the change of variable @xmath141 and apply ( [ ] ) 13 with @xmath111 , @xmath214 and ( [ ] ) 11 with @xmath111 to find @xmath215 where the last inequality follows from lemma [ lem2 ] .
[ lem11 ] let @xmath42 be defined as in ( [ ] ) 26 with @xmath152 .
* if @xmath65 is nonnegative and bounded and @xmath57 , then @xmath216 * if @xmath65 is nonnegative , bounded and absolutely continuous with bounded derivative and @xmath57 , then @xmath217 if @xmath58 , then @xmath218 * if @xmath65 is absolutely continuous with bounded derivative and @xmath57 , then @xmath219 if @xmath58 , then @xmath220 from ( [ ] ) 26 of lemma [ lem6 ]
we have that @xmath221 where @xmath172 is as in lemma [ lem10 ] .
if either @xmath65 is bounded or absolutely continuous with bounded derivative , then recall that lemma [ lem10 ] implies @xmath172 is bounded . if @xmath57 , then ( [ ] ) 7 and ( [ ] ) 11 with @xmath111 and @xmath93 imply that @xmath222 is nonincreasing and positive for positive @xmath223 , so that @xmath224 now , again by ( [ ] ) 27 , we have @xmath225 where we have used ( [ ] ) 41 in the first inequality and lemma [ lem9 ] in the second . applying the bound ( [ ] ) 34 proves ( [ ] ) 36 . to bound @xmath226 for @xmath65 having @xmath207 ,
let @xmath227 and differentiate ( [ ] ) 25 to find @xmath228 which implies @xmath229 integrating , we obtain @xmath230 so that lemma [ lem8 ] yields @xmath231 \\[-8pt ] \nonumber & & { } + \frac{1}{s}\exp\biggl ( { \frac{x^2}{2s } } \biggr)\int _ x^\infty\exp\biggl ( { \frac{-y^2}{2s } } \biggr)h'(y)\,dy.\end{aligned}\ ] ] if @xmath205 , then an integration by parts yields a bound on the second term of ( [ ] ) 44 which yields @xmath232 if @xmath57 , then apply the bound ( [ ] ) 36 above on @xmath233 to find ( [ ] ) 37 ; for @xmath58 , ( [ ] ) 38 follows immediately .
now , we can apply lemma [ lem8 ] directly to ( [ ] ) 44 to find @xmath234 finally , ( [ ] ) 43 implies @xmath235 the first term can be bounded by ( [ ] ) 45 , and if @xmath57 , a subsequent application of ( [ ] ) 42 on @xmath233 and then ( [ ] ) 35 on @xmath236 yields ( [ ] ) 39 . if @xmath58 , then ( [ ] ) 40 follows from ( [ ] ) 46 and ( [ ] ) 45 . in order to obtain the bounds for the kolmogorov metric
, we need to introduce the smoothed half - line indicator function @xmath237 \,dt.\ ] ] [ lem12 ] if @xmath120 and @xmath39 is a nonnegative random variable and @xmath57 , then , for all @xmath67 , @xmath238 if @xmath58 , then , for all @xmath67 , @xmath239 the lemma follows from a well - known argument and the following bounds on the density @xmath83 defined by ( [ ] ) 1 .
if @xmath57 , then by ( [ ] ) 7 with @xmath111 and ( [ ] ) 6 with @xmath165 , @xmath83 is nonincreasing in @xmath223 and from ( [ ] ) 12 with @xmath111 , @xmath240 where the inequality is by lemma [ lem4 ] . if @xmath58 , then @xmath241 which has maximum @xmath242 . we will also need the following `` indirect '' concentration inequality ; it follows from the arguments of the proof of lemma [ lem12 ] immediately above .
[ lem13 ] if @xmath120 and @xmath39 is a nonnegative random variable and @xmath57 , then , for all @xmath243 , @xmath244 if @xmath58 , then , for all @xmath243 , @xmath245 [ lem14 ] if @xmath42 satisfies ( [ ] ) 25 for @xmath246 and @xmath57 , then for @xmath138 , @xmath247\,du.\end{aligned}\ ] ] if @xmath58 , then for @xmath138 , @xmath248\,du.\ ] ] using ( [ ] ) 25 , we obtain @xmath249 hence , @xmath250\,du.\end{aligned}\ ] ] applying the bounds of lemma [ lem11 ] yields the claim .
[ lem15 ] let @xmath39 be a nonnegative random variable with @xmath56 and let @xmath52 be the @xmath38-tdsb of @xmath39 as in definition @xmath251 for some @xmath22 .
for every twice differentiable function @xmath42 with @xmath152 and such that the expectations below are well defined , we have @xmath252 the lemma will follow from two facts : * if @xmath44 has the double size bias distribution of @xmath39 , then for all @xmath172 with @xmath253 , @xmath254 * if @xmath172 is a function such that @xmath255 and for @xmath54 as defined in definition [ def1 ] , @xmath256 , then @xmath257 the first item above is easy to verify from the definition of the size bias distribution and the fact that @xmath258 , and the second follows from a simple calculation after noting that @xmath54 has density @xmath259 for @xmath260 . by conditioning on @xmath44 and using the second fact above for @xmath261 , we find @xmath262 and applying the first fact above proves the lemma .
proof of wasserstein bound of theorem [ thm2 ] making use of lemma [ lem6 ] and lemma [ lem15 ] , we have @xmath263 where @xmath42 is given by ( [ ] ) 26 . if @xmath65 is lipschitz continuous , then @xmath42 is three times differentiable almost everywhere and we have @xmath264 we now obtain ( [ ] ) 3 by invoking ( [ ] ) 39 and ( [ ] ) 40 of lemma [ lem11 ] .
proof of kolmogorov bound of theorem [ thm2 ] fix @xmath265 and let @xmath67 , to be chosen later .
let @xmath42 be as in ( [ ] ) 26 with @xmath266 replaced by @xmath267 , where @xmath268 is defined by ( [ ] ) 47 .
define the indicator random variable @xmath269 $ ] .
now , @xmath270 if @xmath57 , using ( [ ] ) 37 from lemma [ lem11 ] implies @xmath271 applying lemma [ lem14 ] , @xmath272 noticing that @xmath273 and applying lemma [ lem13 ] to the integrand , @xmath274 where @xmath275 . from lemma [ lem12 ]
, we have @xmath276 choosing @xmath277 and solving for @xmath278 , @xmath279 which yields ( [ ] ) 4 . a nearly identical argument yields the statement for @xmath58 .
we first reformulate theorem [ thm1 ] in terms of a generalized plya urn model .
an urn initially contains @xmath5 black balls and @xmath280 white balls and at each step a ball is drawn .
if the ball drawn is black , it is returned to the urn along with an additional @xmath281 black balls and @xmath282 white balls ; if the ball drawn is white , the ball is returned to the urn along with an additional @xmath283 black balls and @xmath278 white balls .
we use the notation @xmath284 to denote the distribution of the number of white balls in the urn after @xmath27 draws and replacements .
for example , @xmath285 has a single point mass at @xmath280 and also note that @xmath286 corresponds to the classical plya urn model .
[ th1a ] let @xmath287 and @xmath288 be integers and @xmath289 .
if @xmath290 , then , for some constants @xmath291 independent of @xmath27 , @xmath292 theorem [ thm1 ] follows immediately from theorem [ thm1](a ) after noting that for model 1 with @xmath293 , the degree of vertex @xmath5 in @xmath1 , the graph with @xmath27 vertices and @xmath294 edges , has distribution @xmath295 ; this is because the degree of vertex @xmath5 in @xmath296 is @xmath29 and the sum of the degrees of the remaining vertices is @xmath297 ( since @xmath296 has @xmath298 edges ) . for model 2 with @xmath299 , the degree of vertex @xmath5 in @xmath1 , the graph with @xmath27 vertices and @xmath27 edges , has distribution @xmath300 ; this is because the sum of the degrees of @xmath301 is @xmath302 and vertex @xmath5 has probability @xmath303 of self - attachment when forming @xmath296 from @xmath301 .
the lower bound of the theorem follows from the following general result combined with the fact from lemma [ lem21 ] below that @xmath304 .
let @xmath305 be a probability distribution with a density @xmath42 such that for all @xmath223 in some interval @xmath306 , @xmath307 .
if @xmath308 is a sequence of integer - valued random variables and @xmath309 is a sequence of nonnegative numbers tending to zero , then @xmath310 for some positive constant @xmath311 independent of @xmath27 .
let @xmath312 be the distribution function of @xmath305 and note that the hypothesis on the density @xmath42 implies that if @xmath313 , then @xmath314 since @xmath315 , there exists @xmath316 such that for all @xmath317 , there is an integer @xmath318 such that @xmath319\subset(a , b)$ ] . from ( [ ] ) 48 ,
for @xmath317 we have @xmath320 and now using the continuity of @xmath312 on @xmath306 and the fact that the distribution function @xmath1 of @xmath321 is constant on the interval @xmath322 , it follows that for @xmath317 , @xmath323 since @xmath1 is the distribution function of a discrete random variable and @xmath312 is continuous , it follows that @xmath324 for all @xmath325 ( and in particular @xmath326 ) , so that we may choose @xmath327 . as mentioned in the we write our results as rates , but the constants are recoverable . for the sake of clarity
, we have not been careful to optimize the bounds in our arguments , but it is clear that sharper statements can be read from the proofs below .
for example , the constant in both the lower bound and upper bounds of theorem [ thm1](a ) depend crucially on the scaling factor @xmath328 . for our purposes
lemma [ lem21 ] below is acceptable , but note that exact results are available [ see ( [ ] ) 70 and ( [ ] ) 71 in the proof of lemma [ lem21 ] ] .
now let @xmath329 have distribution @xmath330
. we will use ( [ ] ) 4 to prove the upper bounds of theorem [ thm1](a ) and so we will show that there is a close coupling of @xmath39 and @xmath331 , where @xmath54 is as in definition [ def1 ] with @xmath332 . this result will follow from the following lemmas proved at the end of this section .
[ lem16 ] there is a coupling @xmath333 of @xmath334 and the double size bias distribution of @xmath335 satisfying @xmath336 [ lem17 ] the distribution @xmath337 can be expressed as a mixture of the distributions @xmath338 and @xmath339 with respective probabilities @xmath340 and @xmath341 . in the next lemma
we use the notation @xmath342 for a nonnegative integer - valued random variable @xmath316 to denote a mixture of the distributions @xmath343 for @xmath344 that are mixed with respective probabilities @xmath345 for @xmath346 [ lem18 ] let @xmath347 , let @xmath348 and let @xmath349
. then @xmath350 [ lem19 ] let @xmath46 and @xmath47 be uniform @xmath75 random variables , independent of each other and of @xmath351 , defined as in lemma @xmath352 . then there exist random variables @xmath353 with distribution @xmath354 and @xmath355 with distribution @xmath356 such that @xmath357 from these lemmas we can now prove theorem [ thm1](a )
; here and below we use @xmath358 to denote a generic constant that may differ from line to line .
proof of theorem [ th1a ] let @xmath359 and @xmath360 , let @xmath333 be defined as in lemma [ lem16 ] above and , as per definition [ def1 ] , let @xmath49be a bernoulli(@xmath340 ) random variable and @xmath361 , where @xmath46 and @xmath47 are independent uniform @xmath75 variables independent of @xmath49 .
lemmas [ lem17 ] , [ lem18 ] and [ lem19 ] imply that we can couple @xmath39 and @xmath362 together so that @xmath363 almost surely .
thus , using lemma [ lem16 ] , @xmath364 and recalling that @xmath331 has the @xmath38-tdsb distribution , the theorem follows from ( [ ] ) 4 taking @xmath365 , noting that for @xmath327 , @xmath366 and using @xmath367 from lemma [ lem21 ] below .
we have left to prove lemmas [ lem16][lem19 ] and [ lem21 ] ; lemma [ lem17 ] is immediate after considering the urn process corresponding to @xmath368 and conditioning on the color of the first ball drawn , which is white with probability @xmath340 .
proof of lemma [ lem18 ] consider an urn with @xmath5 green balls , 1 black ball and 2 white balls .
a ball is drawn at random and replaced in the urn along with another ball of the same color plus an additional green ball .
if @xmath99 is the number of times a nongreen ball is drawn in @xmath294 draws , the number of white balls in the urn after @xmath294 draws is distributed as @xmath369 . since @xmath370 is distributed as @xmath371 ( which is also that of @xmath351 ) and the number of white balls in the urn after @xmath294 draws has distribution @xmath338 , the first equation follows .
the second equation follows from similar considerations .
proof of lemma [ lem19 ] we will show that for @xmath46 and @xmath47 independent uniform @xmath75 random variables , there exist random variables @xmath316 and @xmath372 such that @xmath373 , @xmath374 and @xmath375 the lemma follows from these `` conditional '' almost sure statements after noting that @xmath376 since we can think of @xmath377 as the number of black balls in the @xmath378 urn
. the formulas of @xcite [ ( @xcite ) , page 206 ] imply that @xmath378 has distribution function @xmath379 and it is straightforward to verify that @xmath380 has the same distribution .
we find @xmath381 and thus a coupling satisfying the first claim above .
defining @xmath382 ( [ ] ) 49 implies @xmath374 and @xmath383 . before proving lemma [ lem16 ]
, we provide a useful construction for the double size bias distribution of a sum of indicators .
[ lem20 ] let @xmath384 , where the @xmath385 are bernoulli random variables and @xmath386 .
for each @xmath387 , let @xmath388 have the distribution of @xmath389 conditional on @xmath390 and let @xmath391 and @xmath392 be random variables independent of the variables above satisfying @xmath393 then , @xmath394\ ] ] has the double size bias distribution of @xmath39 .
we have @xmath395 \biggr ) \\ & = & b^{-2}\sum_{j , k}{\mathbb{e}}(x_jx_k ) { \mathbb{e}}\bigl\ { f ( w ) |
x_j = x_k=1\bigr\ } \\ & = & b^{-2}\sum_{j , k}{\mathbb{e}}\bigl \{x_jx_k f ( w ) \bigr\}=b^{-2}{\mathbb{e}}\bigl \{w^2 f ( w ) \bigr\ } ; \end{aligned}\ ] ] this is exactly ( [ ] ) 2 . to simplify the notation we consider @xmath5 fixed in what follows .
we write @xmath396 where for @xmath397 , @xmath398 is the indicator that a white ball is drawn on draw @xmath280 from the @xmath399 urn and @xmath400 to represent the initial white ball in the urn .
we will then define random variables @xmath401 such that @xmath402 so that by lemma [ lem20 ] , if @xmath391 and @xmath392 are random variables independent of @xmath401 satisfying @xmath403 for @xmath386 , then @xmath404 has the double size bias distribution of @xmath39 . in order to generate a variable satisfying ( [ ] ) 50 for @xmath405 , we use the following lemma that yields a method to construct an urn process having the law of the @xmath399 urn process up to time @xmath27 conditional on @xmath406 .
this conditioned process follows the law of the @xmath407 urn process up to ( and including ) draw @xmath408 . at draw @xmath280 , exactly one _ black _ ball is added and then draws @xmath409 through @xmath410 follow the @xmath411 urn law .
again at draw @xmath412 exactly one _ black _ ball is added and then the process continues to draw @xmath27 following the @xmath411 urn rule .
we write @xmath401 to denote the number of white balls in the urn after @xmath27 draws in this process , and we refer to this process as the @xmath413 process .
our next main result shows that this construction of @xmath401 has the distribution specified in ( [ ] ) 50 .
first we state a technical lemma ; the proof can be found at the end of this section .
[ lem21 ] fix @xmath414 and let @xmath415 where for @xmath397 , @xmath398 is the indicator that a white ball is drawn on draw @xmath280 from the @xmath399 urn and @xmath400 . if @xmath416 , then @xmath417 and if @xmath418 , @xmath419 for all @xmath420 , @xmath421 if @xmath422 , then for some constant @xmath358 independent of @xmath423 and @xmath5 , @xmath424 let @xmath425 and @xmath426 , @xmath427 and @xmath428 be defined as in lemma @xmath429 and the preceding two paragraphs .
then @xmath430 let @xmath431 , where for @xmath432 , @xmath433 is the indicator that draw @xmath423 in the @xmath413 urn process is white and @xmath434 . from the definition of the process , for @xmath435 , @xmath436 and for @xmath437 , since @xmath438 , @xmath439 note
also that @xmath440 , @xmath434 and draw @xmath441 in the @xmath413 urn process follows the @xmath411 urn law . now comparing ( [ ] ) 55 to ( [ ] ) 53 and ( [ ] ) 56
to ( [ ] ) 52 , we find the sequential conditional probabilities agree and so the lemma follows .
we are now ready to prove lemma [ lem16 ] , and we first give the following remark about the argument .
the @xmath442 process and the @xmath413 process defined above differ only in that , in the latter process , after each of draws @xmath280 and @xmath412 a single black ball is added into the urn regardless of what is drawn ; in the former process , the two balls added to the urn in these draws depend on the color drawn .
this difference turns out to be small enough to allow a close coupling as stated in the lemma .
proof of lemma [ lem16 ] for each @xmath423 we construct @xmath443 to , respectively , denote the indicator for the event that a white ball is added to the urn after draw number @xmath423 for the @xmath442 process and for the @xmath413 process , and we write @xmath444 to denote the number of white balls in the urn after draw @xmath294 and @xmath27 , respectively , for each process .
let @xmath445 be independent uniform @xmath75 random variables .
we define @xmath446\ ] ] and for @xmath447 we define @xmath448-{{\mathrm{i}}}[t > k ] } \biggr].\ ] ] we also set @xmath449 since a single black ball is added after draws @xmath280 and @xmath412 . writing the event @xmath450 as a union of the events that index @xmath423 is the least index such that @xmath451 , and also using that @xmath449 , we find that for @xmath452 , @xmath453 from ( [ ] ) 57 , @xmath454 , so that we find @xmath455 where we have used ( [ ] ) 54 .
defining @xmath391 and @xmath392 as in ( [ ] ) 51 we now have @xmath456 \\[-8pt ] \nonumber & & \qquad\leq{\mathbb{p}}(j=0)+{\mathbb{p}}(k=0)+ { \mathbb{p}}(j = k)+\frac{2c}{b^2 } \sum _ { j <
k}j^{- { { 1}/{2}}}{\mathbb{e}}x_jx_k.\end{aligned}\ ] ] since @xmath457 , @xmath458 and using ( [ ] ) 51 , we have @xmath459 and now using the bounds on @xmath460 , @xmath461 and @xmath462 from lemma [ lem21 ] , we find that ( [ ] ) 58 is bounded above by @xmath463 proof of lemma [ lem21 ] let @xmath464 . by the definition of conditional probability , @xmath465 and @xmath466 and we
next will calculate the probabilities above . for @xmath397 , we have @xmath467 which implies that for @xmath468 , @xmath469 and @xmath470 now to compute the conditional expectations appearing above note first that @xmath471 conditioning on @xmath472 and taking expectations yields @xmath473 and then iterating and substituting @xmath410 for @xmath412 yields @xmath474 using ( [ ] ) 65 with @xmath280 substituted for @xmath408 , we also find @xmath475 note here that conditioning on @xmath476 and @xmath472 is equivalent to conditioning on @xmath477 and the event @xmath478 .
we use a similar approach to obtain @xmath479 which can then be added to ( [ ] ) 64 while letting @xmath480 to obtain @xmath481 and thus @xmath482 iterating and substituting @xmath410 for @xmath412 gives @xmath483 and using ( [ ] ) 67 with @xmath280 substituted for @xmath408 , we also find @xmath484 letting @xmath485 and applying ( [ ] ) 61 , ( [ ] ) 63 , ( [ ] ) 66 and ( [ ] ) 67 we have @xmath486 and by substituting @xmath487 for @xmath488 in ( [ ] ) 68 , we also find @xmath489 substituting ( [ ] ) 61([])63 , ( [ ] ) 65 , ( [ ] ) 66 , ( [ ] ) 68 and ( [ ] ) 69 appropriately into ( [ ] ) 59 and ( [ ] ) 60 proves ( [ ] ) 52 and ( [ ] ) 53 .
from ( [ ] ) 65 we have @xmath490 and from ( [ ] ) 61 we find @xmath491 now using ( [ ] ) 63 and ( [ ] ) 66 yields @xmath492 and using ( [ ] ) 67 we find @xmath493 lemma [ lem4 ] applied to ( [ ] ) 70 implies @xmath494 and collecting the appropriate facts above yields the bounds on @xmath461 , @xmath462 , @xmath460 and @xmath495 . for the bound on @xmath496 , an argument similar to ( [ ] )
65 leading to ( [ ] ) 70 yields that @xmath497 which can be bounded by lemma [ lem4 ] resulting in inequalities which are similar to ( [ ] ) 72 .
we are grateful to carl morris for valuable advice on this project and an anonymous referee for very helpful and detailed comments .
the authors have also learned that robert gaunt ( university of oxford ) has independently been working on a second order stein operator for the variance - gamma distribution , but there appears to be little overlap between his work and ours .
|
we provide optimal rates of convergence to the asymptotic distribution of the ( properly scaled ) degree of a fixed vertex in two preferential attachment random graph models .
our approach is to show that these distributions are unique fixed points of certain distributional transformations which allows us to obtain rates of convergence using a new variation of stein s method . despite the large literature on these models ,
there is surprisingly little known about the limiting distributions so we also provide some properties and new representations , including an explicit expression for the densities in terms of the confluent hypergeometric function of the second kind . ,
|
the idea that quantum gravity can appear at the tev energy scale well below the planck mass @xmath9gev was proposed in the 1990s@xcite .
the large extra dimensions(led ) model@xcite introduced by arkani - hamed , dimopoulos and davli has attracted much attention .
it has been emphasized that the presence of large extra dimensions brings a new solution to the hierarchy problem , which can take the place of other mechanisms , for example , low - energy supersymmetry .
however , it also interesting to examine a scenario which combines new physics beyond the standard model(sm ) such as the 2hdm@xcite and led .
this new possibility leads to different phenomenology than the usual led scenario , which we explore here . in this
extended led scenario , as in the usual led scheme , the total space - time has @xmath10 dimensions .
the sm and new particles beyond the sm live in the usual @xmath11dimensional space , while gravity can propagate in the additional @xmath12-dimensional space , which is assumed for simplicity to be compactified on the @xmath12-dimensional torus @xmath13 with a common radius r. then the 4-dimensional planck scale @xmath14 is no longer the relevant scale but is related to the fundamental scale @xmath15 as follows@xcite : @xmath16 where @xmath17 .
according to eq .
( [ scale ] ) , deviations from the usual newtonian gravitational force law can be expected at distances smaller than @xmath18@xcite . for @xmath19 , led is consistent with the current experiments@xcite since gravitational forces are not yet well probed at distances less than sbout a millimeter ( however for @xmath20 , there are constraints arising from , e.g. , supernova cooling , which require @xmath21 if @xmath20 @xcite ) .
2hdm led can be tested at future high energy colliders . in 2hdm
led , just as in led , there exist kk towers of massive spin-2 gravitons and scalars which can interact with the sm and beyond sm fields .
there are two classes of effects that can probe led : real graviton emission and virtual kk tower exchange . at future linear colliders
the search for one or more higgs bosons will be a central task . in the sm ,
the higgs boson mass is a free parameter with an upper bound of @xmath22 800gev @xcite . beyond the sm ,
the 2hdm is of particular theoretical interest , in which the two complex higgs doublets correspond to eight scalar states and spontanoues symmetry breaking leads to five physical higgs bosons : two neutral cp - even bosons @xmath23 and @xmath24 , one neutral cp - odd boson @xmath25 , and two charged bosons @xmath26 .
the @xmath23 is the lightest and is a sm - like higgs boson especially in the decoupling region ( @xmath27 ) .
the other four are not sm - like and their discovery , particularly the charged higgs bosons , would provide evidence for the 2hdm .
if the @xmath26 bosons have mass @xmath28 , they will be produced mainly through the @xmath29 decays of top quarks , which can be produced singly or in pairs at an @xmath30 lc @xcite .
if there is sufficient center of mass energy available , @xmath31 , then charged higgs pair production , @xmath0 , will be the dominant production mechanism@xcite .
however , if @xmath32 , then @xmath26 bosons can only be produced singly .
and @xmath1@xcite is one of the most important single charged higgs production processes that can also be used to measure the relevant yukawa couplings .
therefore , for this process , we will focus on the regime where @xmath33 . for @xmath34 gev ( 1000 gev ) , this implies that @xmath35 gev ( @xmath36 gev ) .
there are several kinds of such 2hdm models . in the model
called type i , one higgs doublet provides masses for both the up - type and down - type quarks . in the type ii model
, one higgs doublet gives masses to the up - type quarks and the other one to the down - type quarks . in the type iii model ,
both doublets contribute to generate the masses for up - type and down - type quarks . in this paper
, we will concentrate on the type ii 2hdm(2hdm ii ) , which are is favored by the minimal supersymmetric standard model ( mssm)@xcite .
we will also briefly discussed the results in the type i 2hdm . in the following we consider the the contributions of virtual kk gravitons to the 2hdm charged higgs production , especially in the two important production processes @xmath0 and @xmath1 at future linear colliders .
the presentation is organized as follows : in sect .
ii we present the calculations . in sect .
iii we give the numerical results and discuss them .
iv contains a brief conclusion .
in this section we derive the cross section for 2hdm charged higgs production .
the 2hdm diagrams , including the additional virtual kk gravitons(@xmath37 ) , which contribute to the processes @xmath0 and @xmath4 are presented in figs . 1 and 2 , respectively .
in the four dimensional description the interaction lagrangian between the scattering fields and the kk gravitons ( @xmath38 ) or kk scalars ( @xmath39 ) is given by @xcite @xmath40 where @xmath41 with @xmath42 s being integers , @xmath43gev is the reduced four dimensional planck scale , and @xmath44 is the energy - momentum tensor of the scattering fields .
the @xmath45-th kk mode graviton and scaler masses squared are both characterized by @xmath46 . since the trace of the energy - momentum tensor is proportional to the mass of the fields due to the field equations , we neglect processes mediated by the kk scalars in our study for the linear @xmath47 collider , and consider only the processes mediated by the kk gravitons . from eq
.([il ] ) we can derive the relevant feynman rules to be used in our calculations , which can be found in ref .
the numerator of the graviton propagator @xmath48 in the unitary gauge@xcite , is given by : @xmath49where @xmath50 is the minkowski metric .
the dots represent terms proportional to the graviton momentum @xmath51 , and since @xmath52 , give a vanishing contribution to the amplitude .
note that the the numerator of the graviton propagator in ref .
@xcite is twice eq .
( [ prop ] ) and is the same as presented in ref .
however , the coupling constant squared in ref .
@xcite , i.e. , @xmath53 in eq .
( [ il ] ) is twice the one in ref .
hence the results in the two references are consistent and we have been careful of this issue in our calculations .
we can write the amplitudes for the feynman diagrams shown in figs.1 and 2 as follows : @xmath54u(p_2 ) \nonumber \\
{ \cal m}_{1b}&= & \frac{i{\cal g}p^{\mu\nu\alpha\beta}}{4}{\chi}^{a}_{\alpha\beta}{\bar v(p_1)}c_{\mu\nu } u(p_2)\nonumber \\ % -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-= { \cal m}_{2a}&= & \frac{-ie^2g}{3\sqrt{2}m_ws[(k_1+k_3)^2-m_b^2]}{\bar v(p_1)}\gamma_{\mu}u(p_2){\bar u(k_1,m_t)}\bigg [ m_b\tan\beta p_r+m_t\cot\beta p_l \bigg ] \nonumber\\ \hspace{1.0 cm } & & ( \not{\!k}_1+\not{\!k}_3+m_b)\gamma^{\mu}v(k_2,m_b ) + \nonumber\\ \hspace{1.0 cm } & & \frac{-ie^2g}{\sqrt{2}m_w(s - m^2_z)c^2_ws^2_w[(k_1+k_3)^2-m_b^2]}{\bar v(p_1)}\gamma_{\mu}\big [ ( \frac{1}{2}-s^2_w)p_l -s^2_wp_r\big]u(p_2).{\bar u(k_1,m_t)}\nonumber\\ \hspace{1.0 cm } & & \bigg [ m_b\tan\beta p_r+m_t\cot\beta p_l \bigg ] ( \not{\!k}_1+\not{\!k}_3+m_b)\gamma^{\mu}[(\frac{1}{2}-\frac{1}{3}s^2_w)p_l -\frac{1}{3}s^2_wp_r\big ] v(k_2,m_b)\nonumber\\ % -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-= % -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-= { \cal m}_{2b}&= & \frac{-2ie^2g}{3\sqrt{2}m_ws[(k_2+k_3)^2-m_t^2]}{\bar v(p_1)}\gamma_{\mu}u(p_2){\bar u(k_1,m_t)}\gamma^{\mu}(\not{\!k}_2+\not{\!k}_3-m_t ) \nonumber\\ \hspace{1.0 cm } & & \bigg [ m_b\tan\beta p_r+m_t\cot\beta p_l \bigg ] v(k_2,m_b ) + \nonumber\\ \hspace{1.0 cm } & & \frac{ie^2g}{\sqrt{2}m_w(s - m^2_z)c^2_ws^2_w[(k_2+k_3)^2-m_t^2]}{\bar v(p_1)}\gamma_{\mu}\big [ ( \frac{1}{2}-s^2_w)p_l -s^2_wp_r\big]u(p_2).{\bar u(k_1,m_t)}\nonumber\\ \hspace{1.0 cm } & & \gamma^{\mu}[-(\frac{1}{2}-\frac{2}{3}s^2_w)p_l + \frac{2}{3}s^2_wp_r\big](\not{\!k}_2+\not{\!k}_3-m_t)\bigg [ m_b\tan\beta p_r+m_t\cot\beta p_l \bigg ] v(k_2,m_b ) \nonumber\\ % -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-= % -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-= { \cal m}_{2c}&= & \frac{-ie^2g}{\sqrt{2}m_ws[(k_1+k_2)^2-m_{h^-}^2]}{\bar v(p_1)}(2\not{\!k}_3-\not{\!p}_1-\not{\!p}_2)u(p_2)\nonumber\\ \hspace{1.0 cm } & & { \bar u(k_1,m_t)}\bigg [ m_b\tan\beta p_r+m_t\cot\beta p_l \bigg ] v(k_2,m_b ) + \nonumber\\ \hspace{1.0 cm } & & \frac{-ie^2g(c^2_w - s^2_w)}{2\sqrt{2}m_w(s - m^2_z)c^2_ws^2_w[(k_1+k_2)^2-m_{h^-}^2]}{\bar v(p_1)}(2\not{\!k}_3-\not{\!p}_1-\not{\!p}_2)[(\frac{1}{2}-s^2_w)p_l -s^2_wp_r\big]u(p_2)\nonumber\\ \hspace{1.0 cm } & & { \bar u(k_1,m_t)}\bigg [ m_b\tan\beta p_r+m_t\cot\beta p_l \bigg ] v(k_2,m_b)\nonumber\end{aligned}\ ] ] @xmath55 } { \bar v(p_1)}c_{\mu\nu}u(p_2)\nonumber\\ \hspace{1.0 cm } & & { \bar u(k_1,m_t)}\bigg [ m_b\tan\beta p_r+m_t\cot\beta p_l \bigg ] ( \not{\!k}_1+\not{\!k}_3+m_b){\chi}^{b}_{\alpha\beta}v(k_2,m_b)\nonumber \\ % -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-= { \cal m}_{2e}&= & \frac{ig{\cal g}p^{\mu\nu\alpha\beta}}{16\sqrt{2}m_w[(k_2+k_3)^2-m_{t}^2 ] } { \bar v(p_1)}c_{\mu\nu}u(p_2)\nonumber\\ \hspace{1.0 cm } & & { \bar u(k_1,m_t)}{\chi}^{c}_{\alpha\beta}(\not{\!k}_2+\not{\!k}_3-m_t)\bigg [ m_b\tan\beta p_r+m_t\cot\beta p_l \bigg ] v(k_2,m_b)\nonumber\\ % -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-= { \cal m}_{2f}&= & \frac{-ig{\cal g}p^{\mu\nu\alpha\beta}{\chi}^{d}_{\alpha\beta}}{4\sqrt{2}m_w[(k_1+k_2)^2-m_{h^-}^2 ] } { \bar v(p_1)}c_{\mu\nu}u(p_2)\nonumber\\ \hspace{1.0 cm } & & { \bar u(k_1,m_t)}\bigg [ m_b\tan\beta p_r+m_t\cot\beta p_l \bigg ] v(k_2,m_b)\nonumber\\ % -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-= { \cal m}_{2g}&= & 0\end{aligned}\ ] ] with @xmath56 \\ % ------------------------------------------------------------------------------- { \chi}^{a}_{\alpha\beta } & = & m_{h^-}^2 \eta_{\alpha\beta } - ( k'_1)^\mu ( k'_2)^\nu \left [ \eta_{\mu \alpha } \eta_{\nu \beta } + \eta_{\nu \alpha } \eta_{\mu \beta } - \eta_{\mu \nu } \eta_{\alpha \beta } \right]\\ % --------------------------------------------------------------------------------- { \chi}^{b}_{\alpha\beta } & = & \left [ \gamma_{\alpha } \big(k_1+k_3-k_2)_{\beta } - \eta_{\alpha\beta } \big ( k_1{\!\!\!\!/}+k_3{\!\!\!\!/}-k_2{\!\!\!\!/}- 2 m_b\big ) \right ] + ( \beta \leftrightarrow \alpha ) \\ % ----------------------------------------------------------------------- { \chi}^{c}_{\alpha\beta } & = & \left [ \gamma_{\alpha } \big(k_1-k_3-k_2)_{\beta } - \eta_{\alpha\beta } \big ( k_1{\!\!\!\!/}-k_2{\!\!\!\!/}-k_3{\!\!\!\!/}- 2 m_t\big ) \right ] + ( \beta \leftrightarrow \alpha ) \\ % ----------------------------------------------------------------------- { \chi}^{d}_{\alpha\beta } & = & m_{h^-}^2 \eta_{\alpha\beta } - k_3^\mu ( k_1+k_2)^\nu \left [ \eta_{\mu \alpha } \eta_{\nu \beta } + \eta_{\nu \alpha } \eta_{\mu \beta } - \eta_{\mu \nu } \eta_{\alpha \beta } \right],\end{aligned}\ ] ] where @xmath57 , @xmath58 , @xmath59 , @xmath60 , @xmath61 and @xmath62 is the ratio of the two vacuum expectation values in the 2hdm .
note that the contribution from feynman diagram fig.2(g ) vanishes since the trace of the graviton appears in this diagram and , therefore , in the limit of vanishing electron mass , this contribution also vanishes@xcite .
@xmath63 in eq .
( [ coe ] ) represents the summation of the kk excitation propagators .
if the summation over the infinite tower of the kk modes is performed , one will encounter ultraviolet divergences .
this happens because led is an effective theory , which is only valid below an effective energy scale . in the following calculations we naively introduce an ultraviolet cutoff for the highest kk modes and replace the summation by@xcite @xmath64 where @xmath2 is a cutoff scale naturally being of the order of the fundamental scale @xmath15 .
finally , the cross sections for the charged higgs production processes in led following from the amplitudes are : @xmath65 and the final state phase space hypercube elements are defined as @xmath66
in the numerical calculations , we used the following set of sm parameters@xcite : @xmath67 for the yukawa coupling of the bottom quark at the @xmath68 vertex , as suggested by ref.@xcite , we used the top quark pole mass and the qcd improved running mass @xmath69 with @xmath70gev , which were evaluated using the nlo formula @xcite as follows : @xmath71 here the evolution factor @xmath72 is @xmath73 , \nonumber \\ d^{(f)}=\frac{12}{33 - 2f } , \hspace{1.0 cm } j^{(f)}=-\frac{8982 - 504f+40f^2}{3(33 - 2f)^2 } \ , , \end{aligned}\ ] ] and @xmath74 is the number of active light quarks . for the energy scale @xmath75
, we chose @xmath76{m_tm_bm_{h^-}}$ ] as in ref .
@xcite for @xmath1 . in fig .
3 we show the dependence of the the cross sections in 2hdm ii and led for the process @xmath0 on @xmath77 assuming @xmath78gev .
the dashed lines represent the 2hdm ii results and the solid lines represent the cross section in 2hdm ii led for different values of @xmath2 . from the figures
one can see that when @xmath77 is small(@xmath79gev ) the led effects can be very large up to @xmath80tev .
for example , when @xmath81tev and @xmath82gev the cross section in the 2hdm is 23.8fb , while for 2hdm ii led it is 152.4fb for @xmath83tev and 36.6fb for @xmath80tev . in fig . 4 , the cross section in the 2hdm ii and led for the process @xmath0 are plotted as functions of @xmath84 assuming @xmath82gev
. from this figure
we see that the effects of virtual graviton exchange can substantially modify the @xmath0 cross section compared to its 2hdm ii value , especially at large @xmath84(@xmath85tev ) , enabling led up to @xmath86tev to be probed . in fig .
5 we show the dependence of the cross sections in the 2hdm ii and led for the process @xmath0 on the effective energy scale @xmath2 assuming @xmath5gev and @xmath78gev , 2000gev , and 3000gev . as expected , when @xmath84 is fixed the cross section in led tends to the value in the 2hdm ii when @xmath2 tends becomes very large .
in addition , the figure also shows that for large @xmath84 and small @xmath2 the led effects are very large .
for example , for @xmath84=3tev and @xmath2=4tev the cross section in the 2hdm ii is 3.4fb while the value in led is 58.7fb . in fig . 6
, we present the differential cross section @xmath87 in the 2hdm ii and led for the process @xmath0 as a function of @xmath88 , where the scattering angle @xmath89 is the angle between @xmath90 and the incoming positron assuming @xmath82gev , @xmath78gev , and @xmath91gev .
both differential cross sections in the 2hdm ii and led vanish at @xmath92 .
however , their shapes are different .
note that the led cross section is not symmetric under @xmath93 while the 2hdm ii one is symmetric .
we further have plotted the different contributions to the led cross section and find the asymmetry comes from the interference of the led and 2hdm ii amplitudes , which is in proportion to @xmath94 ( @xmath95 and @xmath96 are mandelstam variables ) and thus to @xmath97 . moreover , the led contribution has peaks in both the forward and backward regions and also vanishes at three points : @xmath98 and @xmath99 , as found in ref .
@xcite , since it is proportional to @xmath100 and thus to @xmath101 . in fig . 7
, we show the dependence of the the cross section in the 2hdm ii and led for the process @xmath1 on @xmath77 assuming @xmath78gev for @xmath7 .
the dashed lines represent the 2hdm ii results and the solid lines represent the cross sections for led with different values of @xmath2 .
this figure shows that when @xmath2 is small(@xmath102tev ) and @xmath77 is also small(@xmath103gev ) , the led effects are very large .
for example , when @xmath7 and @xmath104gev the cross section in the 2hdm ii is 0.017fb . in fig . 8 , we present the dependence of the cross section in the 2hdm ii and for led for the process @xmath4 on @xmath62 assuming @xmath78gev and @xmath105gev .
observe that the cross section in both the 2hdm ii and led exhibit minima close to @xmath106 , since the average strength of the @xmath107 coupling , which is proportional to @xmath108 , is then minimal @xcite . from these figures
one also sees that when @xmath80 is small(@xmath102tev ) and @xmath77 is also small(@xmath103gev ) the led effects are large for @xmath81tev . in fig .
9 the cross sections in the 2hdm ii and led for the process @xmath1 are plotted as functions of @xmath84 , assuming @xmath109gev and @xmath7 .
since we are interested in the case where @xmath110 , @xmath84 should be in the range from about 983gev to 1600gev . from this figure
we see that , as in the case of @xmath0 , the effects of virtual graviton exchange can significantly modify the @xmath4 cross section compared to its 2hdm ii value , especially at large @xmath84(@xmath111tev ) , where led up to about @xmath112tev can be probed . in fig .
10 we show the dependence of the cross section in the 2hdm ii and led for the process @xmath1 on the effective energy scale @xmath2 assuming @xmath113gev , and @xmath114 , and @xmath115gev , respectively .
as expected , when @xmath84 is fixed the led cross section tends to the value in the 2hdm ii when @xmath2 becomes very large .
this figure also shows that for large @xmath84 and small @xmath2 the led effects can be very large .
for example , when @xmath84=2tev and @xmath2=4tev the cross section in the 2hdm ii is 0.006fb , while in led it is 0.013fb . in fig .
11 we show the differential cross sections @xmath116 in the 2hdm ii and led for the process @xmath1 as functions of the transverse momentum of the charged higgs boson @xmath117 assuming @xmath105gev , @xmath78gev , and @xmath118gev , @xmath115gev , @xmath119gev and @xmath120gev .
the shape of the differential cross sections in the 2hdm ii and for led are slightly different .
the led effects generally enhance the differential cross sections , especially when @xmath2 is small .
we briefly note the relevant results in type i 2hdm .
the only change is in the fermionic couplings .
thus , the 2hdm and led results for @xmath0 remain the same .
however , the 2hdm results for @xmath1 in the type i model are different compared with the type ii model .
for example , the results in type i model will decrease with increasing @xmath62 , since the @xmath68 coupling is proportional to @xmath121 in the type i model .
however , the led effects will be similar : large effects when @xmath84 is large and @xmath2 is small .
finally , we briefly discuss the experimental sensitivity .
we will assume that the led effects can be observed , if@xcite @xmath122 where the integrated luminosity @xmath123 are assumed to be @xmath124 . for @xmath0 , from fig . 5
, we see that the led can be probed , for example , up to 2800gev and 7200gev with @xmath5gev for @xmath78gev and @xmath120gev , respectively . for @xmath1 , from fig .
10 , we see that the led can be probed , for example , up to 1400gev and 3600gev for @xmath125 and @xmath126gev , and @xmath127 , and @xmath115gev , respectively .
in the 2hdm ii with large extra dimensions we investigated the contributions of virtual kk gravitons for the 2hdm ii charged higgs production in the two important production processes : @xmath0 and @xmath1 at future linear colliders .
we found that kk gravitons can significantly modify these total cross sections and their differential cross sections compared to the corresponding 2hdm ii values and , therefore , can be used to probe the effective scale @xmath2 up to several tev .
for example , at @xmath3tev the cross sections for @xmath0 and @xmath1 in the 2hdm ii are 7.4fb for @xmath5gev and 0.006fb for @xmath6tev with @xmath7 , while in led they are 12.1fb and 0.013fb , respectively , for @xmath8tev .
we thank prasanta kumar das for useful discussions .
this work was supported in part by the national natural science foundation of china , under grant nos.10421003 and 10575001 , and the key grant project of chinese ministry of education , under grant no.305001 , and the u.s .
department of energy , division of high energy physics , under grant no.de-fg02-91-er4086 .
99 n. arkani - hamed et al . , phys.lett.b429 ( 1998 ) 263 ; n. arkani - hamed et al . , phys.rev.d59 ( 1999 ) 086004 ; n. arkani - hamed et al . , phys.lett.b436 ( 1998 ) 257 .
l. randall , r.sundrum , phys.rev.lett.83 ( 1999 ) 3370 ; l. randall , r.sundrum , phys.rev.lett.83 ( 1999 ) 4690 .
j. lykken , phys.rev.d54 ( 1996 ) 3693 .
e. witten , nucl.phys.b471 ( 1996 ) 135 .
p. horava and e. witten , nucl.phys.b460 ( 1996 ) 506 ; nucl.phys.b475 ( 1996 ) 94 .
i. antoniadis , phys.lett.b246 ( 1990 ) 377 .
s. glashow and s. weinberg , phys.rev.d15 ( 1977 ) 1958 ; j. f. gunion et .
al . , 1990 , higgs hunter s guide(reading , ma : addison - wesley ) .
c. csaki , tasi lectures on extra dimensions and branes , hep - ph/0404096 .
e. g. adelberger [ eot - wash group collaboration ] hep - ex/0202008 .
t. hambye and k. riesselmann , phys.rev.d55 ( 1997 ) 7255 .
s. komamiya , phys.rev.d38 ( 1988 ) 2158 ; j. guasch et al .
, nucl.phys.b596 ( 2001 ) 66 .
s. kanemura et al . , jhep0102 ( 2001 ) 011 ; a. djouadi
z. phys.c54 ( 1992 ) 255 b. a. kniehl et al . , phys.rev.d66 ( 2002 ) 054016 .
h. e. haber and g. l. kane , phys.rep.117 ( 1985 ) 75 .
g. f. giudice et al . , nucl.phys.b544 ( 1999 ) 3 . t. han , j. d. lykken and r. j. zhang , phys.rev.d59 ( 1999 ) 105006 .
d. choudhury , n. g. deshpande , d. k. ghosh , jhep 0409 ( 2004 ) 055 .
particle data group , s. eidelman , et al .
b592 ( 2004 ) 1 .
m. carena et al . , nucl.phys.b577 ( 2000 ) 88 .
n. delerue et al . , phys.rev.d70 ( 2000 ) 091701 .
a. krause et al . , nucl.phys.b519 ( 1998 ) 85 .
s. hao et .
al . , phys.rev.d71 ( 2005 ) 075005 .
|
in the two - higgs - doublet model(2hdm ) with large extra dimensions(led ) , we study the contributions of virtual kaluza - klein(kk ) gravitons to 2hdm charged higgs production , especially in the two important production processes @xmath0 and @xmath1 , at future linear colliders ( lc ) .
we find that kk graviton effects can significantly modify these total cross sections and also their differential cross sections compared to their respective 2hdm values and , therefore , can be used to probe the effective scale @xmath2 up to several tev .
for example , at @xmath3tev , the cross sections for @xmath0 and @xmath4 in the 2hdm are 7.4fb for @xmath5gev and 0.003fb for @xmath6tev and @xmath7 , while in led they are 12.1fb and 0.01fb , respectively , for @xmath8tev .
|
the rich open cluster m37 ( ngc 2099 ) in the constellation auriga has been photometrically monitored to search for new variables with different motivations @xcite . the combination of short and deep exposures with similar sky coverage allowed time - series photometry of hundreds to thousands of stars from the brightest to the dimmest within the cluster field ( @xmath1 ) .
their studies indicate that the fraction of variable sources @xmath2 increases as photometric precision of the survey gets better .
@xcite performed the first variability survey for the central ( @xmath3 ) field of the m37 with a @xmath4-filter photometry and found 7 variable stars among 2300 stars ( @xmath5 ) .
due to poor photometric quality , their variables were with large variability and included three w uma - type eclipsing binaries ( ebs ) , two high - amplitude pulsating stars , and two long - period detached eb candidates .
@xcite and @xcite used the same data set to search for short - period pulsating variables ( e.g. , @xmath6 scuti- and @xmath7 doradus - type stars ) and rotating variables of later spectral types ( f k ) , respectively . using @xmath8-band time - series observations
, these two studies found the total of 153 variable stars among 12,000 stars ( @xmath9 ) , including seven previously known variables , in the field of the cluster ( @xmath10 ) .
lastly , the most extensive variability studies in this field were carried out by @xcite as the by - products of a deep survey for transiting planets in the m37 field ( @xmath11 ) with the 6.5 m mmt telescope . the number of known variable stars increased by a factor of @xmath1210 . by applying several statistical algorithms
, they identified 1445 variables among 23,000 stars ( @xmath13 ) , of which 99% were new discoveries @xcite .
this catalog of variable stars provides the largest homogeneous set of stellar rotation periods for a cluster older than 500 myr , which showed the crucial role of the rotation in the stellar angular momentum evolution @xcite . in this paper , we search new variable phenomena from the same data set of @xcite with vastly improved photometric method ( @xcite , hereafter paper i ) . in section 2 , we give a brief description of new light curves and its temporal properties .
we present a comprehensive re - analysis of photometric variability in section 3 . in section 4 ,
we present a new catalog of variable stars and various variability characteristics .
we compare our findings with previous works and discuss the nature of each variability class separately in section 5 . between two consecutive frames ( top left ) and the exposure times @xmath14 ( top right ) .
we use logarithmic scale for the horizontal axis in these two panels .
the bottom panel is the distribution of exposure times as a function of the modified julian date ( mjd).,scaledwidth=50.0% ]
we refer the reader to paper i for a detailed description of our photometric reduction and light curve production applied to the archival imaging data of mmt transit survey . here
we provide only a brief summary of the key features . in paper
i , we introduced new methods for robust high - precision photometry from well - sampled images of a non - crowded field with strongly varying psf .
the archival light curves from the original image subtraction procedure exhibit many unusual outliers , and more than 20% of data get rejected by the simple filtering algorithm adopted by early analysis .
we note that some of these outliers are results of intrinsic variability of short time scales . in order to achieve better photometric precisions and also to utilize all available data , the entire imaging database were re - analyzed with our multi - aperture indexing photometry and a set of sophisticated calibration procedures .
this algorithm finds an optimal aperture for each star with a maximum signal - to - noise ratio , and also treat peculiar situations where photometry returns misleading information with more optimal photometric index .
we also applied photometric de - trending algorithm based on a hierarchical clustering method .
our method removes systematic variations that are shared by light curves of nearby stars while true variabilities are preserved .
consequently , our method utilizes nearly 100% of available data and reduce the rms scatter several times smaller than original analysis for brighter stars . our new data set includes 30,294 objects , including both point and extended sources .
for blended targets , the original image subtraction light curve is still of higher precision than the present aperture photometry light curves , and thus for these objects the archival light curve is used . in order to pick out such cases , we utilized a simple method for quantifying the level of blending ( see @xcite ) .
@xmath15 where @xmath16 for light curve points @xmath17 with uncertainties @xmath18 , and @xmath19 is the same statistic measured with respect to a 4th order polynomial fit .
we adopt the value @xmath20 for selecting significant blended light curves , and then count the number of blended ones ( 3746 objects ) .
figure [ fig1 ] shows that the cumulative fraction function of temporal intervals @xmath21 between two consecutive frames and the exposure times @xmath22 .
this observing cadence is sensitive to variability on timescales ranging from tens of seconds ( 30 seconds ) to one month ( @xmath1231.7 days ) . in particular
, the time resolution allows an analysis of the short - term variability , which is less than @xmath1280 seconds over 50% of the data and @xmath12150 seconds over 90% of the data .
variability features can be divided into two broad categories : periodic and non - periodic feature .
the former is easier to detect and model using statistical and model - specific fitting procedures as a form of fourier series ( e.g. , @xcite ) .
but the latter is rather ambiguous , since it is not always possible to characterize flux variation and to define a reliable range for some parameters . to overcome this difficulty
, @xcite calculated many basic statistics that can describe the distribution of fluxes even in the limit of few data points .
one interesting approach is to select stochastically varying features in quasar - like sources .
the correlated variability of quasars is well parameterized by a damped random walk model , providing a simple statistical description of their characteristic timescales and amplitudes ( e.g. , @xcite ) . in this work
we use a modified version of the fast @xmath0 periodogram algorithm ( f@xmath0 ) and the change - point analysis ( cpa ) method for detecting and assessing the significance of periodic and non - periodic signals , respectively .
we first searched a periodic feature by f@xmath0 technique that calculates the minimized @xmath0 as a function of frequency at the desired number of harmonics @xcite . with the best - fit frequency , we compared the ratio between the variance of the data after subtracting the best - fit model and full - amplitude of model function .
this algorithm use a fourier series truncated at harmonic @xmath23 as a model function : @xmath24 the fitting coefficients ( @xmath25 , @xmath26 , and @xmath27 ) obtained in each harmonic model provide a reasonable description of a generic periodic feature , irrespective of the feature shape , as discussed by @xcite .
figure [ fig2 ] shows example light curves of both periodic ( new variable star ; v1983 ) and long - term ( v1092 ) variable stars in our sample , respectively .
because the optimal choice of @xmath23 depends on the true shape of the signal , the fit quality is evaluated for each model with standard @xmath0 statistics , and then checked the residual ( phase - folded ) light curves after the subtraction of the best - fit model . to test the significance of a signal , we use the ratio between the full - amplitude of the model function ( @xmath28 = @xmath29 ) and the rms scatter ( @xmath30 ) of the data after model subtraction : @xmath31 the lower limit for the ratio is empirically set to @xmath32 = 0.95 based on light curve inspections .
@xmath32-values less than 0.95 indicate that we do not expect any significant ( semi- ) periodic signals for a given light curve .
even when the source is poorly described by the harmonic fit , this indicator provides a hint of a generic periodic signal ( see right panel of figure [ fig2 ] ) .
most variability detection methods require conventional models that are mainly focused on the periodic signals , and are not suitable for the study of arbitrary - shaped , non - periodic , and sporadic variations .
also , in many cases , signal estimation is equated with smoothing of data for de - noising .
this sometimes discards vital information in time series data .
we introduce a non - parametric method to extract all significant features based on the cpa using local statistics .
change - point analysis is a method to identify abrupt variations in sequential data .
it is widely used in the statistics and data mining communities as well as in the field of time domain astronomy ( see e.g. , @xcite and reference therein )
. using a combination of cumulative sum scheme ( cusum ) and bootstrap rank statistics @xcite , we produce a series of estimated change - points which correspond to the moments of apparent systematic changes . for a given dataset \{@xmath33 , @xmath34 , @xmath35 , @xmath36 } of size @xmath37 ,
the cusum values are given by @xmath38 for @xmath39 = @xmath40 , where @xmath41 = 0 and the mean value @xmath42 for the sample of @xmath37 values of @xmath43 .
an inflection point at which the sign of the cusum slope changes is used to determine whether a given interval of data should be kept as one ( @xmath44=@xmath45=@xmath35=@xmath46=@xmath42 ) or subdivided into two subintervals ( @xmath44=@xmath35=@xmath47@xmath48@xmath49=@xmath46 ) . based on @xmath50 bootstrap samples that are randomly re - ordered original values , we estimate the confidence level ( c.l . ) to reduce the false positives by random noise . here
are the main procedure of each bootstrap process : * we generate a bootstrap sample of @xmath37 dataset , denoted \{@xmath51 , @xmath52 , @xmath35 , @xmath53 } , by randomly re - ordering the original @xmath37 values without replacement .
* based on the bootstrap dataset , we calculate the bootstrap cusum @xmath54 by using equation ( 4 ) , and then obtain the maximum , minimum , and difference of the bootstrap cusum .
@xmath55 where @xmath56 @xmath57 * we check whether the bootstrap difference @xmath58 is less than the original difference @xmath59 .
it is straightforward to derive estimates of confidence level as follows : @xmath60 where @xmath61 typically more than 90% confidence level is suitable for most cases .
when a sudden change is detected , the location of change - point is initially determined as follow : @xmath62 } \left|s_{t}\right|,\ ] ] where @xmath63 denotes the last point before the change occurred .
the time - series data is split into two segments on each side of the change point , and the analysis is repeated for each segment until no more significant change point is detected . once this set is generated , all the change - point locations and their confidence levels are re - estimated in a backward elimination manner .
using only the sample in the segment bounded by adjacent change - points , we re - calculate these parameters detected by the forward procedure to obtain more accurate estimations .
when any initial change - point fails the criteria in the backward procedure , we eliminate this change - point estimate .
this process repeats until no further improvement is possible . the final change
points thus define the segments ( i.e. , piecewise constant level sets ) , characterized by the start and end time of a given interval with its mean and variance . after this procedure
, signal detection process is performed for each segment separately . to detect significant outlying features occurring at specific levels in light curve
, we adopt a simple criteria similar to micro - lensing alert system @xcite in the presence of hetero - scedastic measurement errors ( @xmath64 ) : @xmath65 @xmath66 @xmath67 @xmath68 where the mean @xmath69 and deviation @xmath70 are the local statistics for a given segment , @xmath64 is the photometric error at epoch @xmath71 , and @xmath72 is the number of consecutive points which satisfy the equations ( 8a8c ) . the different combination of @xmath73 values are used for maximize the detection efficiency of significant outlying features .
the values of @xmath73 were taken to be at least larger than 3 , 1 , and 3 , respectively . as a result
, statistically significant events were flagged as candidate flare sources if at least three contiguous points satisfy the above criteria .
example of flare - like feature detection is shown in the left panel of figure [ fig3 ] . by applying the cpa method to the entire light curves
, we efficiently identified several hundred instances of abrupt brightness changes without any smoothing or interpolation of the raw data .
we use the same limits of cpa thresholds to find eclipsing - like event detection , except the first term of equation ( 8) , i.e. , @xmath74 .
the cpa method is useful to detect the moments of eclipse ingress , center , or egress in cases where the eclipsing pattern is not repeated or the coverage is not sufficient for the detection through conventional period analysis ( e.g. , box - fitting least - squares method ) .
example of eclipsing - like feature detection is shown in the right panel of figure [ fig3 ] .
we will discuss this issue further in section 5.1 . for periodicity , we first used the vartools light curve analysis program to determine likely periods of variable stars .
this code computes three different periodograms for the light curves;(i ) lomb - scargle ( ls : @xcite ) , ( ii ) analysis of variance ( aov : @xcite ) , and ( iii ) box - fitting least - squares ( bls : @xcite ) algorithms .
because each method is sensitive to different types of periodic variables , it is often useful to combine results from many different analysis of periodicity .
basically we adopt the similar treatment as in @xcite .
we limited the search range of these algorithms to periods between 0.01 and 30 days with relatively fine frequency resolution of less than 0.01/@xmath75 ( @xmath75 is the time - span of the light curve ) .
we also obtained period estimates from the statistical technique sigspec @xcite .
it calculates the spectral significance ( sig ) levels based on the analytical solution for discrete fourier transform amplitude spectrum , including dependencies on frequency and phase . in recent years , this package is mainly used in the study of stellar pulsation ( i.e. , asteroseismology ) , which shows up as small , periodic variations in the brightness of the stars ( e.g. , @xcite ) . in order to further crosscheck results of periodicity analysis
, we also performed a multi - step period search algorithm msperiod @xcite .
figure [ fig4 ] shows examples of three selected light curves ; power spectrum and phased diagram corresponding to the highest peak .
the differences in the various algorithms depend strongly on the shape of variation .
one case ( v2256 ) shows that all periods are essentially identical ( @xmath76 0.0784 days ) , while in the other cases the derived period differ a lot .
visual inspections are necessary for v2173 and v1482 to choose the most plausible periods ; @xmath77 and 2.595 respectively .
we provide a catalog of the total 2306 stars that exhibits convincing variations that are induced by flares , pulsations , eclipses , starspots , or in some cases , unknown causes .
this catalog is made available in electronic form at http://stardb.yonsei.ac.kr/. table [ taba1 ] gives the basic information about previously known and new variable stars that we detected in the field of m37 .
the columns are as follows : : the variable star ids for the first 1483 stars are from the numbering system of @xcite .
new variable i d numbers are given in ascending order by their right ascension from v1484 to v2306 .
: each object is designated by the combination of right ascension and declination for j2000 epoch ( e.g. , 055536.78 + 323345.97 ) .
: we basically used the dominant periods which were highly scored by several methods . for ebs
, we checked the existence of secondary eclipse when phased at twice the period . in cases where our periodicity analysis is affected by a large scatter of light curves due to photometric bias or source blending , we used periods taken from the previous catalog @xcite , and then denoted by @xmath78 in that column .
we also found evidences of multiperiodicity in some pulsating stars ( see section 5.2 ) . :
the range of variability for periodic variables was derived from phase diagram in which the shape of the light curve was modeled by equation ( 2 ) .
this amplitude is defined as the full - amplitude of model function ( @xmath79 = @xmath29 ) .
when this was not possible , i.e. , cases of ebs and irregulars , we used the original light curve to estimate the amplitude . :
@xmath80 magnitudes for many of our stars are available from the sloan digital sky survey data release 7 ( sdss dr7 ) and two micro all sky survey ( 2mass ) .
each catalog is positionally matched to mmt sources with an average distance of 0.6@xmath81 . in the case of sdss , the photometric quality at low galactic latitudes
is not guaranteed to be accurate to the sdss quoted limits of 2% in color , and photometric depths are about two magnitude shallower than our data .
similarly , 2mass is not deep enough to detect low - mass members in the cluster . :
we assigned five variability types similar to the variability classification scheme proposed by the gcvs @xcite .
variable stars are grouped according to major astrophysical reasons for variability , such as flare ( f ) , pulsating ( p ) , rotating ( r ) , eclipsing binary ( eb ) , and aperiodic variables ( a ) , including variable candidates ( :) . further details are given in section 5 . if a variable belongs to more than one type of variability , it is indicated by combination of individual types , e.g. , f+eb , p+eb : . a total of 30 stars listed as variables in @xcite turned out to be not real variable in our study .
their new amplitude spectrum often does not show anything significant ( figure [ fig5 ] ) and no apparent periodicity in the light curve .
these false positives are listed in table [ tab1 ] .
rcrcrr 1 & v27 & 055118.34 + 323201.71 & 32 & 20.938 & 16.364 + 2 & v85 & 055125.32 + 324341.29 & 28 & 20.862 & 7.668 + 3 & v88 & 055125.46 + 324315.32 & 28 & 21.820 & 1.646 + 4 & v120 & 055128.52 + 322339.10 & 35 & 17.196 & 6.859 + 5 & v155 & 055131.10 + 323756.98 & 30 & 16.847 & 7.424 + 6 & v166 & 055131.96 + 323600.24 & 31 & 20.895 & 10.451 + 7 & v224 & 055137.29 + 324011.78 & 29 & 18.165 & 4.571 + 8 & v230 & 055138.02 + 324011.65 & 29 & 16.690 & 4.595 + 9 & v250 & 055139.59 + 324013.27 & 29 & 17.282 & 4.551 + 10 & v330 & 055151.25 + 323527.70 & 22 & 16.831 & 5.428 + 11 & v337 &
055151.72 + 323510.02 & 22 & 17.920 & 5.442 + 12 & v342 & 055152.18 + 323522.44 & 22 & 19.061 & 5.438 + 13 & v347 & 055152.55 + 323536.03 & 22 & 15.574 & 5.500 + 14 & v367 & 055153.88 + 323459.77 & 22 & 17.463 & 5.515 + 15 & v390 & 055155.42 + 323523.56 & 22 & 18.268 & 5.481 + 16 & v408 & 055156.77 + 323534.61 & 22 & 15.733 & 5.481 + 17 & v433 & 055157.99 + 323528.59 & 22 & 16.617 & 5.481 + 18 & v492 & 055202.16 + 324517.30 & 19 & 20.849 & 16.586 + 19 & v973 & 055235.48 + 322104.58 & 18 & 16.847 & 5.089 + 20 & v1076 & 055241.54 + 322107.71 & 18 & 20.262 & 10.927 + 21 & v1237 & 055258.14 + 322152.09 & 9 & 16.775 & 5.692 + 22 & v1260 & 055300.53 + 323445.47 & 4 & 19.851 & 15.895 + 23 & v1282 & 055303.05 + 323621.95 & 4 & 21.530 & 1.666 + 24 & v1305 & 055305.05 + 323447.83 & 4 & 18.166 & 7.966 + 25 & v1330 & 055306.74 + 323442.52 & 4 & 18.061 & 8.143 + 26 & v1371 & 055310.61 + 323748.13 & 3 & 20.902 & 9.934 + 27 & v1372 & 055310.65 + 323740.82 & 3 & 20.985 & 10.633 + 28 & v1400 & 055312.89 + 323005.46 & 6 & 16.770 & + 29 & v1421 & 055314.64 + 322656.12 & 7 & 20.735 & 15.013 + 30 & v1423 & 055315.29 + 323051.00 & 6 & 20.728 & 12.388 + in this work , we found 823 new variables ( v1484v2306 ) and identified types of their variability .
the discovery rate of new variables is increased by 60% in comparison with the existing catalog .
the new variables are primarily 436 flaring stars , 30 sharp ebs , 48 low - amplitude pulsating variables , and 65 aperiodic variables , as well as many other rotating variables .
we briefly explored how variability fractions depends on their brightness / period and variable types .
as shown in figure [ fig6 ] , short - period , low - amplitude pulsating stars and episodic flare events are dominantly found at the bright- and faint - end of magnitude , respectively . in figure
[ fig7 ] , we plot variability fraction as a function of period .
the period range of new variables is very similar to that reported by @xcite . of these ,
most of new pulsating candidates show @xmath6 scuti - like characteristics ( @xmath82 0.3 days ) . also the majority of periodic variability can be explained by rotational modulation of surface features ( i.e. , starspots ) . magnitude .
previously known variables ( v1v1483 ) , newly detected variables ( v1484v2306 ) , and total number of observed stars are marked with dashed lines , gray solid lines , and black solid lines , respectively.,scaledwidth=45.0% ] .,scaledwidth=50.0% ] time - scales and amplitudes of variability are intrinsic properties of the star and are independent of the distance @xcite .
these two quantities can easily be measured from the light curves .
we extracted the basic properties ( e.g. , period , amplitude ) of all periodic variables using the fourier decomposition technique .
for comparison purposes , we utilized the flare parameters : ( i ) flare duration @xmath83 that is defined as the time - scales of flare events at 80% of its peak level , and ( ii ) flare amplitude @xmath84m@xmath85 that is magnitude at the moment of flare maximum ( s. -w .
chang et al . , in preparation , hereafter paper iii ) .
we separate the variable objects into photometrically selected candidate cluster and field members provided by @xcite , respectively .
since the @xmath86 color covers a relatively wide range from @xmath87 to 4.5 , we used @xmath86 as a separator between a sequence of cluster and that of field population .
as can be seen in figure [ fig8 ] , short period ( @xmath88 1 day ) , small amplitude pulsating stars are found at the blue - end of the distribution of both cluster and field sources , while episodic flare events are dominantly found from the sources in the red color range .
our new data strengthen the relationship between the color of stars and their variability parameters derived for cluster members previously ( figure [ fig8 ] ) .
the cluster members show a tight and nearly linear relation between color index and rotation period in the range of 1 @xmath88 @xmath89 @xmath88 20 days , although the change in slope occurs after the knee point at @xmath90 @xmath91 7 days . this sequence forms a curved , diagonal band from bright blue stars to faint red ones , whose periods are increasing with increasing color index ( or magnitude ) , while for non - cluster members there is no clear correlation .
the color - amplitude relation for the cluster members shows that , statistically , cooler stars have greater amplitudes in the case of both periodic variables and flares . as in the previous study on stellar rotation ( see section 6 and figures 14 , 15 and 17 in @xcite ) , there is a strong correlation between rossby number and amplitude , which can be interpreted as stars rotating faster having more spots , larger spots , and/or greater spot - photosphere contrast than slower rotating stars .
cooler stars have on average lower rossby numbers at a particular age , so they end up having larger amplitudes .
figure [ fig9 ] illustrates the color - magnitude diagram ( cmd ) for all objects and all variable stars in the m37 field . in the new catalog
, we found that about half of variable objects are distributed along the cluster main - sequence , and the other half are located in the field or in the outskirts of the cluster sequence . in order to identify the latter stars , we compared our data with the trilegal model to reproduce the cmd at this location of the galaxy .
most of the input parameters are described in the @xcite . for this analysis
, we set the pointing parameters for the field center ( @xmath92 ; @xmath93 ) and total field area ( @xmath94 = 0.16 deg@xmath95 ) , a limiting magnitude as @xmath96 = 23.5 and a salpeter initial mass function with cutoff at 0.01 m@xmath97 .
the remaining parameter values are fixed to default values that correspond to their model calibration .
the model output consists mostly of thin disk stars ( 96% ) with small contribution of old stars in the thick disk and halo ( 4% ) .
non - member , variable stars may be related in some way to the thin disk population , but their distributions are not fully explained by the model . vs. @xmath71 diagram of both observed ( left ) and simulated ( right ) data .
photometrically selected cluster region is plotted as red line in both panels . ]
in the following subsections , we investigate the nature of each variability class : ( i ) 61 eb systems , ( ii ) 92 multiperiodic variable stars , ( iii ) 132 aperiodic variables , and ( iv ) 436 flare stars , respectively . in order to understand variability properties , we utilized both color - magnitude and color - color diagrams to separate variable objects into broad magnitude ranges or spectral classes , to the extent possible .
our photometric time - series data provide the light curves of 61 ebs .
figure [ fig10 ] shows phased light curves of 8 newly discovered ebs .
we categorize ebs into three groups according to the shape of phased light curves , and discuss each group below .
ebs in this category are either algol - type systems or @xmath98 lyrae - type systems .
to derive their geometric parameters , we used the detached eclipsing binary light curve ( debil ) fitter by @xcite , which employs a simple model of a perfectly detached binary system in a classical 2-body orbit .
table [ tab2 ] lists parameters derived for all detached / semi - detached binaries in the m37 field .
ebs with well determined periods are included in debil analysis .
special notes are given below for interesting features .
[ [ v2 ] ] v2 + + this bright eb system was missed by previous study due to magnitude cut , but it clearly shows evidences of eclipse in our new light curve .
[ [ v587 ] ] v587 + + + + the light curve of this eb system contains two distinct periodic signals , the 3.6173 days modulation with a small amplitude ( @xmath99 @xmath91 0.056 mag ) and an additional period at 3.2857 days ( orbital period ) . since an ellipsoidal variation should have the same period as the orbit , the first signal is more likely a consequence of starspots and rotational modulation ..
after 3.6173-d modulation is removed , the light curve clearly shows the signature of a typical algol - type binary with both primary and secondary eclipses .
[ [ v1482 ] ] v1482 + + + + + this eb system is a newly recognized algol - type binary with a @xmath6 sct - type pulsating component .
we have clearly detected an orbital period of 2.5972 days including multiple primary and secondary eclipses .
the short period variability caused by stellar pulsation is shown in the out - of - eclipse light curves . in order to determine the oscillation modes of the pulsation ,
a frequency analysis was performed in the out - of - eclipse data using the program period04 @xcite .
our analysis not only shows peaks as a consequence of binary effects ( @xmath100 = 2.5638 and @xmath101 = 1.3006 ) but also reveals the presence of the dominant pulsation frequency @xmath102 = 0.0465 with a small amplitude ( @xmath99 = 0.0035 mag ) .
rcrrrrrrrrrrl 1 & & 13.677 & & & & & & & & & & not enough data + 2 & & 14.840 & & & & & & & & & & not enough data + 3 & & 22.549 & 0.7780 & @xmath1030.08 & 0.17 & 0.15 & 22.81 & 24.12 & 1.00 & 0.65 & 283.4 & d + 4 & & 20.463 & 0.6477 & & & & & & & & & sd ; no secondary minimum ?
+ 5 & & 22.121 & 0.9123 & @xmath1030.01 & 0.10 & 0.04 & 22.23 & 24.59 & 1.00 & 0.53 & 129.5 & d + 6 & & 17.392 & 3.4460 & 0.01 & 0.10 & 0.08 & 17.96 & 18.35 & 1.00 & 0.33 & 232.1 & d + 7 & & 19.474 & 2.1995 & @xmath1030.08 & 0.08 & 0.07 & 20.15 & 20.30 & 1.00 & 0.75 & 239.7 & d + 8 & & 17.080 & 1.6676 & & & & & & & & & d ; no secondary minimum ?
+ 9 & & 16.572 & 3.2857 & 0.52 & 0.06 & 0.02 & 16.58 & 21.24 & 1.00 & 0.31 & 209.4 & d + rm ( p=3.6173 days ) ( new ) + 10 & & 16.669 & 1.8324 & 0.06 & 0.28 & 0.06 & 16.73 & 19.80 & 0.98 & 0.01 & 77.2 & sd + 11 & & 21.635 & 2.0272 & & & & & & & & & d ; no secondary minimum ? + 12 & & 17.383 & 0.7310 & @xmath1030.04 & 0.43 & 0.34 & 17.61 & 18.91 & 0.99 & 0.77 & 73.3 & d + 13 & & 16.736 & 1.9732 & 0.00 & 0.16 & 0.05 & 16.75 & 21.05 & 1.00 & 0.27 & 357.0 & sd ( new ) + 14 & & 18.730 & 5.4956 & 0.03 & 0.04 & 0.04 & 19.21 & 19.84 & 1.00 & 0.55 & 267.5 & d + rm + 15 & & 16.977 & 3.6948 & @xmath1030.66 & 0.06 & 0.01 & 16.98 & 22.60 & 1.00 & 0.29 & 268.3 & d + 16 & & 21.869 & 2.2834 & 0.05 & 0.07 & 0.06 & 22.40 & 22.87 & 1.00 & 0.16 & 268.3 & d ( new ) + 17 & & 19.742 & 1.1919 & @xmath1030.41 & 0.12 & 0.02 & 19.77 & 23.56 & 1.00 & 0.86 & 271.6 & d + 18 & & 14.874 & 2.1898 & @xmath1030.05 & 0.16 & 0.06 & 15.01 & 17.19 & 0.99 & 0.84 & 280.1 & d + @xmath7 doradus component ( p=0.941/1.174 days ) + 19 & & 17.006 & 5.9680 & 0.12 & 0.07 & 0.07 & 17.47 & 18.15 & 1.00 & 0.54 & 89.3 & d + 20 & v1482 & 15.597 & 2.5972 & 0.02 & 0.14 & 0.08 & 15.71 & 18.14 & 0.99 & 0.17 & 76.1 & d + @xmath6 scuti component ( p=0.0465 days ) ( new ) + 21 & v2138 & 14.181 & 3.4051 & 0.04 & 0.09 & 0.04 & 14.28 & 16.85 & 1.00 & 0.63 & 191.7 & d ( new ) + 22 & v2297 & 15.863 & 3.5109 & @xmath1030.01 & 0.12 & 0.02 & 15.88 & 20.42 & 1.00 & 0.32 & 229.7 & d ( new ) + the light curves of contact binary stars ( also known as w uma - type variable stars ; ew ) are very much different from those for detached binaries .
our new catalog lists 22 contact binaries of which four are new discoveries ( v840 , v1031 , v1250 , v1463 ) .
[ [ v37-v706-v1031 ] ] v37 , v706 , v1031 + + + + + + + + + + + + + + + + these systems are one of the fastest orbiting contact binary systems ( @xmath90 @xmath91 0.22 days ) .
it is known that observed distribution of orbital periods has a cut - off period limit at 0.22 days ( e.g. , @xcite ) .
we have not found anything faster in our study .
rccrcrrr 1 & & 055223.89 + 322826.47 & 15.604 & 1 & 53732.28145 & 1.0 & 0.011 + & & & & 2 & 53734.26454 & 2.6 & 0.142 + & & & & 3 & 53756.09692 & 3.1 & 0.134 + 2 & & 055226.24 + 324044.62 & 15.774 & 1 & 53731.14064 & 1.1 & 0.074 + & & & & 2 & 53745.26403 & 4.7 & 0.171 + 3 & & 055232.65 + 322208.16 & 16.348 & 1 & 53755.08741 & 2.2 & 0.046 + 4 & & 055236.13 + 323001.13 & 16.012 & 1 & 53734.15888 & 2.2 & 0.054 + & & & & 2 & 53756.24427 & 1.1 & 0.051 + 5 & & 055237.96 + 323415.64 & 16.569 & 1 & 53732.14053 & 1.5 & 0.019 + & & & & 2 & 53756.17928 & 1.8 & @xmath88 0.030 + 6 & & 055311.98 + 322353.58 & 17.717 & 1 & 53726.20141 & 0.4 & 0.208 + & & & & 2 & 53730.29695 & 0.7 & 0.146 + & & & & 3 & 53733.22359 & 1.6 & 0.184 + & & & & 4 & 53734.41413 & 0.4 & 0.159 + & & & & 5 & 53737.34213 & 1.2 & 0.161 + & & & & 6 & 53740.33536 & 0.9 & 0.124 + & & & & 7 & 53753.25941 & 0.7 & 0.203 + & & & & 8 & 53756.17828 & 1.2 & 0.157 + 7 & v1731 & 055157.49 + 324534.90 & 15.823 & 1 & 53728.25931 & 3.3 & 0.233 + & & & & 2 & 53732.11914 & 2.5 & 0.282 + 8 & v1748 & 055159.96 + 322433.64 & 14.846 & 1 & 53740.46821 & 0.5 & 0.012 + & & & & 2 & 53742.46911 & 0.5 & 0.015 + 9 & v1757 & 055201.06 + 322408.79 & 17.189 & 1 & 53731.14064 & 0.4 & 0.022 + 10 & v1889 & 055216.11 + 322822.95 & 14.971 & 1 & 53742.42521 & 1.6 & 0.100 + & & & & 2 & 53756.13038 & 3.8 & 0.082 + 11 & v1892 & 055216.39 + 322609.00 & 19.328 & 1 & 53734.15888 & 2.0 & 0.178 + 12 & v1909 & 055218.62 + 324100.16 & 13.281 & 1 & 53731.34461 & 0.0 & 0.019 + & & & & 2 & 53753.15826 & 2.0 & 0.291 + 13 & v1915 & 055218.84 + 324434.08 & 21.189 & 1 & 53755.21194 & 1.9 & 0.273 + 14 & v2213 & 055301.96 + 323744.91 & 16.433 & 1 & 53725.17392 & 2.0 & 0.230 + & & & & 2 & 53733.15900 & 0.5 & 0.025 + 15 & & 055302.02 + 322327.62 & 15.094 & 1 & 53732.11914 & 2.6 & 0.091 + & & & & 2 & 53735.16000 & 1.9 & 0.087 + & & & & 3 & 53754.14737 & 0.8 & 0.051 + with our cpa analysis , we discovered 15 new eclipsing binary candidates that have eclipsing - like or transit - like features undergoing eclipse , eclipse - ingress , or eclipse - egress phases .
table [ tab3 ] summarize observed eclipsing features of the light curves , which are missed by the standard analysis software .
their orbital periods are not well constrained .
the light curve of each event can be described by the first time of dimming points @xmath104 , the time duration between the start time of eclipse and the end time of eclipse @xmath105 , and the depth of observed dimming points @xmath106 .
figure [ fig11 ] illustrates examples of light curves of this category .
[ [ v791 ] ] v791 + + + + this detached eb system has deep primary and secondary eclipses ( 1213% and 1% , respectively ) , and has out - of - eclipse variability at the 2% level with period @xmath124.0631 days .
after pre - whitening with the rotational modulation , our analysis indicates possible orbital period @xmath107 = 2.4277 days .
[ [ v926 ] ] v926 + + + + this object was known to be a long - period ( 16.5777 days ) variable star .
our ls analysis supports the conclusion of previous study , but it has sinusoidal - shaped periodic signal in our light curve .
the ls periodogram shows two power peaks with periods similar to previous one ( rank 1 = 15.7291 days ) and another shorter than one day ( rank 2 = 0.9376 days ) . one sudden eclipse event was caught during its return to quiescence brightness .
[ [ v1018 ] ] v1018 + + + + + we caught two eclipses , but the star seems to vary all the time in a strange manner . by using the plavchan algorithm @xcite provided in nasa exoplanet archive periodogram service , we identified two types of systems which can be confused with a transiting planetary system .
one type is connected with the grazing ebs , and it shows v- or u - shape eclipses .
the other type is the eclipsing binary systems with a large primary star , in which case the eclipse produces a flat bottomed light curve with small deeming .
figure [ fig12 ] shows the phased light curves of actual cases of these contaminants .
the observed shapes of the eclipsing events resemble transit - like signals , but depths of mid - eclipse are more than a few % of the total light ( v985 : 5% ; v1483 : 14% ; v1731 : 24% ; v1889 : 10% ) .
multi - periodicity is a common feature found in nearly all type of pulsating stars such as @xmath6 scuti stars , rr lyraes , and cepheids .
our catalog contains many examples of multiperiodic variable stars which deserve in - depth analysis .
we performed multiple - frequency analysis of 92 short - periodic variables with the program period04 @xcite , looking for the presence of multi - periodicity . with an iterative pre - whitening procedure ,
discrete fourier transformations are calculated until there is no meaningful change in the fit residuals .
each light curve was fitted using an equation of the form : @xmath108 where @xmath25 is a mean level and @xmath109 , @xmath110 , and @xmath111 are the frequency , amplitude , and phase for each successive peak found in the amplitude spectrum .
we adopt a conservative approach in selecting the statistical significant peaks from the amplitude spectrum . in general , a signal / noise ( s / n ) amplitude ratio of 4.0 is a good criterion for independent frequencies , equivalent to 99.9% certainty of variability @xcite . for combination modes with known values ,
the s / n amplitude ratio of @xmath123.5 is used which is equivalent to 90% certainty .
after removal of all significant peaks in the amplitude spectrum , the noise level was estimated on the residual amplitude spectrum over 5 c / d boxes around each frequency .
the resulting light curve of one example multiperiodic variable is shown in figure [ fig13 ] .
the multi - period model is in excellent agreement with the observed light curve .
is shown as a solid curve .
the full amplitude of this variation is about 0.04 mag in @xmath96-band light curve . ]
table [ taba2 ] summarizes all candidate frequencies identified in 92 multiperiodic light curves with their fitted fourier parameters .
all the detected frequencies are classified as independent modes ( @xmath100 , @xmath101 , @xmath102 , @xmath35 ) , combination modes ( @xmath112 ) , and harmonic mode ( @xmath113 ) . by using the method described in section 3.1
, we detected long - term aperiodic variability of 132 variable candidates with time - scales similar to the total observing time span ( @xmath1231 days ) .
we visually inspected each light curve to confirm its variable nature , and this result is illustrated in terms of time sequence ( figure [ fig14 ] ) . due to the lack of periodicity ,
the light curves of aperiodic variables are simply characterized by the level of photometric variability ( @xmath99 ) , the dispersion about the mean level ( @xmath114 ) , and the reduced chi - square assuming a constant median value for the light curve ( @xmath115 ) .
@xmath116 where @xmath117 is the photometric uncertainty of each data point @xmath43 , and @xmath118 is the total number of measurements .
the full parameters are described in table [ taba3 ] .
these variables show relatively low levels of variability with @xmath99 @xmath88 0.03 mag ( 60% ) and @xmath99 @xmath88 0.05 mag ( 80% ) , respectively .
31 days ) . ]
values for sdss photometry system , we used the extinction coefficients compiled in @xcite . ]
we used the combined sdss and 2mass photometry to identify basic properties of these sources .
since color ranges are much smaller in the near - infrared color - color diagram , the stellar locus is not as well defined as in optical color - color diagram ( figure [ fig15 ] ) .
we infer spectral types from @xcite , in which stars in this sample cover spectral types from approximately a5 to m5 and can be either main - sequence and giant stars .
the cmd indicates that many of these aperiodic stars are field stars not associated with m37 .
the origin of aperiodic nature is difficult to explain .
possible speculations include that ( i ) some of these may be cases where long - period field giants are detected as low - amplitude photometric variables on timescales of weeks or longer ( e.g. , @xcite ) , ( ii ) some of them may be semi - regular variables exhibiting both fast and slow variabilities ( e.g. , @xcite ) , and ( iii ) as mentioned in @xcite , the rapid amplitude variation may be flickering , nova - like outbursts seen in many symbiotic variables ( e.g. , @xcite ) .
further observations of much longer duration will be necessary to properly characterize these aperiodic variables .
earlier analysis of m37 data ( hartman et al .
2008a ) could not address flare events properly because many data points were removed as outliers .
our new pipeline , with nearly 100% data utilization , offers opportunity to investigate flare stars more thoroughly . by cpa analysis
, we found that 436 stars ( 625 flare events ) show flaring activity characterized by a rapid flux increase and a slower decay . as expected , the location of these flaring stars on color - magnitude and color - color diagrams suggest that most of them are late - type main - sequence stars ( k4m6 ) in m37 open cluster ( figure [ fig16 ] ) .
flares are also observed from earlier spectral types ( a0k3 ) , but these are only a few and probably belong to field stars . the underlying mechanism of flares in stars cooler than about f5 is essentially the same as in the sun .
it is believed that these cool stars can generate a magnetic field through a dynamos at the interface between the radiative core and the convective envelope @xcite .
thus , flaring is an inevitable result of strong magnetic field . but flares in hot stars , especially in a - type stars , are not easy to explain in the context of dynamo process because these stars should not generate strong magnetic fields in their thin convection layer and are incapable of storing larger energy in that region @xcite .
the observed flare signature of one a - type star ( v72 ) might be the result of binarity .
if an a - type main sequence star has an active m dwarf secondary star , such binary star would look like a flare - generating a - type star .
the number of flare events is increased by a factor of two in the new analysis because our new calibration procedure provides more accurate @xmath96-band light curves ( figure [ fig17 ] ) . in table
[ taba4 ] and figure [ fig18 ] , we list the parameters of flare events discovered by our cpa analysis .
they have a broad range of peak amplitude ( @xmath119 = @xmath120 mag ) and typical durations of a few minutes to several hours , irrespective of their spectral type .
more detailed analysis of flare statistics and its relation to magnetic activity and stellar rotation will be presented in subsequent papers .
-band data in terms of both photometric accuracy and data recovery rate . ]
we cataloged and investigated a variety of photometric variables in the field of m37 using our new light curves , which were obtained from archival imaging data of one - month long exo - planet transit survey with the 6.5-m mmt telescope .
this new dataset allows a rare opportunity to explore different types of variability of short ( @xmath12minutes ) and long ( @xmath12one - month ) time - scales simultaneously .
we find 61 eclipsing binary systems ( eb ) , 92 multiperiodic variable stars ( mp ) , 132 aperiodic variables ( a ) , and 436 flare stars ( f ) , as well as several hundreds of rotating variables ( r ) . our work has increased the number of known variable stars in this field by 60% .
the new variable catalog has been made available on the web ( http://stardb.yonsei.ac.kr/ ) .
we would like to thank the anonymous referee for helpful suggestion .
this research was supported by basic science research program of the national research foundation of korea ( 2011 - 0030875 ) .
i.b . acknowledges the support from kasi - yonsei drc program of korea research council of fundamental science and technology ( drc-12 - 2-kasi ) .
benk , j. m. , kolenberg , k. , szab , r. , et al .
2010 , , 409 , 1585 benzm , a. o. & gdel , m. 2010 , , 48 , 241 breger , m. , stich , j. , garrido , r. , et al .
1993 , , 271 , 482 butler , n. r. & bloom , j. s. 2011 , , 141 , 93 chang , s .- w . , byun , y .-
i . , & hartman , j. d. 2015 , , 149 , 135 ( paper i ) ciardi , d. r. , braun , k. v. , bryden , g. , et al .
2011 , , 141 , 108 covey , k. r. , ivezi , . , schlegel , d. , et al .
2007 , , 134 , 2398 cutri , r. m. , skrutskie , m. f. , van dyk , s. , et al .
2003 , the irsa 2mass all - sky point source catalog , nasa / ipac infrared science archive , http://irsa.ipac.caltech.edu/applications/gator/ debosscher , j. , sarro , l. m. , aert , c. , et al .
2007 , , 475 , 1159 debosscher , j. , sarro , l. m. , lpez , m. , et al . 2009 , , 506 , 519 devor , j. 2005 , , 628 , 411 eyer , l. & mowlavi , n. 2008 , j. phys .
ser . , 118 , 012010 girardi , l. , grebel , e. k. , odenkirchen , m. , & chiosi , c. 2004 , , 422 , 205 girardi , l. , groenewegen , m. a. t. , hatziminaoglou , e. , & da costa , l. 2005 , , 436 , 895 glicenstein , j. -f .
2001 , in asp conf .
239 , microlensing 2000 : a new era of microlensing astrophysics , ed .
j. w. menzies & penny d. sackett .
( san francisco , ca : asp ) , 28 gromadzki , m. , mikoajewska , j. , whitelock , p. , & marang , f. 2009 , acta astron .
, 59 , 169 gromadzki , m. , mikolajewska , j. , & soszynski , i. 2013 , arxiv:1312.6063 grubrbauer , m. , kolenberg , k. , rowe , j. f. , et al .
2007 , , 379 , 1498 guggenberger , e. , kolenberg , k. , nemec , j. m. , et al .
2012 , , 424 , 649 hartman , j. d. , gaudi , b. s. , holman , m. j. , et al .
2008a , , 675 , 1233 hartman , j. d. , gaudi , b. s. , holman , m. j. , et al . 2008b , , 675 , 1254 hartman , j. d. , gaudi , b. s. , pinsonneault , m. h. , et al .
2009a , , 691 , 342 hartman , j. d. , gaudi , b. s. , holman , m. j. , et al .
2009b , , 695 , 336 henry , gregory w. , fekel , francis c. , henry , stephen m. , & hall , douglas s. 2000 , , 130 , 201 irwin , j. , irwin , m. , aigrain , s. , et al . , , 375 , 1449 kalirai , j. s. , ventura , p. , richer , h. b. , et al .
2001 , , 122 , 3239 kang , y. b. , kim , s .- l .
, rey , s .- c . , et al . 2007 , , 119 , 239 kelly , b. c. , bechtold , j. , & siemiginowska , a. 2009 , , 698 , 895 kiss , l. l. , szab , gy . m. , szildi , k. , frsz , g. , srneczky , k. , & csk , b. 2001 , , 376 , 561 kovcs , g. , zucker , s. , & mazeh , t. 2002 , , 391 , 369 kuschnig , r. , weiss , w. w. , gruber , r. , bely , p. y. , jenkner , h. 1997 , , 328 , 544 lenz p. , breger m. 2005 , coast , 146 , 53 lomb , n. r. 1976 , , 39 , 447 messina , s. , distefano , e. , parihar , p. , et al .
2008 , , 483 , 253 molnar , lawrence a. , van noord , daniel m. , & steenwyk , steven d. 2013 , arxiv:1310.0539 richards , j. w. , starr , d. l. , butler , n. r. , et al .
2011 , , 733 , 10 rucinski , s. m. 1992 , , 103 , 960 rucinski , s. m. 2002 , , 114 , 1124 palmer , d. m. 2009 , , 695 , 496 plavchan , p. , jura , m. , kirkpatrick , d. , cutri , roc m. , & gallagher , s. c. 2008 , , 175 , 191 press , w.h . & rybicki , g.b .
1989 , , 338 , 277 press , w. h. , teukolsky , s. a. , vetterling , w. t. , & flannery , b. p. 1992
, numerical recipes in c. 2nd ed .
( new york : cambridge university press ) reegen , p. 2007
, , 467 , 1353 samus , n. n. , durlevich , o. v. , kazarovets , e. v. , et al .
2009 , vizier online data catalog : b / gcvs , 1 , 2025 scargle , j. d. 1982 , , 263 , 835 schtz , n. & holschneider , m. 2011 , , 84 , 021120 schwarzenberg - czerny , a. 1989 , , 241 , 153 shin , m .- s . &
byun , y .-
2004 , jkas , 37 , 79 taylor , w. 2000 , in change - point analyzer 2.0 shareware program , taylor enterprises , libertyville , illinois . , http://www.variation.com/cpa ccrrrrrrrrrrrrc & 055220.39 + 323319.65 & & & & 13.677 & 0.285 & 0.191 & & 13.406 & 0.609 & 12.447 & 0.270 & 0.063 & eb + & 055216.58 + 322815.32 & & & 16.792 & 14.840 & 0.564 & 0.290 & 14.471 & 14.653 & 0.688 & 13.539 & 0.287 & 0.134 & eb + & 055233.01 + 323241.91 & 0.4225 & 0.320 & 17.982 & 15.789 & 0.560 & 0.329 & 15.284 & 16.051 & 0.728 & 14.267 & 0.355 & 0.012 & eb + & 055253.26 + 323301.47 & 0.5582 & 0.235 & 18.079 & 15.634 & 0.762 & 0.477 & 15.093 & 15.948 & 0.938 & 13.951 & 0.497 & 0.101 & eb + & 055300.63 + 322450.81 & 0.2788 & 0.310 & 18.036 & 15.915 & 0.459 & 0.481 & 15.235 & 16.184 & 0.585 & 14.421 & 0.359 & 0.084 & mp + & 055150.53 + 323234.70 & 0.1098 & 0.343 & 18.218 & 15.973 & 0.591 & 0.173 & 15.622 & & & 14.739 & 0.342 & -0.009 & mp + & 055239.09 + 323631.27 & 0.3577 & 0.433 & 19.858 & 17.252 & 1.039 & 0.056 & 17.065 & 17.890 & 0.815 & 15.779 & 0.446 & 0.196 & eb + &
055234.32 + 323218.70 & 0.1195 & & & & & & & & & & & & u + & 055214.91 + 322441.35 & 0.0909 & & 14.873 & 13.233 & 0.137 & 0.128 & 13.114 & 13.298 & 0.370 & 12.348 & 0.171 & 0.042 & p + & 055200.49 + 323648.19 & 0.9432 & 0.075 & 16.844 & 14.151 & 0.928 & 0.436 & 13.493 & 14.762 & 0.531 & 12.356 & 0.567 & 0.131 &
p + rccrrcrrrl 1 & & 055616.79 + 322514.20 & 15.915 & 3.5872 & f@xmath121 & 310.1 & 0.3331 & 68.9 & + & & & & 7.1768 & f@xmath122(=2f@xmath121 ) & 35.4 & 0.2906 & 11.3 & + & & & & 10.7619 & f@xmath123(=3f@xmath121 ) & 22.8 & 0.0418 & 10.3 & + & & & & 5.6768 & f@xmath124 & 19.5 & 0.2369 & 5.7 & + & & & & 14.3473 & f@xmath125 & 13.8 & 0.9990 & 7.1 & + 2 & & 055506.89 + 323303.20 & 15.973 & 9.1040 & f@xmath121(=f@xmath123-f@xmath122 ) & 343.4 & 0.5433 & 172.7 & + & & & & 18.2081 & f@xmath122(=f@xmath123-f@xmath121 ) & 89.9 & 0.2201 & 53.6 & + & & & & 27.3122 & f@xmath123(=f@xmath121+f@xmath122 ) & 27.6 & 0.9237 & 19.7 & + & & & & 8.7545 & f@xmath124 & 15.4 & 0.5801 & 8.1 & + & & & & 11.9319 & f@xmath125 & 14.5 & 0.5557 & 6.3 & + & & & & 36.4178 & f@xmath126(=2f@xmath122 ) & 10.1 & 0.0527 & 10.4 & + & & & & 14.9868 & f@xmath127 & 8.8 & 0.9887 & 4.5 & + & & & & 16.8695 & f@xmath128 & 7.9 & 0.7280 & 4.5 & + 3 & & 055519.81 + 323541.23 & 15.245 & 8.4261 & f@xmath121 & 79.0 & 0.9052 & 84.8 & + & & & & 16.8561 & f@xmath122(=2f@xmath121 ) & 7.6 & 0.6488 & 8.3 & + & & & & 17.7044 & f@xmath123 & 8.0 & 0.5895 & 9.1 & + & & & & 16.4751 & f@xmath124 & 6.9 & 0.5617 & 7.3 & + & & & & 8.8787 & f@xmath125 & 4.9 & 0.4440 & 5.2 & + & & & & 12.0801 & f@xmath126 & 4.8 & 0.8357 & 5.0 & + & & & & 9.2640 & f@xmath127 & 5.9 & 0.6453 & 6.4 & + & & & & 10.6455 & f@xmath128 & 5.4 & 0.0597 & 6.0 & + & & & & 13.5604 & f@xmath129 & 4.8 & 0.8005 & 4.9 & + & & & & 16.2578 & f@xmath130 & 5.1 & 0.6379 & 5.3 & + & & & & 8.0470 & f@xmath131 & 4.5 & 0.8268 & 4.7 & + & & & & 2.3825 & f@xmath132 & 5.2 & 0.7274 & 3.6 & + & & & & 4.3812 & f@xmath133 & 4.3 & 0.3942 & 4.2 & + & & & & 14.9483 & f@xmath134 & 3.9 & 0.8404 & 4.0 & + 4 & & 055600.99 + 323041.25 & 15.873 & 12.7176 & f@xmath121 & 14.4 & 0.0994 & 8.1 & + & & & & 7.8896 & f@xmath122 & 13.8 & 0.4901 & 6.0 & + & & & & 12.7698 & f@xmath123 & 10.5 & 0.0175 & 6.0 & close to f@xmath121 + & & & & 14.4564 & f@xmath124 & 9.6 & 0.0884 & 6.5 & + 5 & & 055527.32 + 322543.01 & 16.511 & 10.0341 & f@xmath121(=f@xmath132-f@xmath122 ) & 23.0 & 0.4259 & 137.9 & + & & & & 10.4706 & f@xmath122(=f@xmath132-f@xmath121 ) & 13.0 & 0.8171 & 77.9 & + & & & & 17.1389 & f@xmath123 & 8.7 & 0.9050 & 51.9 & + & & & & 14.0376 & f@xmath124 & 6.1 & 0.6801 & 33.6 & + & & & & 13.0968 & f@xmath125 & 3.7 & 0.1837 & 20.9 & + & & & & 12.8019 & f@xmath126 & 3.1 & 0.2056 & 17.4 & + & & & & 15.0841 & f@xmath127 & 2.1 & 0.7887 & 11.6 & + & & & & 13.6263 & f@xmath128 & 1.3 & 0.1260 & 7.1 & + & & & & 11.1091 & f@xmath129 & 1.4 & 0.2454 & 8.7 & + & & & & 27.1718 & f@xmath130 & 1.0 & 0.7296 & 6.7 & + rccrrrrrrc 1 & & 055122.14 + 322942.40 & 16.742 & 0.009 & 0.096 & 443.5 & 4730 & + 2 & & 055123.93 + 324357.22 & 15.837 & 0.003 & 0.017 & 14.0 & 4706 & + 3 & & 055124.73 + 324312.36 & 18.360 & 0.004 & 0.028 & 1.8 & 4730 & 13.260 + 4 & & 055125.42 + 323600.45 & 17.564 & 0.002 & 0.017 & 1.6 & 4730 & 21.199 + 5 & & 055127.04 + 323225.57 & 15.458 & 0.002 & 0.010 & 7.7 & 4592 & + 6 & & 055128.19 + 322931.14 & 17.854 & 0.004 & 0.029 & 3.4 & 4730 & + 7 & & 055129.26 + 323235.44 & 18.515 & 0.005 & 0.029 & 1.4 & 4730 & + 8 & & 055130.48 + 324053.05 & 18.348 & 0.005 & 0.033 & 1.9 & 4730 & 7.709 + 9 & & 055131.24 + 323059.61 & 17.688 & 0.008 & 0.046 & 12.9 & 4730 & + 10 & & 055131.87 + 324233.83 & 17.482 & 0.004 & 0.026 & 5.1 & 4730 & + rccrcrrrc 1 & & 055120.47 + 322200.75 & 20.690 & 1 & 53742.40307 & 0.093 & 0.17 & + & & & & 2 & 53742.46247 & 0.135 & 0.08 & + 2 & & 055121.46 + 322856.90 & 19.990 & 1 & 53730.34740 & 0.221 & 0.55 & + & & & & 2 & 53733.19093 & 0.176 & 0.21 & + & & & & 3 & 53735.19767 & 0.039 & 0.49 & + & & & & 4 & 53737.33376 & 0.188 & 0.30 & + 3 & & 055123.45 + 322948.83 & 20.809 & 1 & 53732.22482 & 0.061 & 0.32 & + & & & & 2 & 53732.35513 & 0.133 & 1.11 & + & & & & 3 & 53742.35427 & 0.281 & 0.23 & + 4 & & 055124.44 + 324407.71 & 15.143 & 1 & 53730.32306 & 0.010 & 1.47 & + 5 & & 055124.78 + 323440.56 & 18.368 & 1 & 53728.24480 & 0.026 & 0.17 & + & & & & 2 & 53742.44376 & 0.117 & 0.24 & + & & & & 3 & 53753.25302 & 0.139 & 0.16 & + 6 & & 055124.82 + 322937.32 & 20.017 & 1 & 53726.39533 & 0.046 & 0.10 & var + 7 & & 055125.41 + 323800.03 & 18.086 & 1 & 53726.27563 & 0.041 & 0.58 & var + & & & & 2 & 53727.30272 & 0.025 & 0.46 & + 8 & & 055125.71 + 323106.38 & 20.553 & 1 & 53734.23990 & 0.072 & 0.20 & var + & & & & 2 & 53756.08268 & 0.079 & 0.60 & + 9 & & 055126.78 + 324024.73 & 17.971 & 1 & 53726.43908 & 0.030 & 0.29 & + 10 & & 055127.86 + 322854.79 & 22.181 & 1 & 53727.19210 & 0.360 & 0.58 & var +
|
we present a comprehensive re - analysis of stellar photometric variability in the field of the open cluster m37 following the application of a new photometry and de - trending method to mmt / megacam image archive .
this new analysis allows a rare opportunity to explore photometric variability over a broad range of time - scales , from minutes to a month .
the intent of this work is to examine the entire sample of over 30,000 objects for periodic , aperiodic , and sporadic behaviors in their light curves .
we show a modified version of the fast @xmath0 periodogram algorithm ( f@xmath0 ) and change - point analysis ( cpa ) as tools for detecting and assessing the significance of periodic and non - periodic variations .
the benefits of our new photometry and analysis methods are evident .
a total of 2306 stars exhibit convincing variations that are induced by flares , pulsations , eclipses , starspots , and unknown causes in some cases .
this represents a 60% increase in the number of variables known in this field .
moreover , 30 of the previously identified variables are found to be false positives resulting from time - dependent systematic effects .
new catalog includes 61 eclipsing binary systems , 92 multiperiodic variable stars , 132 aperiodic variables , and 436 flare stars , as well as several hundreds of rotating variables .
based on extended and improved catalog of variables , we investigate the basic properties ( e.g. , period , amplitude , type ) of all variables .
the catalog can be accessed through the web interface ( http://stardb.yonsei.ac.kr/ ) .
|
omc-2 is an active star - forming region ( e.g. , @xcite ) in the orion a molecular cloud , located at a distance of 414 @xmath37 pc @xcite .
@xcite identified six bright mm / ir sources ( fir 1 - 6 ) within a region of about @xmath4 in size that have been associated with young stellar objects ( ysos ) through subsequent studies ( @xcite , @xcite and references therein ) .
the region has been imaged at mm and submm wavelengths by @xcite and @xcite , and in the near- and mid - ir by @xcite , @xcite , and @xcite . at mm and submm
wavelengths the brightest source is fir 4 , which has been associated with the hops 108 class 0 protostar ( @xcite , @xcite ) .
this source is connected through a filamentary cloud structure to the bright source fir 3 , also known as hops 370 , an intermediate - mass class i yso with an @xmath5 ( @xcite , @xcite ) located about @xmath6 to the ne ( see fig .
[ fig:2014 ] for the positions and nomenclature of the sources ) .
the region was observed with the very large array ( vla ) at 3.6 cm in the d configuration ( angular resolution @xmath7 ) by @xcite .
these authors found a weak elongated source ( vla 12 ) toward fir 4 ( hops 108 ) that was interpreted as a radio jet originating from this protostar .
a stronger , but unresolved source ( vla 11 ) , was detected toward fir 3 ( hops 370 ) .
the nature of the source fir 4 is still uncertain .
@xcite through mm and submm observations proposed that fir 4 was composed of 11 dusty cores .
these authors detected molecular line emission from shock tracers ( sio and ch@xmath8oh ) in the proximity of fir 4 and proposed that the interaction of a powerful molecular outflow driven by fir 3 ( e.g. , @xcite , @xcite ) with a dense clump could be triggering the formation of a next generation of stars at the position of fir 4 .
on the other hand , @xcite based on 2 mm observations proposed that fir 4 was composed of at least three cores , and suggested that its associated 3.6 cm source vla 12 traces an hii region photoionized by an early - type ( b3-b4 ) star with a luminosity of 700 - 1000 @xmath9 embedded within one of these cores .
a bolometric luminosity @xmath10 for fir 4 was obtained by @xcite but this estimate , based on low angular resolution iras data , is very uncertain because of the difficulties in separating the emission of the protostar from that of the surrounding molecular cloud and other neighboring objects . in a rigorous analysis of the available photometry of hops 108 , including herschel measurements that constrain better the peak of the spectral energy distribution ( sed ) than the iras fluxes , @xcite give @xmath11 .
this is much lower than the earlier estimate by @xcite or the value adopted by @xcite .
due to uncertainties in the amount of external heating and the inclination of the protostar , @xcite find that they can fit the sed with models having values for the intrinsic total luminosity ranging from 15 @xmath9 to 540 @xmath9 , although models with luminosities @xmath12 are favored .
@xcite detected far - ir co lines indicative of shock - heated gas at high temperatures ( @xmath13 k ) toward both hops 108 and hops 370 .
recently , @xcite imaged in far - ir [ oi ] lines and in the submm co(6 - 5 ) line , a powerful ( @xmath14 yr@xmath1 , as estimated by these authors ) bipolar jet / outflow originating from hops 370 .
the outflow extends to the ne of hops 370 ( fir 3 ) in the direction where a 4.5 @xmath15 m spitzer / irac image ( which traces shock - excited h@xmath16 emission ) shows an extended jet that terminates in a bow shock @xcite . to the sw
, the outflow terminates near the location of hops 108 , with a bright [ oi ] emission peak that @xcite argue originates in the terminal shock produced by the jet . in summary , @xcite , @xcite , and
@xcite suggest that the protostar hops 108 results from the interaction of an outflow driven by hops 370 with a dense clump . within this scenario ,
@xcite assume that a star cluster is being formed , with the radio source vla 12 tracing an hii region associated with the most massive star ; for this star they adopt a luminosity of @xmath01000 @xmath17 , similar to earlier estimates of @xcite . on the other hand , @xcite , @xcite , and @xcite
interpret vla 12 as a radio jet driven by a modest luminosity protostar associated with hops 108 . in this paper
we analyze new and archive vla observations that shed new light on the origin of the hops 108 protostar , the nature of its radio emission , and its relationship with other nearby sources .
the observations were carried out with the karl g. jansky vla of the national radio astronomy observatory ( nrao ) in the c configuration in 2014 ( project 14b-296 ) at c ( 5 cm ) , x ( 3 cm ) , k ( 1.3 cm ) and q ( 0.7 cm ) bands , and in the a configuration in 2015 ( project 15a-369 ) at c - band ( 5 cm ) . in both cases ,
the phase center was close to the position of hops 108 .
further details of the observational setup are given in table [ tab : observations ] .
the data were edited and calibrated using the common astronomy software applications ( casa ; version 4.2.2 ) package .
we also analyzed several epochs of vla archive data at x - band ( 3.6 cm ) obtained with different configurations between 1994 and 2000 ( see table [ tab : observations ] for details ) .
these data were edited and calibrated using the astronomical image processing system ( aips ) , and then were concatenated using the casa package .
all the images were made with the casa task clean using multi - frequency synthesis @xcite and fitting the frequency dependence of the emission with a taylor series expansion with nterms = 2 during the deconvolution .
there is extended emission of the bright hii region m43 ( ngc 1982 ) @xmath18 to the south of hops 108 that makes it difficult the imaging of the archive ( x - band ) data and the new data at c - band . in order to do a proper cleaning of our target sources we downweighted the extended emission by removing the shortest baselines . for the archive and the 2014 c - band data we used only baselines @xmath195 k@xmath20 and the multiscale deconvolution algorithm to make the images . for the 2015 a - configuration data , which are more sensitive
, we used only baselines @xmath1925 k@xmath20 .
in our new data of epochs 2014 and 2015 we detect the radio sources vla 11 , vla 12 , and vla 13 previously reported at 3.6 cm by @xcite , but our observations cover additional bands and are more sensitive , revealing further details .
the fwhm of our primary beam , centered near vla 12 , ranges from @xmath21 at q - band to @xmath22 at c - band .
therefore , we detected additional sources , including several of the ir sources in the field . in this paper
we will discuss only the sources detected in the proximity ( within @xmath23 ) of hops 108 .
positions of these sources are given in table [ tab : positions ] .
the remaining detected sources will be discussed in a forthcoming paper ( a. k. daz - rodrguez et al .
2017 , in preparation ) . in figure [ fig:2014 ]
we show the maps of vla 11 and vla 12 at different wavelengths obtained in our 2014 vla c - configuration observations .
vla 12 ( in the proximity of fir 4/hops 108 ) is clearly detected as a very elongated source at 5 and 3 cm , showing three knots that we call vla 12n , vla 12c , and vla 12s , for the northern , central and southern observed emission peaks , respectively . at first glance
, these results appear to confirm vla 12 as a radio jet , as was first suggested by @xcite .
these knots coincide with the southern , brightest part of the jet / outflow traced by the far - ir [ oi ] and submm co(6 - 5 ) lines , which is assumed to originate in hops 370 ( @xcite ; see fig .
[ fig : overlays ] , left ) .
the ir position of the hops 108 protostar ( @xcite ) falls close to ( within @xmath24 ) the position of knot vla 12c , apparently favoring hops 108 as the driving source of the proposed vla 12 radio jet . at shorter wavelengths ( 1.3 and 0.7 cm ; bottom panels in fig .
[ fig:2014 ] ) the emission of the vla 12 knots decreases making them undetectable , indicating that they have a negative spectral index @xmath25 ( where @xmath26 ) , characteristic of non - thermal emission .
this is confirmed by the analysis of the spectra obtained from our full dataset ( table [ tab : pmotions ] , fig .
[ fig : sed ] ) . despite the possible time variability both in flux density
( that can be more prominent in non - thermal emission ) and morphology , as well as the difficulties inherent in the measure of extended emission from data taken with different angular resolutions , the obtained spectral indices in the centimeter range are negative for the three knots of vla 12 , with values of @xmath271.1 , @xmath271.3 , and @xmath270.6 .
the behavior is clear even if only the c - configuration data from october 2014 , which were taken almost simultaneously , are considered .
also , we have checked in these data that the negative spectral indices obtained are not spurious results of differences in the uv coverage and/or angular resolution , that could make data at shorter wavelengths less sensitive to extended emission . to do
that , we obtained pairs of images of contiguous bands using only baselines in the same uvrange . using these pairs of maps we confirmed that the flux density truly decreases at shorter wavelengths , resulting in negative spectral indices for the vla 12 knots .
these negative spectral indices can not be explained by thermal emission ( see @xcite ) and should be attributed to non - thermal emission , probably arising from relativistic electrons accelerated in strong shocks ( @xcite ; see below ) .
this rules out the hii region scenario proposed by @xcite for vla 12 . although the position of the knot vla 12c at 6 and 3 cm appears somewhat ( @xmath28 ) displaced to the ne of the position of hops 108 ( as obtained from 24 and 8 data ; @xcite ) , at 1.3 and 0.7 cm the radio emission appears to be closer to the hops 108 ir position ( see fig . [ fig:2014 ] ) .
this shift in the position could indicate either an opacity gradient or that the radio emission observed at shorter wavelengths originates in a different object than the emission observed at longer wavelengths .
this last suggestion is confirmed by our data of subarcsecond angular resolution at long wavelengths ( fig .
[ fig : movprop ] ) that reveal two radio sources spatially separated by @xmath24 .
we identify the northeasternmost of these two radio sources with the vla 12c knot and the other source , whose position coincides with the ir source , with the radio counterpart of the hops 108 protostar .
recent alma observations with an angular resolution of 012 ( j. tobin et al . 2017b , in preparation )
reveal a compact source at 870 @xmath15 m whose position coincides within @xmath29 01 with the cm position ( table [ tab : pmotions ] ) confirming the nature of hops 108 as a protostar and its association with the cm radio source .
furthermore , the spectral index of hops 108 obtained from the cm data that separate its emission from that of vla 12c shows that hops 108 has a positive spectral index ( @xmath30 ; table [ tab : pmotions ] and fig .
[ fig : sed ] ) and , thus it is associated with thermal ( free - free ) emission , in contrast to the knots of vla 12 , whose emission is non - thermal . from the empirical correlation between cm flux density and bolometric luminosity for protostellar objects , @xmath31 @xcite , and using our measured flux density at 3 cm ( table [ tab : pmotions ] ) , we estimate a bolometric luminosity @xmath32 for hops 108 .
this value of the bolometric luminosity is close to the lowest end of the range considered by @xcite in the modeling of this object , being much smaller than early estimates ( @xmath33 ; @xcite ) that were based on very low angular resolution data , and significantly smaller than the value obtained if the contribution of the vla 12 knots is not removed ( @xmath34 ; @xcite ) .
thus , separating the emission of hops 108 from that of the vla 12 knots confirms hops 108 as a low - luminosity object .
the radio source vla 11 , that was associated with fir 3 ( hops 370 ) and reported as unresolved by @xcite , is clearly resolved as a very elongated bipolar radio jet along a pa @xmath35 in our a - configuration data at 5 cm ( fig .
[ fig : movprop ] ) .
the radio source is also resolved in the 3.6 cm image obtained by combining archive data from 1994 - 2000 .
the spectral index is @xmath36 ( table [ tab : pmotions ] , fig .
[ fig : sed ] ) , which is a reasonable value for a thermal free - free radio jet @xcite . hops 370 is known to be associated with an outflow extending along a pa similar to that of the radio jet ( @xcite ; see fig .
[ fig : overlays ] , left ) .
the morphology of vla 11 appears to change in the 0.7 cm image ( fig .
[ fig:2014 ] ) , that shows weak extended emission in a direction perpendicular to the jet , suggesting a noticeable dust contribution at this wavelength . to prevent dust contamination ,
the 0.7 cm datapoint has been excluded in the fits to determine the spectral index of this and the remaining thermal sources . in the lower resolution images at 6 and 3
cm ( fig .
[ fig:2014 ] ) , vla 11 appears blended with the radio emission of mips 2301 ( @xcite , an ir source also known as mir 22 , @xcite ) located @xmath37 south of vla 11 .
mips 2301 is well separated , but only marginally detected , in the higher resolution 3.6 cm image shown in figure [ fig : movprop ] and in the a - configuration images at 5 cm obtained with the robust weighting parameter set to values @xmath19 1 ( not shown in fig . [ fig : movprop ] ) ) .
thus , its radio emission can be distinguished from the knots of the radio jet in the high angular resolution images . ] . with the data currently available
we obtain a positive spectral index , @xmath38 ( table [ tab : pmotions ] , fig .
[ fig : sed ] ) , consistent with thermal free - free emission from a yso .
more sensitive observations are required for a better characterization of the nature of the radio emission of this object .
the source mips 2297 ( @xcite , also known as mir 23 , @xcite ) is marginally detected at 1.3 cm and 0.7 cm ( fig .
[ fig:2014 ] ) , but clearly detected in the 0.9 cm observations of j. tobin et al .
( 2017a , in preparation ) .
it shows a positive spectral index ( @xmath39 ) in the cm regime ( table [ tab : pmotions ] and fig .
[ fig : sed ] ) , confirming its nature as a yso .
so far , mips 2297 has not been studied in great detail , but it is classified as a young star with disk ( class ii ) by @xcite .
we also detect hops 64 ( fig .
[ fig:2014 ] ) , a class i object @xcite that in the cm range presents a positive spectral index of @xmath40 ( table [ tab : pmotions ] , fig .
[ fig : sed ] ) , suggesting optically - thick free - free emission from a yso .
the position of the radio source falls in between the fir 4d and fir 4e cores imaged by @xcite , and is close to , but slightly displaced from , one of the main emission peaks of the dust cloud imaged by alma at 3 mm ( @xcite ; see fig .
[ fig : overlays ] , right ) . nevertheless , the position of the radio source coincides accurately ( within @xmath29 01 ) with that of a compact 870 @xmath15 m source recently detected with alma ( j. tobin et al . 2017b , in preparation ) .
a few arcsec to the sw of the knot vla 12s we detect a weak source .
the source has not been reported previously at other wavelengths and we call it vla 16 .
the detection is marginal in the individual bands ( @xmath0 5@xmath41 at most ) but we consider that the source is real because it appears in four of the five observed bands ( see table [ tab : pmotions ] ) .
the spectral index is @xmath0 0.7 ( fig .
[ fig : sed ] ) , consistent with that of a yso / jet .
the position of the radio source falls near a secondary emission peak in the 3 mm alma image @xcite shown in figure [ fig : overlays ] and coincides very well ( within @xmath29 01 ) with the position of a weak 870 @xmath15 m source recently detected with alma observations of higher angular resolution ( j. tobin et al .
2017b , in preparation ) . from the empirical correlation between cm emission and bolometric luminosity @xcite
we infer @xmath42 .
all these results confirm the reality of the source vla 16 and its nature as an embedded low - luminosity yso . finally , we report the detection in all the observed bands of a compact source , located @xmath43 sw of vla 12s , and that we call vla 15 ( tables [ tab : positions ] , [ tab : pmotions ] , figs .
[ fig:2014 ] , [ fig : sed ] , [ fig : movprop ] ) .
the position of the source coincides with the fir 4j core imaged by @xcite and falls close to ( but slightly offset from ) one of the main emission peaks of the dust cloud imaged at 3 mm by alma ( @xcite ; see fig .
[ fig : overlays ] , right ) .
no counterparts at other wavelengths had been reported so far , but the source has been recently imaged at 870 @xmath15 m with alma ( j. tobin et al . 2017b , in preparation ) .
the radio source has a positive spectral index ( @xmath44 ) and we suggest that it traces a very young embedded stellar object associated with the omc-2 star - forming region that deserves further study at other wavelengths .
we searched for proper motions by comparing the positions of the radio sources detected in the high - angular resolution maps obtained from archive data ( epochs 1994 - 2000 ) and from our last observations ( epoch 2015 ) .
unfortunately , the sensitivity of the archive data in a single epoch was insufficient , and we combined in the same map data taken over @xmath06 years , which is not optimal to measure accurate proper motions , because of motions and changes in morphology of the sources over this period .
also , the archive data were observed in x - band ( 8.4 - 8.5 ghz ) while the new data have been obtained in c - band ( 4 - 8 ghz ) ( table [ tab : observations ] ) .
although the frequency ranges of the two bands are almost contiguous , the differences in frequency could result in apparent motions due to variations of opacity with frequency , particularly in extended sources .
despite all these factors that contribute to the uncertainty in our measurements , the relatively high velocity of the jet and long time ( @xmath45 yr ) elapsed between the two maps , make the proper motions clearly detectable .
our results show that , while vla 11 , and vla 15 remain static within the uncertainties , the vla 12n , vla 12c and vla 12s knots present measurable proper motions of @xmath0 40 - 100 km s@xmath1 roughly in the s - sw direction ( table [ tab : pmotions ] and fig .
[ fig : movprop ] ; see also left panel in fig .
[ fig : overlays ] ) .
hops 108 shows a measurable proper motion ( @xmath46 km s@xmath1 ) , but in the ne direction , whose nature is uncertain ( see below ) .
the proper motion velocity of vla 11 , the source with the best signal - to - noise ratio , is @xmath47 km s@xmath1 ( table [ tab : pmotions ] ) .
since this velocity is obtained from the absolute positions of the source , its small value indicates that systematic errors are effectively very small and that their possible residual effects on the proper motions determination are at a level of @xmath0 2 km s@xmath1 or smaller . as discussed above ,
the knots of vla 12 are characterized by negative spectral indices , indicating non - thermal emission ( see fig . [
fig : sed ] ) , but their overall morphology with three aligned knots , suggests that they pertain to a radio jet .
radio jets associated with ysos are characterized by thermal ( free - free ) emission ( e.g. , @xcite ) , but in a few cases , non - thermal emission has been found in the jet lobes , at relatively large distances from the jet core that shows a positive spectral index and thermal emission ( @xcite , @xcite , @xcite and references therein ) .
the most spectacular case of this selected sample of non - thermal radio jets is the hh 80 radio jet , where linearly polarized emission , indicative of synchrotron radiation , was imaged in the jet lobes @xcite , while the central core of the jet , with a positive ( thermal ) spectral index , remains unpolarized .
as the velocities of protostellar jets are not relativistic ( @xmath48 km s@xmath1 ) , it is assumed that the synchrotron emitting electrons are accelerated up to relativistic velocities via the diffusive shock acceleration ( dsa ) mechanism @xcite working in the strong shocks present in the interaction of the jet with the surrounding medium .
thus , it is expected that the vla 12 knots are associated with strong shocks , as indeed is suggested by the detection of several shock tracers in the proximity of this source ( @xcite , @xcite , @xcite ; see fig .
[ fig : overlays ] , left panel ) .
actually , the region near hops 108 shows the brightest far - ir [ oi ] , co , h@xmath16o and oh line emission among the hops spectroscopic sample of sources ( p. manoj et al .
2017 , in preparation ) , indicating the presence of strong shocks .
these results are consistent with the vla 12 knots being part of a non - thermal radio jet either originating from the class 0 protostar hops 108 ( fir 4 ) , which falls in between the vla 12c and the vla 12s knots , or from the intermediate - mass class i yso hops 370 ( fir 3 ) , as the vla 12 knots are roughly aligned in the direction of the vla 11 radio jet associated with hops 370 ( fig .
[ fig:2014 ] ) and fall on the sw lobe of the submm and far - ir outflow ( @xcite , @xcite ) driven by this object .
if hops 108 was the driving source of the vla 12 knots , one would expect proper motions of these knots away from hops 108 .
in particular , the vla 12n and 12c knots would show proper motions with a component pointing to the north .
however , as figure [ fig : movprop ] shows , the proper motions of the vla 12 knots point to the south , consistent with an origin from hops 370 ( vla 11 ) , and excluding hops 108 as the driving source .
thus , we conclude that the vla 12 non - thermal radio knots belong to a radio jet , driven by the intermediate - mass class i object hops 370 , whose central thermal region is traced by the collimated radio source vla 11 .
vla 11 is bipolar , extending @xmath49 both to the ne and to the sw of the position of the protostar ( see top - right panel in fig .
[ fig : movprop ] ) .
however , only the sw side of the radio jet shows a distant ( @xmath50 ) non - thermal lobe ( vla 12 ) , without a ne counterpart .
the absence of a ne non - thermal radio lobe is probably due the lack of a dense clump along the ne jet path where a strong shock interaction , similar to that observed in the sw lobe , could take place .
the jet is also bipolar in the far - ir [ oi ] lines @xcite , but its sw lobe is more extended , with the brightest part associated with the non - thermal vla 12 radio lobe . adopting 1997.65 as an average epoch for the archive data map , the resulting plane - of - the - sky velocities of the vla 12n , 12c , and 12s knots are @xmath0100 , 101 , and 37 km s@xmath1 , respectively .
the velocities of knots 12n and 12c fall in the range of values measured for other protostellar radio jets ( @xcite and references therein ) ; the velocity of vla 12s is smaller , suggesting that this distant knot could have suffered a significant deceleration , particularly if the jet has interacted strongly with the ambient medium , as indicated by other observations ( @xcite , @xcite ) . in this respect ,
recent proper motion measurements of the radio knots associated with the herbig - haro objects hh 80 , hh 81 , and hh 80n , far away from the central source , have shown that these knots have significantly slower motions than the radio knots located a few arcsec from the central source @xcite .
this result has been interpreted as indicating strong jet interactions with the ambient cloud that significantly slow down the jet material .
since the position of the hops 108 protostar falls in the path of the vla 11-vla 12 radio jet , but it can be ruled out as its driving source , an appealing alternative possibility is that the jet is related to the origin of hops 108 .
it has been proposed that shocks associated with the interaction of jets with the surrounding medium could compress the gas and induce local instabilities that may trigger star - formation ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
in particular , this scenario has been proposed for hops 108 , as it is associated with strong shocks ( @xcite , @xcite ) and probably interacting with a nearby dense core @xcite .
interestingly , the shape of the jet in the region of the vla 12 knots appears to follow the eastern edge of a clump of enhanced dust emission , as imaged by alma at 3 mm ( @xcite ; see fig .
[ fig : overlays ] , right panel ) .
thus , it seems plausible that the formation of the hops 108 protostar was triggered by the compression of the material in this clump after a strong shock interaction with the vla 11-vla 12 jet , driven by hops 370 .
the evolutionary status of hops 370 and hops 108 are also fully consistent with the suggestion of the latter being triggered by the former .
hops 370 , with a bolometric temperature @xmath51 k , has been classified as an intermediate - mass young class i object ( @xcite , @xcite ) and we would expect an age ( time elapsed since the onset of collapse ) of a few times @xmath52 years for this object .
hops 108 , with a lower bolometric temperature @xmath53 k , is supposed to be younger , and it has been classified as a class 0 protostar @xcite .
we can then hypothesize that some @xmath52 years ago the jet of hops 370 started to impact the molecular clump ( see fig .
[ fig : overlays ] , right panel ) from which hops 108 later formed . after an interaction of a few times @xmath54 years ,
enough time had passed to allow the collapse of the clump and the start of the protostellar stage of hops 108 .
the travel time of the gas from hops 370 to hops 108 is only 500 years ( as derived from our measured proper motions ) and the delay in the influence of the first object on the second can be taken to be much shorter than the other timescales .
we then have a scenario in which the evolutionary status of the sources is consistent with one of them triggering the formation of the other .
it should be noted that the mips 2297 source also falls on the path of the hops 370 jet , but closer to the origin . since mips
2297 seems to be older than hops 108 one could think that it is the result of an earlier episode of triggered star - formation . if mips 2297 is really a class ii object , as proposed by @xcite , it would be even older than hops 370 , making the possibility of triggered formation inviable .
however , mips 2297 is not well studied yet , and if it happens to be younger this possibility can not be fully discarded . on the other hand , @xcite proposed that the formation of hops 64 ( @xmath55 nw from hops 108 ) could have also been triggered by the outflow from hops 370 . however , in the recent study by @xcite , both hops 370 and hops 64 are classified as class i sources , suggesting a similar evolutionary stage , which would be in conflict with a triggered star - formation scenario for hops 64 .
it is also possible that mips 2297 and/or hops 64 , which are both visible at shorter wavelengths , lie in front of the plane where hops 370 and hops 108 are located ; so , they could be close to hops 370 and its jet just in projection , but in reality be farther away and have formed in omc-2 independently of this source .
the new source vla 16 also falls in the path of the jet , very close to the southernmost detected knot , vla 12s .
we can also not exclude the possibility that star - formation has been triggered , or will eventually be triggered , in other dust cores reported by @xcite which currently do not show evidence of embedded ysos . thus , there are at least four young stars which could be the result of triggering or affected by the hops 370 jet , but only for hops 108 there is further evidence of ongoing shock - interaction .
a puzzling issue in the triggered - formation scenario for hops 108 is its large proper motion velocity ( @xmath0 32 km s@xmath1 ; table [ tab : pmotions ] ) .
such a large velocity implies that it should have been acquired very recently ( @xmath56 200 yr ) in order the star to be formed in the proximity ( within a few arcsec ) of its current position , in the region where the shock emission has been identified @xcite . otherwise ,
if this velocity was maintained during a longer time interval , it would imply that the hops 108 protostar originated at a distant position from its current location , making inviable the triggering hypothesis .
relatively low velocities are expected if the protostars are mechanically linked to the molecular gas reservoir and thus obtain their velocities directly from the gas they are forming in , as proposed by @xcite , who found typical radial velocities with respect to the ambient cloud of @xmath0 0.6 km s@xmath1 for protostars and
@xmath0 1.8 km s@xmath1 for pre - main - sequence stars in the integral - shaped filament ( isf ) region in orion .
these authors propose a slingshot - like ejection mechanism to account for the increase in velocity of pre - main - sequence stars relative to protostars .
however , this mechanism can not explain velocities above a few km s@xmath1 .
in contrast , hops 108 presents a much larger proper motion velocity , similar to the @xmath0 27 km s@xmath1 of the bn object in the nearby orion bn / kl stellar system @xcite .
another runaway star , v 1326 ori , with a peculiar large proper motion velocity of about 28 km s@xmath1 has been recently identified in a radio survey of proper motions in the core of the orion nebula cluster ( @xcite ) . for the bn / kl system , in order to explain both the uncollimated explosive molecular outflow ( @xcite ) and the stellar velocities , violent ( proto-)stellar interactions and the subsequent dynamical ejection of said ( proto-)stars have been invoked @xcite .
this acceleration mechanism would require the presence of a binary or multiple stellar system near fir 4/hops 108 to make it compatible with the triggering scenario .
@xcite imaged 11 potential cores within a region of @xmath57 in the proximity of fir 4 and interpreted them as tracing a young stellar cluster whose formation was triggered by the impact of the jet with the ambient cloud .
if this was the case , interactions within this multiple system could have resulted in the recent ejection of hops 108 .
we identify the region imaged by @xcite with the clump of dust imaged by alma at 3 mm ( @xcite ; fig .
[ fig : overlays ] , right panel ) , where the sources hops 108 , hops 64 , vla 15 and vla 16 appear to be embedded .
since hops 108 moves toward the ne , we do not find a suitable candidate that might be receding from a close enough common position and that could be responsible for a past interaction with hops 108 , casting doubt on this possibility .
there are other detected radio sources in the region whose proper motions are compatible with being receding from a common location with hops 108 ( e.g. , vla 13 , located @xmath58 to the sw ; a. k. d az - rodrguez et al .
2017 , in preparation ) , and that could have dynamically interacted with it in the past . however , the location of a potential encounter would fall at a large distance from the current position of hops 108 and , as noted above , would make the origin of this object incompatible with a scenario of ( local ) triggered star - formation . on the other hand , if hops 108 was accelerated in a distant encounter it seems unlikely that our observation occurs just at the moment when it crosses the vla 12 knots of the hops 370 jet , at the point of its maximum interaction with the ambient cloud .
another possibility to explain the observed large proper motion of hops 108 is to assume that it is due to a change in the shape of its radio emission because of blending with the emission of the nearby vla 12 knots .
since these knots move toward the sw , with vla 12c approaching hops 108 and vla 12s going away , a partial blending with these knots can result in an apparent motion of hops 108 toward the ne .
also , a one - sided ejecta of a high velocity cloud of ionized plasma by the protostar could produce a change in the shape of its radio emission and an apparent proper motion .
one - sided high velocity ejecta from young stars have been observed in several other sources ( e.g. , @xcite and references therein ) .
such an ejecta with a velocity of the order of hundreds of km s@xmath1 in a poorly resolved source would result in a one - sided distortion in the source shape and thus a change in the centroid position that could mimic a motion of several tens of km s@xmath1 .
wavelength - dependent opacity effects in an extended source could emphasize these changes in the position of the centroid of the emission .
indeed , we found that the precise position of the hops 108 radio source changes with wavelength .
additional , sensitive high angular resolution observations could provide a more precise measure of the hops 108 proper motions and reveal the possible presence of either a high velocity ejecta from this source or a nearby binary / multiple system whose past interaction with hops 108 had been responsible for its current high proper motions .
we analyzed new ( 2014 - 2015 ) multiwavelength ( 0.7 - 5 cm ) , multiconfiguration vla observations together with archive ( 1994 - 2000 ) data at 3.6 cm that allowed us to obtain spectral indices and proper motions of radio sources in the region enclosing fir 3 ( hops 370 ) and fir 4 ( hops 108 ) in omc-2 .
our main conclusions can be summarized as follows : 1 .
we detect radio emission from the far - ir sources hops 370 , hops 108 , and hops 64 , as well as from the mid - ir sources mips 2297 and mips 2301 .
we also detect two new sources , both associated with dust cores , that we call vla 15 and vla 16 .
for all these sources we obtain positive spectral indices in the centimeter wavelength range , consistent with thermal free - free emission from young stellar objects
the radio source vla 11 , associated with hops 370 , presents a clearly elongated bipolar morphology in the ne - sw direction . because of this morphology and positive spectral index ( @xmath59 0.3 ) we interpret vla 11 as the thermal region closest to the origin of a well collimated bipolar radio jet driven by the hops 370 intermediate - mass protostar .
we detect several knots of emission ( vla 12n , 12c , and 12s ) at @xmath07,500 - 12,500 au to the sw of hops 370 with negative spectral indices and showing proper motions of @xmath040 - 100 km s@xmath1 away from the hops 370 protostar .
we interpret these knots as a non - thermal lobe of the hops 370 jet .
the vla 12 knots are found in a region where previous observations have detected several tracers of strong shocks , suggesting that their non - thermal emission is likely synchrotron emission from relativistic electrons accelerated in shocks , as has been proposed for other non - thermal jet lobes .
although vla 11 is bipolar , a similar non - thermal radio lobe is not found to the ne of hops 370 , probably because there is not a dense clump where the jet interaction could take place .
3 . hops 108 is identified as a compact radio source independent and kinematically separated from the radio emission of vla 12 knots of the hops 370 jet .
its position along the path of the vla 12 knots , and coincidence with different shock tracers , suggest a scenario where the formation of hops 108 has been triggered by the interaction of the hops 370 jet with the surrounding medium , as was already proposed by @xcite .
the more advanced evolutionary stage of hops 370 relative to hops 108 and the short dynamical timescale of the jet , are consistent with this scenario .
however , hops 108 presents a large proper motion velocity of about 30 km s@xmath1 toward the ne whose nature is uncertain .
this velocity is similar to the proper motion velocities found in runaway stellar sources in orion @xcite .
such a large velocity in hops 108 would be inconsistent with the triggered scenario , unless the source had been accelerated to this high velocity very recently .
alternatively , an apparent proper motion could result because of a change in the position of the centroid of the source due to a partial blending with the emission of the nearby vla 12 knots or to a one - sided ejecta of ionized plasma , rather than by the actual motion of the protostar itself .
deep high angular resolution observations at several epochs are required to clarify this issue .
4 . hops 370 and vla 15 do not show detectable proper motions ( @xmath60 2 - 4 km s@xmath1 ) , as expected for embedded protostars .
a.k.d.r . acknowledges an spanish mecd fpu fellowship .
, a.k.d.r . , j.f.g . , and
acknowledge support from mineco ( spain ) aya2014 - 57369-c3 - 3-p grant ( co - funded by feder ) .
the authors thank lvaro sanchz - monge for helpful suggestions .
c@c@ c@c@c@ cccc@c@cc@c[h ! ] 1994 nov 17 & c & x & 8.4649 & 0.100 & ar0323s & 05 35 25.825 & @xmath2705 09 51.38 & 3c286 & 5.06 & j0541 - 0541 & [email protected] & 0.5 + 1998 jan 13 & d & x & 8.4851 & 0.100 & ar0387 & 05 35 24.220 & @xmath2705 10 07.27 & 3c48 & 3.15 & j0541 - 0541 & [email protected] & 1.0 + 2000 jan 14 & b & x & 8.4851 & 0.100 & ar0411 & 05 35 26.970 & @xmath2705 10 01.20 & 3c48 & 3.28 & j0541 - 0541 & [email protected] & 3.2 + 2014 oct 16 & c & k & 22.000 & 8.048 & am1313 & 05 35 27.070 & @xmath2705 10 00.60 & 3c147 & 2.78 & j0541 - 0541 & [email protected] & 1.4 + 2014 oct 16 & c & q & 44.063 & 8.048 & am1313 & 05 35 27.070 & @xmath2705 10 00.60 & 3c147 & 0.91 & j0541 - 0541 & [email protected] & 0.6 + 2014 oct 18 & c & c & 6.000 & 4.048 & am1313 & 05 35 27.070 & @xmath2705 10 00.60 & 3c147 & 7.94 & j0541 - 0541 & [email protected] & 0.5 + 2014 oct 18 & c & x & 10.000 & 4.048 & am1313 & 05 35 27.070 & @xmath2705 10 00.60 & 3c147 & 4.84 & j0541 - 0541 & [email protected] & 0.5 + 2015 aug 06 & a & c & 6.224 & 3.998 & ao0316 & 05 35 27.080 & @xmath2705 10 00.30 & 3c147 & 7.94 & j0541
- 0541 & [email protected] & 0.5 + 2015 sep 11 & a & c & 6.224 & 3.998 & ao0316 & 05 35 27.080 & @xmath2705 10 00.30 & 3c147 & 7.94 & j0541 - 0541 & [email protected] & 0.5 + ccccll vla 11 & fir 3 , hops 370 , mips 2302 , mir 21 & class i yso & 1&05 35 [email protected] & @xmath2705 09 [email protected] + mips 2301 & mir 22 & yso & 2 & 05 35 [email protected] & @xmath2705 09 [email protected] + mips 2297 & mir 23 & class ii yso & 2 & 05 35 [email protected] & @xmath2705 09 [email protected] + vla
12n & & jet knot & 3 & 05 35 [email protected] &
@xmath2705 09 [email protected] + hops 64 & mips 2293 , mir 24 , fir 4d / e & class i yso & 1 & 05 35 [email protected] & @xmath2705 09 [email protected] + vla 12c & & jet knot & 3 &
05 35 [email protected] & @xmath2705 09 [email protected] + hops 108 & fir 4 , mips 2289 & class 0 yso & 1 & 05 35 [email protected] & @xmath2705 09 [email protected] + vla 12s & & jet knot & 3 & 05 35 [email protected] & @xmath2705 10 [email protected] + vla 16 & & yso & 3 & 05 35 [email protected] & @xmath2705 10 [email protected] + vla 15 & fir 4j & yso & 3 & 05 35 [email protected] & @xmath2705 10 [email protected] + ccccccccccc vla 11 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 2@xmath32 & 50@xmath350 + mips 2301 & [email protected] & [email protected] & @xmath29 0.03 & @xmath29 0.03 & @xmath29 0.07 & [email protected] & & & & + mips 2297 & @xmath29 0.03 & @xmath29 0.03 & [email protected] & [email protected] & [email protected] & [email protected] & & & & +
vla 12n & [email protected] & [email protected] & @xmath29 0.03 & @xmath29 0.025 & @xmath29 0.05 & @[email protected] & 22@xmath311 & @xmath2746@xmath311 & 100@xmath321 & 155@xmath312 + hops 64 & [email protected] & @xmath29 0.03 & [email protected] & [email protected] & [email protected] & [email protected] & & & & + vla 12c & [email protected] & [email protected] & @xmath29 0.03 & @xmath29 0.024 & @xmath29 0.04 & @[email protected] & @xmath2711@xmath314 & @xmath2750@xmath315 & 100@xmath330 & 192@xmath316 + hops 108 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 14@xmath37 & 9@xmath37 & 32@xmath314 & 57@xmath325 + vla 12s & [email protected] & [email protected] & [email protected] & [email protected] & @xmath29 0.04 & @[email protected] & @xmath2713@xmath33 & @xmath2714@xmath33 & 37@xmath36 & 223@xmath310 + vla 16 & [email protected] & @xmath29 0.03 & [email protected] & [email protected] & [email protected] & [email protected] & & & & + vla 15 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & @xmath271@xmath33 & 2@xmath33 & 4@xmath36 & 340@xmath390 + 3 , 3 , 6 , 9 , 12 , 15 , 20 , 30 , 50 , 70 , 100 , 150 and 200 times the rms noise of each map .
the robust weighting parameter @xcite was set to 0 .
the synthesized beam is shown at the bottom left corner of each map .
the 10.4 positions ( @xcite ) are indicated by red plus signs whose size corresponds to the positional uncertainty . for hops 108 , that is not reported at 10.4 , the 24 position @xcite is given .
top left : c band ( 5 cm ) ; rms = 11 @xmath15jy beam@xmath1 ; hpbw = 475@xmath61246 , p.a .
top right : x band ( 3 cm ) ; rms = 9 @xmath15jy beam@xmath1 ; hpbw = 268@xmath61173 , p.a .
bottom left : k band ( 1.3 cm ) ; rms = 10 @xmath15jy beam@xmath1 ; hpbw = 084@xmath61071 , p.a .
bottom right : q band ( 0.7 cm ) map ; rms = 15 @xmath15jy beam@xmath1 ; hpbw = 046@xmath61036 , p.a .
= @xmath2720 .
[ fig:2014],title="fig:",scaledwidth=49.0% ] 3 , 3 , 6 , 9 , 12 , 15 , 20 , 30 , 50 , 70 , 100 , 150 and 200 times the rms noise of each map .
the robust weighting parameter @xcite was set to 0 .
the synthesized beam is shown at the bottom left corner of each map .
the 10.4 positions ( @xcite ) are indicated by red plus signs whose size corresponds to the positional uncertainty . for hops 108 , that is not reported at 10.4 , the 24 position @xcite is given .
top left : c band ( 5 cm ) ; rms = 11 @xmath15jy beam@xmath1 ; hpbw = 475@xmath61246 , p.a .
top right : x band ( 3 cm ) ; rms = 9 @xmath15jy beam@xmath1 ; hpbw = 268@xmath61173 , p.a .
bottom left : k band ( 1.3 cm ) ; rms = 10 @xmath15jy beam@xmath1 ; hpbw = 084@xmath61071 , p.a .
bottom right : q band ( 0.7 cm ) map ; rms = 15 @xmath15jy beam@xmath1 ; hpbw = 046@xmath61036 , p.a .
= @xmath2720 .
[ fig:2014],title="fig:",scaledwidth=49.0% ] 3 , 3 , 6 , 9 , 12 , 15 , 20 , 30 , 50 , 70 , 100 , 150 and 200 times the rms noise of each map .
the robust weighting parameter @xcite was set to 0 .
the synthesized beam is shown at the bottom left corner of each map .
the 10.4 positions ( @xcite ) are indicated by red plus signs whose size corresponds to the positional uncertainty . for hops 108 , that is not reported at 10.4 , the 24 position @xcite is given .
top left : c band ( 5 cm ) ; rms = 11 @xmath15jy beam@xmath1 ; hpbw = 475@xmath61246 , p.a .
top right : x band ( 3 cm ) ; rms = 9 @xmath15jy beam@xmath1 ; hpbw = 268@xmath61173 , p.a .
bottom left : k band ( 1.3 cm ) ; rms = 10 @xmath15jy beam@xmath1 ; hpbw = 084@xmath61071 , p.a .
bottom right : q band ( 0.7 cm ) map ; rms = 15 @xmath15jy beam@xmath1 ; hpbw = 046@xmath61036 , p.a .
= @xmath2720 .
[ fig:2014],title="fig:",scaledwidth=49.0% ] 3 , 3 , 6 , 9 , 12 , 15 , 20 , 30 , 50 , 70 , 100 , 150 and 200 times the rms noise of each map .
the robust weighting parameter @xcite was set to 0 .
the synthesized beam is shown at the bottom left corner of each map .
the 10.4 positions ( @xcite ) are indicated by red plus signs whose size corresponds to the positional uncertainty .
for hops 108 , that is not reported at 10.4 , the 24 position @xcite is given .
top left : c band ( 5 cm ) ; rms = 11 @xmath15jy beam@xmath1 ; hpbw = 475@xmath61246 , p.a .
top right : x band ( 3 cm ) ; rms = 9 @xmath15jy beam@xmath1 ; hpbw = 268@xmath61173 , p.a .
bottom left : k band ( 1.3 cm ) ; rms = 10 @xmath15jy beam@xmath1 ; hpbw = 084@xmath61071 , p.a .
bottom right : q band ( 0.7 cm ) map ; rms = 15 @xmath15jy beam@xmath1 ; hpbw = 046@xmath61036 , p.a .
= @xmath2720 .
[ fig:2014],title="fig:",scaledwidth=49.0% ] 63 @xmath15 m line emission ( color scale , @xcite ) overlaid on the 3 cm emission vla map ( contours , see fig . [ fig:2014 ] ) with proper motions indicated by arrows .
both images are tracing the jet ejected from hops 370 .
note that the [ oi ] 63 @xmath15 m emission peaks close to the position of hops 108 , suggesting that this protostar has been formed in a region where a strong shock interaction between the jet and the ambient medium is taking place . plus
signs indicate the positions of ir sources .
right panel : alma image of the dust emission at 3 mm ( color scale , @xcite ) overlaid on the vla maps at 3 cm ( black contours ) and 0.7 cm ( blue contours , see fig . [ fig:2014 ] ) .
note that the radio jet at 3 cm appears to delineate the eastern edge of the dust clump traced by the alma image .
[ fig : overlays],title="fig:",scaledwidth=47.0% ] 63 @xmath15 m line emission ( color scale , @xcite ) overlaid on the 3 cm emission vla map ( contours , see fig . [ fig:2014 ] ) with proper motions indicated by arrows .
both images are tracing the jet ejected from hops 370 .
note that the [ oi ] 63 @xmath15 m emission peaks close to the position of hops 108 , suggesting that this protostar has been formed in a region where a strong shock interaction between the jet and the ambient medium is taking place . plus
signs indicate the positions of ir sources .
right panel : alma image of the dust emission at 3 mm ( color scale , @xcite ) overlaid on the vla maps at 3 cm ( black contours ) and 0.7 cm ( blue contours , see fig .
[ fig:2014 ] ) .
note that the radio jet at 3 cm appears to delineate the eastern edge of the dust clump traced by the alma image .
[ fig : overlays],title="fig:",scaledwidth=53.0% ] .
upper limits are represented by arrows where the central symbol is at 3@xmath41 .
dashed lines represent least - squared fits . to avoid possible dust contamination
the 0.7 cm datapoint has been excluded in the fitting for the positive spectral index sources .
upper limits have been taken into account only when they constrain the fit .
[ fig : sed ] ] ) .
second epoch maps ( c - band ; right panels ) were obtained from the 2015 data .
different weightings have been used to better emphasize the emission of each source .
synthesized beams of @xmath62 and rms of @xmath0 10 @xmath15jy beam@xmath1 were obtained for the first epoch maps , while synthesized beams of @xmath63-@xmath64 , and rms of 3 - 8 @xmath15jy beam@xmath1 were obtained for the second epoch maps .
contour levels are @xmath274 , @xmath273 , 3 , 4 , 5 , 6 , 7 , 8 , 10 , 12 , 15 , 20 , 35 , 60 , 100 , and 150 times the rms noise of each map . positions of the sources derived from elliptical gaussian fits to the first epoch maps are marked with blue plus signs , whose sizes represent their positional errors .
arrows indicate the displacement between the two epochs .
[ fig : movprop],title="fig:",scaledwidth=38.0% ] ) .
second epoch maps ( c - band ; right panels ) were obtained from the 2015 data .
different weightings have been used to better emphasize the emission of each source .
synthesized beams of @xmath62 and rms of @xmath0 10 @xmath15jy beam@xmath1 were obtained for the first epoch maps , while synthesized beams of @xmath63-@xmath64 , and rms of 3 - 8 @xmath15jy beam@xmath1 were obtained for the second epoch maps .
contour levels are @xmath274 , @xmath273 , 3 , 4 , 5 , 6 , 7 , 8 , 10 , 12 , 15 , 20 , 35 , 60 , 100 , and 150 times the rms noise of each map . positions of the sources derived from elliptical gaussian fits to the first epoch maps are marked with blue plus signs , whose sizes represent their positional errors .
arrows indicate the displacement between the two epochs .
[ fig : movprop],title="fig:",scaledwidth=38.0% ] ) .
second epoch maps ( c - band ; right panels ) were obtained from the 2015 data .
different weightings have been used to better emphasize the emission of each source .
synthesized beams of @xmath62 and rms of @xmath0 10 @xmath15jy beam@xmath1 were obtained for the first epoch maps , while synthesized beams of @xmath63-@xmath64 , and rms of 3 - 8 @xmath15jy beam@xmath1 were obtained for the second epoch maps .
contour levels are @xmath274 , @xmath273 , 3 , 4 , 5 , 6 , 7 , 8 , 10 , 12 , 15 , 20 , 35 , 60 , 100 , and 150 times the rms noise of each map . positions of the sources derived from elliptical gaussian fits to the first epoch maps are marked with blue plus signs , whose sizes represent their positional errors .
arrows indicate the displacement between the two epochs .
[ fig : movprop],title="fig:",scaledwidth=38.0% ] ) .
second epoch maps ( c - band ; right panels ) were obtained from the 2015 data .
different weightings have been used to better emphasize the emission of each source .
synthesized beams of @xmath62 and rms of @xmath0 10 @xmath15jy beam@xmath1 were obtained for the first epoch maps , while synthesized beams of @xmath63-@xmath64 , and rms of 3 - 8 @xmath15jy beam@xmath1 were obtained for the second epoch maps .
contour levels are @xmath274 , @xmath273 , 3 , 4 , 5 , 6 , 7 , 8 , 10 , 12 , 15 , 20 , 35 , 60 , 100 , and 150 times the rms noise of each map . positions of the sources derived from elliptical gaussian fits to the first epoch maps are marked with blue plus signs , whose sizes represent their positional errors .
arrows indicate the displacement between the two epochs .
[ fig : movprop],title="fig:",scaledwidth=38.0% ] ) .
second epoch maps ( c - band ; right panels ) were obtained from the 2015 data .
different weightings have been used to better emphasize the emission of each source .
synthesized beams of @xmath62 and rms of @xmath0 10 @xmath15jy beam@xmath1 were obtained for the first epoch maps , while synthesized beams of @xmath63-@xmath64 , and rms of 3 - 8 @xmath15jy beam@xmath1 were obtained for the second epoch maps .
contour levels are @xmath274 , @xmath273 , 3 , 4 , 5 , 6 , 7 , 8 , 10 , 12 , 15 , 20 , 35 , 60 , 100 , and 150 times the rms noise of each map .
positions of the sources derived from elliptical gaussian fits to the first epoch maps are marked with blue plus signs , whose sizes represent their positional errors .
arrows indicate the displacement between the two epochs .
[ fig : movprop],title="fig:",scaledwidth=38.0% ] ) .
second epoch maps ( c - band ; right panels ) were obtained from the 2015 data .
different weightings have been used to better emphasize the emission of each source .
synthesized beams of @xmath62 and rms of @xmath0 10 @xmath15jy beam@xmath1 were obtained for the first epoch maps , while synthesized beams of @xmath63-@xmath64 , and rms of 3 - 8 @xmath15jy beam@xmath1 were obtained for the second epoch maps .
contour levels are @xmath274 , @xmath273 , 3 , 4 , 5 , 6 , 7 , 8 , 10 , 12 , 15 , 20 , 35 , 60 , 100 , and 150 times the rms noise of each map . positions of the sources derived from elliptical gaussian fits to the first epoch maps are marked with blue plus signs , whose sizes represent their positional errors .
arrows indicate the displacement between the two epochs .
[ fig : movprop],title="fig:",scaledwidth=38.0% ] ) .
second epoch maps ( c - band ; right panels ) were obtained from the 2015 data .
different weightings have been used to better emphasize the emission of each source .
synthesized beams of @xmath62 and rms of @xmath0 10 @xmath15jy beam@xmath1 were obtained for the first epoch maps , while synthesized beams of @xmath63-@xmath64 , and rms of 3 - 8 @xmath15jy beam@xmath1 were obtained for the second epoch maps .
contour levels are @xmath274 , @xmath273 , 3 , 4 , 5 , 6 , 7 , 8 , 10 , 12 , 15 , 20 , 35 , 60 , 100 , and 150 times the rms noise of each map . positions of the sources derived from elliptical gaussian fits to the first epoch maps are marked with blue plus signs , whose sizes represent their positional errors .
arrows indicate the displacement between the two epochs .
[ fig : movprop],title="fig:",scaledwidth=38.0% ] ) .
second epoch maps ( c - band ; right panels ) were obtained from the 2015 data .
different weightings have been used to better emphasize the emission of each source .
synthesized beams of @xmath62 and rms of @xmath0 10 @xmath15jy beam@xmath1 were obtained for the first epoch maps , while synthesized beams of @xmath63-@xmath64 , and rms of 3 - 8 @xmath15jy beam@xmath1 were obtained for the second epoch maps .
contour levels are @xmath274 , @xmath273 , 3 , 4 , 5 , 6 , 7 , 8 , 10 , 12 , 15 , 20 , 35 , 60 , 100 , and 150 times the rms noise of each map . positions of the sources derived from elliptical gaussian fits to the first epoch maps are marked with blue plus signs , whose sizes represent their positional errors .
arrows indicate the displacement between the two epochs . [
fig : movprop],title="fig:",scaledwidth=38.0% ]
|
we carried out multiwavelength ( 0.7 - 5 cm ) , multiepoch ( 1994 - 2015 ) very large array ( vla ) observations toward the region enclosing the bright far - ir sources fir 3 ( hops 370 ) and fir 4 ( hops 108 ) in omc-2 .
we report the detection of 10 radio sources , seven of them identified as young stellar objects .
we image a well - collimated radio jet with a thermal free - free core ( vla 11 ) associated with the class i intermediate - mass protostar hops 370 .
the jet presents several knots ( vla 12n , 12c , 12s ) of non - thermal radio emission ( likely synchrotron from shock - accelerated relativistic electrons ) at distances of @xmath0 7,500 - 12,500 au from the protostar , in a region where other shock tracers have been previously identified .
these knots are moving away from the hops 370 protostar at @xmath0 100 km s@xmath1 . the class 0 protostar hops 108 , which itself is detected as an independent , kinematically decoupled radio source , falls in the path of these non - thermal radio knots .
these results favor the previously proposed scenario where the formation of hops 108 has been triggered by the impact of the hops 370 outflow with a dense clump .
however , hops 108 presents a large proper motion velocity of @xmath230 km s@xmath1 , similar to that of other runaway stars in orion , whose origin would be puzzling within this scenario .
alternatively , an apparent proper motion could result because of changes in the position of the centroid of the source due to blending with nearby extended emission , variations in the source shape , and/or opacity effects .
|
synchronization is an ubiquitous natural phenomenon that occurs widely in real systems including those in physics @xcite , chemistry @xcite , biology @xcite and nano - technology @xcite .
this phenomenon is an active topic of current research and is being extensively studied @xcite . nevertheless , synchronization is not always a desirable phenomenon . in some cases it is desirable
while in some other times it is undesirable and hence a mechanism to desynchronize becomes necessary for normal behavior .
for example , synchronization is desirable in the cases of lasers and josephson junction arrays @xcite , coupled spin torque nano - oscillators where coherent microwave power is needed @xcite , and in the brain when synchronization of neuronal oscillations facilitate cognition via temporal coding of information @xcite . on the other hand
, synchronization is undesirable when pedestrians walk on the millennium bridge @xcite and when mass synchronization of neuronal oscillators occurs at a particular frequency band resulting in pathologies like trauma , parkinson s tremor and so on @xcite .
apart from these , synchronization or desynchronization of neuronal oscillations in the brain are found to facilitate task selection and performance .
when such a synchronization or desynchronization occurs in the brain or other physical systems mentioned above due to a particular task or an event or an influence of an event then it is called an event - related synchronization ( ers ) or desynchronization ( erd ) . in this direction in order to understand the phenomena of erd and ers , we study the synchronization and desynchronization dynamics of a system of coupled nonlinear oscillators due to the effect of the strength of the external stimulus which represents some external influence . in particular , we demonstrate the occurrence of erd and ers due to the change in the strength of the external stimulus .
we also find the occurrence of erd / ers in periodic and chaotic systems and also with different forms of coupling .
we therefore find that the occurrence of erd and ers due to external influence is generic and is not limited to the type of the system or the coupling .
the plan of the article is as follows . in the following section we present a general model of coupled nonlinear oscillators in the presence of an external influence to explain erd / ers .
we discuss the case of the limit cycle stuart - landau ( sl ) oscillators in sec .
[ sl ] as an illustration of our model .
we quantify the strength of erd / ers by defining the intensity @xmath0 and identify the analytical stability boundaries of the erd and ers states .
further we find out the stability of the erd and ers states and also identify the analytical and numerical bifurcation boundaries between the different synchronization regimes .
we also demonstrate the occurrence of erd and ers in the presence of other forms of couplings like weighted coupling and time delayed coupling . in sec .
[ chaotic ] we show that erd / ers phenomena also occur in chaotic systems by illustrating the case of coupled rssler oscillators .
we point out potential applications of the model and the results in sec .
finally we present our conclusions in sec .
let us consider a system of coupled limit cycle nonlinear oscillators @xcite subject to _ external stimulus_. the external stimulus represents a dynamic signal from outside the system ( brain or or other physical systems such as spin torque nano oscillators ) or from a distant region inside the system and typically represents the influence of the event ( and not necessarily the actual event itself ) . with this in consideration
, the model equations shall be cast into the following form , @xmath1 where @xmath2 . here
@xmath3 represents the nonlinear limit cycle behavior of the @xmath4th uncoupled oscillator and @xmath5 is the corresponding system parameter .
@xmath6 is the coupling strength between the oscillators in the system and @xmath7 represents the coupling strength between the oscillators in the system and the external stimulus .
our findings on the occurrence of erd in this model has been briefly reported in @xcite . from a detailed analysis of generic models of the form ( [ sl01 ] )
, we have found that the external stimulus @xmath7 typically affects synchronization in the system of coupled oscillators .
if the system of coupled oscillators is completely synchronized , that is , @xmath8 , @xmath9 ( in this state all the oscillators behave as one ) due to the strength of the coupling @xmath6 ( when @xmath10 ) , then a sufficient strength of @xmath7 causes desynchronization ( erd , since external stimulus causes desynchronization ) in the system . typically erd occurs due to a small group of oscillators that separate themselves from the larger synchronized group .
the separated small group either remains desynchronized or synchronizes itself to a frequency different from that of the larger synchronized group . both of these cases will be demonstrated in the following sections .
the above behavior is generic and can also occur in chaotic systems leading to chaotic synchronization / desynchronization .
further , this behavior occurs irrespective of the form of coupling ( weighted , nonlocal , etc . ) and also in delay coupled systems . in all the cases , when the system is synchronized due to the strength of the coupling and when the strength of the external stimulus is sufficient , erd / ers occurs . in the following
we present our theoretical findings and discuss the qualitative connection between our findings and experimental observations .
a well - known and well - studied limit cycle nonlinear oscillator is the stuart - landau ( sl ) oscillator @xcite . a system of coupled sl oscillators
is described by the following set of coupled complex first order nonlinear ordinary differential equations ( odes ) , @xmath11 here @xmath12 is the nonisochronicity parameter , @xmath13 is the hopf bifurcation parameter and @xmath14 is the coupling function . @xmath15
is the complex amplitude of the @xmath4th oscillator with natural frequency @xmath16 .
@xmath17 is another sl oscillator with natural frequency @xmath18 and nonisochronicity parameter @xmath19 . without coupling @xmath20 ,
the dynamical equation of an individual stuart - landau oscillator becomes @xmath21 where @xmath22 .
the fixed points of the radial equation are @xmath23 and @xmath24 . from a linear stability analysis the eigenvalue is found as @xmath25 . for @xmath26
the system has a stable fixed point @xmath27 which is a linearly stable state corresponding to amplitude death . for @xmath28 ,
the fixed point @xmath27 becomes linearly unstable and a limit cycle oscillation is established with radius @xmath29 and phase @xmath30 , where @xmath31 is a constant . for
more detailed studies about synchronization in populations of sl oscillators one may refer to @xcite . , @xmath32 , @xmath33 , @xmath34 , @xmath35,@xmath36 and @xmath37 in equation ( [ sl02 ] ) .
left column shows the time evolution of the real part of the state vectors , @xmath38 , and the right column depicts the corresponding phase portraits on the ( @xmath38,@xmath39 ) plane . here
( top ) @xmath40 , ( middle ) @xmath41 and ( bottom ) @xmath42 .
note @xmath43.,width=226 ] the occurrence of erd and ers in a system of 1000 coupled stuart - landau oscillators is numerically demonstrated in figure [ sl1 ] for a lorentzian distribution of natural frequencies @xmath44^{-1}$ ] , where @xmath45 is the half width at half maximum and @xmath46 is the central frequency .
for the given set of system parameters ( @xmath47 , @xmath32 , @xmath48 , @xmath35 , @xmath36 , @xmath37 ) , erd occurs when moving from the top to the mid panel and ers occurs when moving from the mid to the bottom panel , upon increasing @xmath7 .
the time evolution of the real part of the state vectors , @xmath38 , and the corresponding phase portraits are plotted in the left and the right columns , respectively . when @xmath40 all the oscillators are synchronized in phase ( while there is slight desynchronization in the amplitude ) in the top panel . on increasing @xmath7 from @xmath49 to @xmath50 , in the mid - panel ,
erd occurs when some oscillators separate themselves from the synchronized group and are synchronized to a different frequency .
even though the separated group of oscillators are synchronized among themselves , we refer to this state as being desynchronized because from the point of view of the entire system , this is desynchronization .
further due to this separation , the intensity or the strength of synchronization ( discussed in detail in the following subsection ) of the original synchronized group ( shown in the top panel ) is reduced .
now , in this state , when we increase @xmath7 from @xmath50 to @xmath51 ers occurs ( with reference to the mid - panel ) when the separated group of oscillators again synchronize with the other oscillators .
the separated group of oscillators need not always be synchronized but can be either desynchronized or quasi - periodically synchronized .
the case of quasi - periodic synchronization is shown in figure [ sl2 ] . here
erd occurs in the mid - panel for the same values of ( system and control ) parameters as in figure [ sl1 ] , except that now @xmath52 , where the separated group is quasi - periodically synchronized .
; all the other parameter values are the same as in figure [ sl1].,width=226 ] the strength or intensity of synchronization can be measured by the number of oscillators that are oscillating in synchrony .
the more the number of oscillators that are oscillating in synchrony the more will be the strength of synchronization .
hence an increase in the strength or intensity of synchronization denotes the occurrence of ers and a decrease in the intensity denotes the occurrence of erd . with this reasoning
we quantify the intensity of synchronization using the phases of the oscillators by defining the quantity @xcite @xmath53dt,\end{aligned}\ ] ] where @xmath54 is the phase of the @xmath4th oscillator . here
the bar represents average over all oscillators in the population and the angle brackets represent time average . in the state of complete desynchronization @xmath55 and for complete synchronization @xmath56
( while we neglect very little desynchronization in the amplitude )
. for partial synchronization @xmath0 takes a value between 0 and 1 ; the more oscillators are oscillating in synchrony the higher will be the value of @xmath0 .
we use _ phase _ desynchronization / synchronization to characterize the occurrence of erd / ers because , upon increasing @xmath7 while the oscillators are in complete synchronization , very little amplitude desynchronization occurs first and then for an increase in @xmath7 desynchronization occurs in the phase and the desynchronization in the amplitude is also increased ( erd ) ( see for example figures [ sl1 ] and [ sl2 ] ) . further increase in @xmath7 brings back synchronization in the phase with slight desynchronization in the amplitude ( ers ) . therefore monitoring the intensity of phase synchronization will facilitate monitoring the occurrence of erd / ers .
figure [ slint1 ] shows that there exists a critical strength of the external stimulus for erd / ers to occur . upon varying the stimulus strength @xmath7 ,
erd occurs when there is a decrease in the intensity at a sufficient @xmath7 ( as denoted by a downward arrow in figure [ slint1 ] ) and when there is an increase in the intensity for sufficient @xmath7 , ers occurs ( as denoted by an upward arrow in figure [ slint1 ] ) .
it may also be noted that the erd / ers occurs for a finite window of the stimulus strength which depends upon the coupling strength @xmath6 . when we start with a state where the system is completely phase synchronized due to the coupling strength @xmath6 , the erd / ers occurrence window decreases with stronger @xmath6 .
this means that for very strong coupling @xmath6 there is a very less chance that erd / ers occurs .
for the sl system ( [ sl02 ] ) for varying stimulus strength @xmath7 for different values of the coupling strength @xmath6 .
all other parameters are the same as in figure [ sl2 ] . in the direction of increasing @xmath7 , downward arrow ( that is , a decrease in @xmath0 ) denotes erd and the upward arrow (
that is , an increase in @xmath0 ) denotes the occurrence of ers.,width=226 ] our numerical simulations show that when erd occurs for a sufficient strength of the stimulus , the synchronized group splits into one major synchronized group and a small separated group ( synchronized or desynchronized ) comprising of few oscillators ( say @xmath57 ) compared to the size of the major synchronized group ( say @xmath58 )
. therefore the size ratio of this two cluster state is @xmath59 where @xmath60 and @xmath61 , and in the desynchronized state @xmath62 ( @xmath63 ) .
for the case of sl oscillators ( [ sl02 ] ) with an external force @xmath64 ( so that @xmath65 and @xmath66 ) , the dynamics of the large group @xmath67 and the small group @xmath68 can be written as @xmath69 if we assume that the major synchronized group is completely synchronized with the external force @xmath64 , while the small group is not , we get @xmath70 and @xmath71 . from this reasoning , the dynamics of the small group can be written in the form @xmath72 where @xmath73 .
the occurrence of erd / ers can be understood from the fixed points of this equation @xcite .
due to the cubic nature of equation ( [ anal02 ] ) it has three fixed points which determine the stability of the erd / ers state .
one of the fixed points , namely @xmath74 , is stable .
the eigenvalues of the linearized version of ( [ anal02 ] ) are @xmath75 .
this corresponds to a complete synchronization ( synchronization of all the oscillators ) of the population , that is @xmath76 @xcite .
the other two fixed points ( say @xmath77 and @xmath78 ) determine the stability of the desynchronized state .
the fixed points @xmath77 and @xmath78 exist for @xmath79 where @xmath80 is a saddle - node bifurcation point and the fixed points are given by @xmath81 performing a linear stability analysis on equation ( [ anal02 ] ) we find that the determinant of the jacobian matrix for @xmath77 and @xmath78 is @xmath82 where @xmath83 .
it is easily checked that the determinant ( [ anal07 ] ) is always negative ( positive ) for @xmath77 ( @xmath78 ) .
therefore the fixed point @xmath77 turns out to be a saddle .
the other fixed point @xmath78 is either an unstable node or focus for @xmath84 .
when @xmath85 and @xmath86 , the fixed point @xmath78 is either a stable node or focus ; here @xmath87 is a hopf bifurcation point given by @xmath88 which is determined from the condition @xmath89 . here
phase diagram .
@xmath80 ( solid line ) and @xmath87 ( dashed line ) are the analytically obtained saddle - node and hopf bifurcation boundaries respectively .
the boundary @xmath91 ( dotted line ) is obtained numerically by solving the evolution equation for @xmath92 given in the text.,width=302 ] further for @xmath93 only synchronized solutions are stable and desynchronized solutions do not exist . for given values of parameters
the bifurcation boundaries are plotted in figure [ slanal1 ] . in the regions
i and iv all the oscillators are synchronized , while in the region ii there is frequency desynchronization ( mid - panel of figure [ sl2 ] is an example since the small group oscillates quasi - periodically while the major group oscillates periodically ) between the two clusters and in the region iii there is amplitude desynchronization between the two clusters . at the point @xmath91 ,
both the saddle and the hopf bifurcation points merge and disappear and this point is called a saddle - connection point .
it is obtained numerically by solving equation ( [ anal02 ] ) and is shown as a dotted line in figure [ slanal1 ] .
thus desynchronized solutions exist in the region between @xmath80 and @xmath91 .
this is in agreement with our numerical observations . a movement in the parameter space in the direction @xmath94 or @xmath95 represents erd and a movement in the opposite direction represents ers . of a system of 100 delay
coupled sl oscillators showing the occurrence of erd / ers in panels ( a ) - ( c ) .
here @xmath96 , @xmath97 , @xmath98 , @xmath99 , @xmath34 , @xmath35,@xmath100 and @xmath101 and @xmath102 for ( a ) , @xmath103 for ( b ) and @xmath104 for ( c ) .
occurrence of erd / ers in a system of 100 coupled sl oscillators with random weighted coupling with weight factor 3.5 is shown in panels ( d)-(f ) .
here @xmath98 , @xmath105 , @xmath106,@xmath36 and @xmath101 and @xmath102 for ( d ) , @xmath107 for ( e ) and @xmath108 for ( f).,width=283 ] in order to demonstrate further the validity of the model , we first consider a system of delay coupled sl oscillators described by @xmath109 where @xmath110 is the time delay parameter .
we simulate ( [ sl02de ] ) with @xmath96 and find the occurrence of erd / ers as depicted in figure [ sl_delay ] ( a)-(c ) for @xmath97 , @xmath98 , @xmath99 , @xmath35,@xmath100 and @xmath101 . in panel
( a ) the oscillators in the system are synchronized due to the strength of coupling @xmath6 when the external stimulus is absent . in panel
( b ) when the external stimulus is switched on with a strength @xmath103 erd occurs in the system where a few of the oscillators separate themselves from the major synchronized group .
when the strength of the external stimulus is further increased to @xmath104 in panel ( c ) ers occurs in the system and all the separated oscillators get back to synchronization with the major group .
secondly , erd / ers also occurs in a system with random weighted coupling described by @xmath111 where @xmath112 is the random coupling matrix .
figure [ sl_delay ] ( d)-(f ) demonstrates the occurrence of erd ( panel ( e ) ) and ers ( panel ( f ) ) in this system due to the stimulus strength .
thus we find that the form of the coupling is immaterial in causing the occurrence of erd / ers .
for the non - delayed case , if we reduce our model to a phase model the bifurcations match with those presented in @xcite where the authors have studied a system of coupled phase oscillators in the presence of external forcing .
the occurrence of erd / ers is generic and not limited to limit cycle oscillator systems alone . for example
let us consider a system of coupled rssler oscillators in the presence of an external field ( @xmath113 ) @xmath114 where the external stimulus is described by @xmath115 also this system is found to exhibit erd / ers . here
@xmath13 , @xmath116 , and @xmath12 are the system parameters that determine the periodicity / chaoticity of the system . a lorentzian distribution of the natural frequencies @xmath16 are chosen for numerical simulation .
the external stimulus is represented by @xmath117 and in this case we have introduced coupling with the @xmath118 component of the external stimulus .
however , the phenomenon demonstrated here can also be observed irrespective of the component in which the external stimulus is coupled to .
the system parameters of the coupled rssler oscillators are chosen so that they are in the periodic regimes while those of the external oscillator are chosen so that it is in the chaotic regime .
an illustration of erd and ers in this system of 1000 coupled oscillators is depicted in figure [ ross_erd1 ] for a lorentzian distribution of natural frequencies with @xmath119 , @xmath120 , @xmath121 , @xmath122 and @xmath123 , @xmath124 , @xmath125 and @xmath36 .
the top panel is the state of complete phase synchronization due to the coupling strength @xmath126 when @xmath107 .
erd occurs in the mid - panel for @xmath127 . in this state
some of the oscillators separate from the synchronized group and are desynchronized .
however unlike the case of erd in sl oscillators ( depicted in figure [ sl1 ] ) this erd state is not stable but breathes in time . that is the separated group is desynchronized at certain times while it is synchronized with the major group at other times
this is obvious from figure [ ross_zoom ] which is a zoom of the rectangular regions marked in figure [ ross_erd1 ] .
figure [ ross_zoom ] ( a ) corresponds to the dotted rectangular region , which is the state where some oscillators are separated and desynchronized from the major synchronized group . figure [ ross_zoom ] ( b ) corresponds to the solid rectangular region , which is the state where all the oscillators are synchronized .
we have confirmed that this is not a transient since we have plotted the data after allowing a sufficiently large number of iterations and the breathing phenomenon is found to occur periodically .
thus erd in this state is not stable but time varying . on further increase in the strength of the stimulus to @xmath128 , ers occurs in the system when all the oscillators are synchronized .
it may be noted that for illustrative purpose we have shown here the occurrence of erd / ers in a system of coupled rssler oscillators that are in periodic regime stimulated by a chaotic rssler oscillator .
however , the same can be observed even if the external oscillator is in the periodic regime and coupled oscillators are in the chaotic regime and also for cases where the coupled oscillators and the external oscillator are both in chaotic or periodic regimes . for illustration
we have shown the occurrence of erd in a system of coupled chaotic rssler oscillators with a chaotic stimulus in panels ( a ) and ( b ) and with a periodic stimulus in panels ( c ) and ( d ) of figure [ ross_new ] . , @xmath120 , @xmath121 , @xmath122 and @xmath123 , @xmath124 , @xmath125 , @xmath36 , @xmath129 and ( top ) @xmath107 , ( mid ) @xmath127 and ( bottom ) @xmath128 . left column shows the time evolution of the state vectors @xmath38 and the right column depicts the corresponding phase portraits on the ( @xmath38,@xmath39 ) plane . a blow up of the dotted and solid rectangular regions marked in the mid - panel
is shown in figure [ ross_zoom].,width=283 ] is shown in panels ( a ) and ( b ) , respectively . in panel
( a ) the separated group is desynchronized while in ( b ) it is synchronized with the major group.,width=245 ] , @xmath130 , @xmath131 . in panels ( c ) and
( d ) the external oscillator is periodic with @xmath123 , @xmath132 , @xmath133 .
erd occurs in panels ( b ) and ( d ) . here
@xmath126 , @xmath134 , @xmath135 , @xmath136 , @xmath36 , @xmath129 and @xmath40 for ( a ) , @xmath137 for ( b ) , @xmath138 for ( c ) and @xmath139 for ( d).,width=283 ] the same stimulus can result in ers and erd depending upon the strength of the stimulus . fig .
[ sl_ers_erd ] shows this phenomenon where we choose the initial state of the system to be in a partially synchronized state ( since we want to demonstrate both ers and erd ) .
panel ( a ) represents the initial state before the application of the stimulus .
when the stimulus is applied , depending upon the strength of the stimulus @xmath7 erd occurs in panel ( b ) while ers occurs in panel ( c ) .
the change in intensity @xmath0 for different values of stimulus strength @xmath7 is shown in panel ( d ) .
the reference line corresponds to the initial partial synchronized state and an intensity @xmath0 above and below the line denotes the occurrence of ers and erd , respectively .
thus we find that for a given external stimulus signal , both erd and ers can occur depending upon the strength of the stimulus . , @xmath32 , @xmath140=2.0 , @xmath52 , @xmath141 , @xmath142 and @xmath102 for ( a ) , @xmath143 for ( b ) and @xmath144 for ( c ) .
( a ) , ( b ) and ( c ) depict the phase portraits in the ( @xmath145 ) plane .
( d ) shows the change in the intensity @xmath0 for different values of @xmath7 , representing erd and ers.,width=283 ]
in this paper we have discussed the occurrence of event - related synchronization and desynchronization in a system of coupled nonlinear oscillators in the presence of an external stimulus .
this type of model and the results discussed here can be applied to various real world systems .
we point out a few applications below .
synchronization of neurons at a particular frequency band may facilitate the performance of a particular task while desynchronization at the same frequency band may leave it unattended .
the vice verse can also happen .
brain oscillations at different frequency bands are found to be one of the most crucial mechanisms that control higher level information processing , motor functioning and even large scale integration of information across various regions of the brain .
it is therefore highly important to understand the dynamics of event related neuronal oscillations in relation to specific tasks and/or pathologies . in particular , in pathological conditions like tremors , it is of great importance to understand the occurrence of synchronization or desynchronization of neuronal oscillations at specific frequency bands in order to be able to control the synchronization of neuronal oscillations and hence the pathological states .
the occurrence of synchronization or desynchronization due to an event is not limited to brain and there are many other systems that exhibit this phenomenon .
for instance , an open problem in the field of nano - technology is that the microwave power emitted by a single spin torque nano - oscillator ( stno ) is very small , of the order of nano - watts . in order to increase the power one needs to find ways to synchronize a group of coupled nano - oscillators to obtain coherent power .
the model discussed in this paper can be used to find ways to avoid desynchronization in such systems and can also explain the occurrence of synchronization and/or desynchronization caused by a stimulated microwave current @xcite .
we also note here that this model , apart from serving as a qualitative mathematical representative for erd / ers phenomena in the brain , can also explain synchronization and/or desynchronization due to external stimulus in other systems ; for instance , polariton condensates in semiconductor micro - cavities that interact both among themselves and with the reservoir and quantum coherence of condensates @xcite can also be represented by this model .
studies on these systems will be discussed separately .
even more generally , the model can account for synchronization / desynchronization due to external stimulus , while the system is already synchronized due to coupling .
possible applications in this direction include the establishment of synchronization or desynchronization due to external stimulus in neuronal networks , bose einstein condensates , josephson junction arrays and lasers and so on
. therefore it will also be of interest to investigate the occurrence of erd / ers in systems of diffusively coupled sub - populations .
such a model will facilitate the understanding of the simultaneous occurrence of erd / ers at multiple frequency bands in the brain and in systems of coupled stnos in multiple columns .
in this article we have proposed a general model of coupled nonlinear oscillators which can represent interesting physical or biological systems such as a system of coupled spin torque nano - oscillators or neurons , by generalizing and building upon our previous letter @xcite .
we subject the system to an external field , which represents an event or an external stimulus .
the model results provide an understanding of the occurrence of erd / ers due to different types of stimuli .
we have found that the occurrence of erd / ers is generic to all types of coupled oscillator systems including chaotic systems .
hence we hope that the model results can be applied to quantitative description and prediction of the occurrence of synchronization and desynchronization due to external stimulus in real world systems including bose
einstein condensates , josephson junction arrays and lasers and so on .
the work is supported by the department of science and technology ( dst)ramanna program , dst
irhpa research project and a dae raja ramanna program , government of india .
jhs is supported by the dst
fast track young scientist research project .
o. v. popovych , c. hauptmann , and p. a. tass , engineering in medicine and biology 27th annual conference shanghai , china , proceedings of the 2005 ieee , p. 7656
( 2005 ) ; phys .
lett . * 94 * , 164102 ( 2005 ) .
p. c. matthews and s. h. strogatz , phys .
* 65 * 1701 ( 1990 ) ; d. v. ramana reddy , a. sen , and g. l. johnston , phys .
. lett . * 80 * , 5109 ( 1998 ) ; j - n .
teramae and d. tanaka , phys .
lett . * 93 * , 204103 ( 2004 ) .
|
changes in the level of synchronization and desynchronization in coupled oscillator systems due to an external stimulus is called event related synchronization or desynchronization ( ers / erd ) .
such changes occur in real life systems where the collective activity of the entities of a coupled system is affected by some external influence . in order to understand the role played by the external influence in the occurrence of erd and ers , we study a system of coupled nonlinear oscillators in the presence of an external stimulus signal .
we find that the phenomena of ers and erd are generic and occur in all types of coupled oscillator systems .
we also find that the same external stimulus signal can cause ers and erd depending upon the strength of the signal .
we identify the stability of the ers and erd states and also find analytical and numerical boundaries between the different synchronization regimes involved in the occurrence of erd and ers .
|
the standard model has three free parameters in the bosonic sector : the electromagnetic coupling constant , @xmath5 , the mass of the z boson , @xmath6 , and the fermi coupling constant , @xmath2 .
the theory becomes predictive when these and several other fundamental parameters have been determined experimentally . by progressively improving the precision of these parameters ,
the theoretical predictions become increasingly precise and , in turn , the experimental measurements are increasingly sensitive to new physics beyond the standard model . therefore , on quite general grounds , it is important for each of the fundamental parameters of the standard model to be measured with the highest possible experimental precision .
the fermi coupling constant is determined from the measurement of the positive muon lifetime , @xmath7 , through the relationship @xmath8 in order to avoid uncertainties in the capture rate of negative muons on the target nuclei , the more precise value of @xmath2 is derived from the positive muon lifetime . in this equation
@xmath9 encapsulates the higher order qed and qcd corrections calculated in the fermi theory , in which the weak charged current is described by a contact interaction .
this is discussed in detail by van ritbergen and stuart @xcite , who have computed the second - order qed corrections and reduced the theoretical uncertainty of the radiative corrections to 0.3 ppm .
in contrast to the case for @xmath5 , @xmath2 is not afflicted with a hadronic uncertainty until 2-loops is reached , and consequently its influence is negligible . among the other parameters in eq.[eq_gf ] ,
the error on the muon mass , @xmath10 , contributes an uncertainty of 0.21 ppm to @xmath2 .
the contribution from the uncertainty in the muon neutrino mass is negligible , assuming the mass limits implied by neutrino oscillation experiments .
the largest uncertainty in @xmath2 comes from the error on @xmath7 , which is 18 ppm in the 2006 review of particle properties @xcite , and recently reduced by a new measurement from the mulan experiment at psi to a precision of 11 ppm @xcite . in this paper , an initial measurement of @xmath7
is presented from the fast experiment with a precision of 16 ppm .
the final goal of fast is an uncertainty of 2 ppm in @xmath7 , which will determine @xmath2 to 1 ppm precision .
the fast detector ( fig . [ fig_fast_detector ] ) operates in the @xmath11m1 secondary beamline of the 590 mev proton cyclotron at the paul scherrer institute , which generates a 2 ma dc proton beam ( 100% duty cycle ) .
the detector is a fast imaging target of @xmath12 pixels constructed from plastic scintillator bars of dimension @xmath13 mm@xmath14 .
the target is read out by wavelength shifter fibres attached to hamamatsu h6568 - 10 position sensitive photomultipliers ( pspms ) .
there are 96 pspms in total , each one viewing @xmath15 pixels .
a dc @xmath16 beam of momentum 165 mev / c is stopped in the target and the decay times of each @xmath17 decay chain are registered .
a wedge - shaped beam degrader ensures that the pion stopping positions are distributed uniformly through the target .
a local copper beam collimator defines a beam aperture of @xmath18 mm@xmath19 and suppresses pion stops near the edges of the target .
an array 16 finger counters ( z counters ) , each @xmath20 mm@xmath19 , measures the vertical position of the beam particle as it enters the target . in order to reach the final goal of a 1 ppm measurement of @xmath2 ,
fast must handle a factor 100 higher event rate than earlier muon lifetime experiments and , at the same time , reduce the systematic errors by at least an order of magnitude .
the high rate is achieved by means of a highly granular and fast detector that can measure several muon decays in parallel .
the imaging target ( fig.[fig_fast_detector ] ) essentially comprises a large number of replicated mini - detectors placed side - by - side , each one capable of making an independent and simultaneous measurement of a @xmath21 decay . with regards systematic uncertainties , the detector is designed to suppress several effects that limited previous experiments .
firstly , pile - up effects are reduced by a fast , imaging detector . secondly ,
possible gain shifts and additional event losses at the start of the muon decay period due to the higher counting rate are avoided by operating in a dc beam ( uniform counting rate ) .
thirdly , muon spin polarisation systematics are highly suppressed by a ) using a @xmath16 beam ( spin 0 ) and tagging the @xmath22 transition , b ) applying range cuts that further suppress residual beam muons , c ) designing a uniform detector with large angular acceptance , and d ) providing an 80 g magnetic field across the target , which causes rapid precession of the muons so that any residual spin effects can be observed and measured .
the pspm signals are passed through pre - amplifiers into custom - designed updating discriminators with dual - threshold differential ecl outputs .
the low level ( ll ) threshold is satisfied by minimum ionizing particles , and the ll pulses are sent to the time - to - digital converters ( tdcs ) .
the high level ( hl ) threshold is set to efficiently detect the large pulses from stopping pions or @xmath22 decays ( 4 mev muon kinetic energy ) , but to suppress minimum ionizing particles .
the hl pulses are sent to the level 2 ( lv2 ) trigger .
the time measurement is performed with 16 caen v767 128-channel multihit tdcs @xcite .
they provide a fast time stamp for each pspm pulse and then load the pixel address and time into memory .
dual hit registers on each input channel of the tdcs ensure a short minimum time separation of 10 ns for two hits to be registered on the same channel , although a longer deadtime of up to several 100 ns occurs before a third hit can be registered on the same channel since the contents of the hit registers must first be written into the local event buffer .
the tdcs are driven by an external 30 mhz rb atomic clock .
this defines a precise so - called _ coarse _ tdc tick which is then divided by the delay locked loop of the tdc chip into 32 _ fine _ tdc ticks , each of 1.041667 ns .
the time window of the tdcs is set to measure decays between -8 @xmath1s and + 22 @xmath1s relative to the pion stop time .
the level 1 ( lv1 ) trigger defines an incident beam particle by a coincidence of three beam counters ( bc1 , bc2 and z ) with the accelerator rf signal .
the leading edge of the lv1 trigger is determined by the machine rf pulse ( period @xmath2319 ns ) .
the lv1 deadtime is 50 ns , implying that no further triggers are accepted in the two subsequent rf buckets .
the lv2 trigger uses the hl data from the target to a ) reconstruct the pion stopping pixel and b ) identify the subsequent @xmath22 decay . having identified a prompt stopping beam pion , in - time with lv1 , the lv2 trigger is satisfied by a delayed second pulse within a 15100 ns window on any of the @xmath24 pixels centred on the pion stop pixel .
the lv2 trigger then provides a selective trigger pulse for the tdcs subtended by a @xmath25 pixel array centred on the pion stop pixel .
this serves to reduce the data bandwidth requirements of the data acquisition system ( daq ) since , on average , only 2.4 out of the 16 tdc modules are triggered per event .
the @xmath26 and @xmath27 coordinates of the muon stop pixel are also sent to tdc data inputs for use in the subsequent event analysis .
the experiment reads out only tdc data ; each 128 ch tdc has 96 ch devoted to target readout and 32 ch devoted to other data , including the pion stop coordinates .
the lv2 trigger operates at up to 1 mhz trigger rate and can process multiple lv1 triggers in parallel , provided they are separated by at least 50 ns .
the very high data rate of the fast detector makes it impractical to store all events for later analysis offline .
not only would the disk space requirements be huge about 7 tb / day but the reading and reprocessing time for these events would far exceed the original time taken for their collection .
the adopted solution is to carry out a full analysis of all data online , and to record approximately each hour a set of about 1200 separate muon lifetime histograms that contain all the information required for the lifetime measurement and for the study of systematic errors .
in addition , a pre - scaled fraction of the raw events of order 1% depending on the rate are recorded for monitoring purposes and to study specific aspects of the systematic errors .
a web - based online monitor program accumulates control histograms to verify the proper functioning of the detector and provides a warning if any distribution falls outside its pre - defined tolerance .
the monitor program includes a software emulation of the lv2 trigger ( albeit with ll data ) and continuously verifies the correct functioning of the lv2 hardware trigger .
the analysis procedure uses the muon coordinates given by lv2 and identifies any positron candidate emerging from the muon pixel in the -8 to 22 @xmath1s window .
positron tracks are identified using a set of 512 mask topologies in a so - called @xmath28 _ superpixel _ matrix centred on the muon pixel .
the mask topologies are chosen to be highly efficient for accepting true positrons from muon decays , while suppressing false positrons due to overlapping tracks from other beam particles or decays elsewhere in the target .
the absolute positron time is taken as the average of all the positron pixels within the superpixel .
the readout of the tdc s is done via pvic boards @xcite which provide an interface between the daq pcs and the 4 vme crates containing the tdc modules .
the data are transferred at the full vme bandwidth limit of 20 mb / s per crate into 4 daq pcs . a gigabit ethernet switch and collector pc
then build the events from the daq pcs into time slices and pass these slices in turn to a farm of event analyser pcs .
the event analyser pcs process the events and write the histograms and other information to disk .
the data were recorded in december 2006 at an average lv2 trigger rate of 30 khz .
the muon lifetime histogram for the full data sample of @xmath29 events is shown in fig.[fig_lifetime]a , measured with the resolution of a fine tdc tick ( 1.041667 ns ) . the positron time is measured relative to the time of the beam particle as defined by the lv1 pulse ( which is synchronised with the accelerator rf time ) .
the analysis allows one and only one positron candidate to be found in the -8 to 22 @xmath1s window ; events failing this criterion are rejected .
events are also rejected if the pion stop position falls outside an @xmath26 range of pixels defined by the associated z counter ( this suppresses a small contamination of beam muons since their range is 21 pixels further than pions ) .
after some additional geometrical and quality criteria are applied , 42% of the lv2-triggered events are accepted for inclusion in the lifetime histograms ( figs.[fig_lifetime ] ) .
the appearance of muon decays at negative times involves the loss of the true positron and the gain of a fake positron .
as seen in the insert of fig.[fig_lifetime]a , the fake positron background is structured with the accelerator rf period and dominated by incoming beam particles .
the events with negative decay times allow a precise description to be obtained for the accidental background .
the events with positive lifetime are used to perform the muon lifetime measurement .
figure [ fig_lifetime]b shows the same data after rebinning by exactly the accelerator rf period ( `` rf rebinning '' ) , as described below , which eliminates the beam structure .
a slight exponential behaviour can be seen in the background near the boundaries of the measurement window , as anticipated in the original study of systematic errors for the fast proposal @xcite .
the increase in background around -8 @xmath1s is due to positrons from another pion that stops in the signal muon pixel _ before _ the early edge of the window , whereas the increase in background around + 22 @xmath1s is due to another pion that stops in the signal muon pixel ( thereby faking a positron ) but whose subsequent positron emerges _
after _ the late edge of the window . in both cases
the background grows as the edge of the window is approached .
the measurement of the muon lifetime is performed in three steps .
first , the beam period is obtained with high precision from the negative decay region in fig.[fig_lifetime]a .
then , the lifetime histogram is rebinned using the measured beam period as the bin width ( fig.[fig_lifetime]b ) .
finally a fit to the rebinned histogram is performed with the muon lifetime as one of the free parameters .
the rebinning minimises the influence of the periodic background structure on the measurement of the lifetime
. a systematic error due to this fit method is assessed by comparing the lifetime value obtained from the data without rebinning ( [ sec_fine_bin_fit ] ) .
two different methods were used to obtain the beam period . in both cases , the decay interval from -7000 to -1000 tdc ticks was used , since this maximised the statistics while minimising the possible systematic uncertainties related to the exponential tail .
the first method consists of folding the full interval into a single period .
the dispersion of the points around the mean shape of the background is a minimum for the real beam period ( [ sec_fine_bin_fit ] ) .
the second method uses the fourier transform of the data to determine the frequency with the highest power , which corresponds to the beam frequency .
both methods agree and determine the beam period to be 18.960051 tdc ticks , with a precision of better than 1 ppm . since this does not correspond to an integer number of tdc ticks , the data are `` rf rebinned '' by dividing fractional tdc ticks into the two adjacent bins according to a linear sharing algorithm .
a further periodicity is introduced in the data by a non - linearity of the tdcs .
although the coarse ticks are perfectly paced by the 30 mhz rb atomic clock , the 32 fine ticks are determined by the tdc chip interpolation .
this was found to have a non - linearity with an amplitude of 1 per mil , introducing periodicities of 16 and 32 ticks in the fit residuals .
an identical tdc non - linearity was measured in both the beam data ( by folding the residuals with a period of 1 rb coarse clock tick ) and in laboratory measurements ( using random flat data ) .
the positive region of the lifetime distribution in fig.[fig_lifetime]b is fitted with the function @xmath30 where @xmath31 is the electron time relative to the rf bucket of the beam pion , @xmath32 accounts for the tdc non - linearity , and the parameters @xmath33 , @xmath34 and @xmath35 describe the amplitude of each component .
the first term represents the muon decay signal , the second term accounts for the small exponential rise of the background at the positive edge of the window , and the final term accounts for the flat , uncorrelated background .
the free parameters of the fit are @xmath33 , @xmath34 , @xmath35 and @xmath7 .
a binned maximum likelihood fit to the data is performed over the interval from 600 to 20000 tdc ticks .
the lower limit is chosen to avoid the `` third hit '' tdc inefficiency at earlier times , and the upper limit avoids undue sensitivity to the rising background near the edge of the time window .
the result for the lifetime is @xmath36 tdc ticks [ 2.197 083 ( 32 ) @xmath1s ] , with @xmath37 for 1020 degrees of freedom , corresponding to 40.1% probability .
the residuals of the fit show no systematic trends versus decay time ( fig.[fig_residuals]a ) and are gaussian distributed , with the expected mean and width ( fig.[fig_residuals]b ) .
the fitted signal fraction is 96.82% , while the positive exponential contribution is 0.02% and the flat background accounts for 3.15% .
several further tests of the quality of the fit have been performed .
the fourier transform of the residuals reveals no periodicities .
furthermore , the fit has been separated into four regions to check the stability and quality in samples that are statistically independent .
the four regions are 6003000 , 30006000 , 600010000 , and 1000020000 tdc ticks . within statistical errors ,
the fitted lifetime is the same in all four regions , as is the fitted constant background term .
as expected , the late decay region above 10000 ticks is most sensitive to the correct description of the background , while the early decays are essentially insensitive to background uncertainties .
overall , no evidence has been found to suggest any problems with the quality of the fit .
an alternative analysis has been carried out using the original lifetime histogram with a fine bin size of 1 tdc tick ( fig.[fig_lifetime]a ) .
the data are fitted in two steps .
first , a description of the shape of the background is obtained from the negative region .
second , this description plus the decay exponential and some other small corrections are used to fit the data in the positive lifetime region .
the systematic uncertainties of this analysis are somewhat larger than for the nominal fit with the rf rebinned data since a more precise description is required for the dc background structure and for the pulsed , positive exponential background at the end of the tdc window . nevertheless , this second analysis is important in order to check the consistency of the nominal fit and to assign a systematic error to the fit method .
a precise description of the negative region is obtained by folding all the data from -7000 tdc ticks to -1000 tdc ticks into one single rf period .
the folded data ( fig.[fig_spline]a ) clearly show the three components of the beam background : muons ( -15 tdc ticks ) , electrons ( -9 ticks ) and pions ( bump near -6 ticks ) .
the different timings between the particles are due to their different times - of - flight from the production target .
the structure is well - described using a spline with 150 knots .
the negative decay region is then fitted by the sum of an exponential , a flat background and the spline , repeated with the measured rf frequency .
the fit a region of which is shown in fig.[fig_spline]b has a @xmath38 for 5995 degrees of freedom .
once the shape of the background is fixed using the negative region , the positive lifetime region is fitted using the function : @xmath39 where @xmath31 is the electron time relative to the beam pion , the parameters @xmath33 , @xmath34 and @xmath35 describe the amplitude of each component , @xmath40 is the correction due to the tdc non - linearity and @xmath41 describes the shape of the increasing beam pion background near the positive edge of the window .
the free parameters of the fit are @xmath33 , @xmath34 , @xmath35 and @xmath7 .
a binned maximum likelihood fit is performed to the data between 600 and 16000 tdc ticks .
the fit region excludes longer - lived muons since they are most sensitive to uncertainties in the background description .
the muon lifetime obtained with these fine binned data is 5.2 ppm below the nominal value obtained with the rf rebinned data .
possible sources of systematic error have been identified and their influence on the lifetime has been evaluated in turn with the experimental data , as discussed below . [ [ fit - method ] ] 1 .
fit method : + + + + + + + + + + + + + + the stability of the measured lifetime is checked by varying the starting point and the end point of the fit over a large region .
the results are stable and no associated systematic error can be discerned .
the influence of the algorithm for rf rebinning has been checked using three different methods : simple sharing , linear interpolation and quadratic interpolation .
the observed difference between the measured lifetimes is below 0.1 ppm .
the systematic error on the tdc non - linearity correction has been estimated using two different corrections , with different amplitudes and phases .
no effect is observed on the fitted lifetime .
the sensitivity to the phase of the periodic beam background has been estimated by varying this parameter over one full beam period .
the associated systematic error is @xmath42 ppm .
furthermore the central value of the period has been varied by @xmath43 ppm , which is far outside the range of experimental uncertainty .
the influence in the measured lifetime is @xmath44 ppm .
four different methods have been used to perform the fit to test if there is any sensitivity to the analytical method : minimise the @xmath45 , maximum the likelihood , minimise the kullback - leibler discrepancy and minimise the absolute value of the residuals .
the various methods are sensitive to different features of the distribution .
since no differences are found , we conclude that the fitted lifetime is insensitive to the analytical method . finally ,
in order to evaluate possible biases of the rebinned method , the results have been compared with an independent measurement using the fine binned data .
the method used to obtain the lifetime from the fine binned histogram is described above .
the results are shown in fig.[fig_systematics]a for bin sizes between 1 and 34 tdc ticks . to account for the difference in the lifetime measurement between the rf rebinned method and the fine - binned ( 1 tdc tick ) method ,
we ascribe a systematic error @xmath46 ppm . summing all contributions ,
the total systematic error associated with the fit method is therefore @xmath47 ppm .
[ [ reference-time-pi-vs.mu ] ] 2 . reference time ( @xmath11 vs.@xmath1 ) : + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + the nominal analysis uses the beam pion as the reference start time ( @xmath48 ) , whereas the muon time could be used as an alternative .
the resultant lifetime distributions are sensitive to different systematic errors .
a pion start time makes the distribution independent of the muon time and also sharpens the beam structure of the background events ; a muon start time makes the analysis less sensitive to the shape of the beam background .
the difference of the fitted lifetime values for the two methods ( @xmath1 start minus nominal @xmath11 start ) is @xmath49 ppm and is taken as the associated systematic uncertainty .
[ [ detector - uniformity ] ] 3 .
detector uniformity : + + + + + + + + + + + + + + + + + + + + + + + the systematic error from detector non - uniformities is estimated as follows .
first the muon lifetime is determined from independent sub - samples that divide the data according to the particular parameter under study , such as , for example , location of the muon stop pixel in the target .
any sub - samples having lifetime measurements that are statistically incompatible ( @xmath503@xmath51 ) with the nominal lifetime measurement are then removed and a new lifetime determined from the remaining total sample .
the systematic error is then estimated as the signed full difference of the new lifetime minus the nominal lifetime .
this procedure has been used to evaluate non - uniformities with respect to the position of the pion stop pixel in the target as a function of the pspm ( 96 units ) , tdc module ( 16 ) , tdc chip ( 48 ) , @xmath26 coordinate ( 26 ) , and @xmath27 coordinate ( 42 ) .
no systematic effects are observed except as a function of pspm ( fig.[fig_systematics]b ) .
the measured lifetimes in two pspms deviate by more than @xmath52 from the nominal value , and the associated systematic error is @xmath53 ppm .
the position of the muon decay inside the pspm has also been checked by folding all the measured decays into one single pspm ( fig.[fig_systematics]c ) .
no systematic effect is observed . the total systematic error due to target non - uniformity
is therefore taken as @xmath53 ppm .
[ [ time - stability ] ] 4 .
time stability : + + + + + + + + + + + + + + + + + + the stability of the experiment with respect to calendar time has been checked by looking at the evolution of the measured lifetime during the 3 weeks over which the data taking was performed . no variation is observed .
the stability of the rb clock has been checked against the accelerator rf clock using the fourier transform of the negative decay time region of the lifetime plot in order to measure the accelerator rf period .
the two clocks are found to have a relative stability of better than 0.5 ppm over the duration of the data taking period .
[ [ beam - rate ] ] 5 . beam rate : + + + + + + + + + + + + + data were recorded at lv2 trigger rates between 30 and 42 khz . no systematic effect is observed with respect to the trigger rate .
[ [ tdc - performance ] ] 6 .
tdc performance : + + + + + + + + + + + + + + + + + + + occasionally during data taking , a tdc chip could lose synchronisation of the fine tick phase lock .
this was automatically identified when it happened by a characteristic broadening of the time distribution of the beam pion pixels .
the daq then ended the run , declared that run as bad , re - initialised the tdcs , and started a new run . to eliminate the possibility of any residual effect from events with broadened timing
, the bad runs have been collected together and fitted for the muon lifetime .
the value obtained is compatible with the nominal lifetime , within the relatively large statistical errors , @xmath54 ppm .
further evidence against a residual error from this source is that there was a significant difference in susceptibility to this problem among the 16 tdcs , yet there is no sign of any different lifetime measurement from the susceptible tdcs . in conclusion , there is no evidence for any systematic effect from this source . [ [ muon - spin - rotation - mu - sr ] ] 7 .
muon spin rotation ( @xmath55 ) : + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + muon spin rotation effects are highly suppressed in fast as a result of the basic experimental design .
nevertheless , spin effects can be extracted from the data by selection of special event topologies .
these make use of the fact that the muon spin is 100% polarised in the direction pointing back to the parent pion .
a @xmath55 signal can be obtained , for example , by selecting events where the @xmath11 , @xmath1 and positron are in the same pixel , or where the @xmath11 and @xmath1 are in the same pixel but not the electron . from these data ,
the @xmath55 period is measured to be 931 tdc ticks ( 970 ns ) , in agreement with that expected for @xmath2380 g magnetic field . in order to understand @xmath55 effects in the lifetime measurement , it is helpful to consider an idealised case where the pion is exactly centred on a pixel . in this case
the daughter muon will lie somewhere on a sphere centred on the pion and of radius about 1.4 mm ( corresponding to the range of a 4.12 mev muon ) , with its spin pointing radially inwards at the time of the @xmath22 decay . when projected onto the readout plane
, this effectively produces a circular distribution of muon stop positions located ( in this idealised case ) near the edge of the 4 mm pixel , with spins pointing radially inwards .
the net muon spin is zero and so , assuming uniform detection efficiency , no asymmetry should be observed in the positron emission directions .
the muons then begin to precess about the magnetic field , with the largest precession angle for the forward
backward muons ( i.e.along the beam direction and with spin direction perpendicular to the magnetic field ) and the the smallest for the left
right muons ( transverse to the beam direction and with spin direction along the magnetic field ) .
the first check of @xmath55 effects is therefore to see if there is any variation of the lifetime with respect to the relative position of the @xmath1 and the @xmath11 .
this is shown in fig.[fig_systematics]d ; no systematic effects are observed .
the second check of @xmath55 effects is to look at the time dependence of the residuals of the lifetime fit with respect to the direction of positron emission .
this is shown in fig.[fig_systematics_musr]a and reveals a clear @xmath55 signal in the forward
backward ( @xmath56 and @xmath57 ) and left
right ( @xmath58 and @xmath59 ) directions .
the @xmath55 amplitude is about 1 per mil and the phase at @xmath48 gives a peak deficiency for @xmath60 and @xmath57 , and a peak excess for @xmath58 and @xmath59 .
this is due to a small difference in the solid angle for positrons to be classified as @xmath60 between muons located near the upstream compared with downstream edges of the pion pixel , which arises from the relatively coarse pixel granularity .
for example , at @xmath61 , the downstream muons have a higher solid angle for classification as @xmath62 decays than do the upstream muons , which leads to a net deficiency . the mirror effect ( with the same phase )
occurs for positrons at @xmath57 .
the small @xmath61 deficiency at @xmath60 and @xmath57 appears as a symmetric excess at @xmath58 and @xmath59 , and cancels almost completely along the diagonal directions .
this pattern then oscillates with the @xmath55 frequency and gradually decays as the muons depolarise . in short , the small @xmath55 signal in the @xmath60 and @xmath57 directions due to muons emitted in the forward
backward direction is causing and therefore exactly compensated by an anti - phase @xmath55 signal in the @xmath58 and @xmath59 directions .
this is confirmed by the central distribution in fig.[fig_systematics_musr]a , which shows no @xmath55 signal when all positron directions are combined .
figure [ fig_systematics_musr]b shows the corresponding fourier transform for each of the positron directions . the @xmath55 signal is seen to be completely absent in the summed data .
we conclude that no systematic errors are present due to @xmath55 effects .
furthermore , the equality of the @xmath55 signal in the @xmath60 and @xmath57 directions confirms that there is a negligible residual component of beam muons , which would produce a forward
backward @xmath55 asymmetry .
.summary of the systematic errors in the muon lifetime measurement .
the total systematic error is obtained by first summing the same - sign errors in quadrature and then taking the average of the absolute values of the positive and negative errors .
[ cols= " > , < , > " , ]
a summary of the estimated contributions from the various sources of systematic uncertainty is shown in table [ tab_systematics ] , leading to a total systematic error of 6.5 ppm .
the measured value of the @xmath21 lifetime is therefore 2109.200 @xmath63 0.031 @xmath63 0.014 tdc ticks , i.e. @xmath64 where the first error is statistical and the second is systematic .
this measurement , with 16 ppm overall precision , is in good agreement with both the previous world average ( 18 ppm precision ) @xcite and the new result from mulan ( 11 ppm precision ) @xcite . from eq.[eq_gf ] , our measurement of the muon lifetime determines the fermi constant to be @xmath65 which may be compared to the world average value of @xmath66 gev@xmath4 in the 2006 review of particle properties @xcite .
the uncertainty of our present measurement of @xmath2 to 8 ppm precision is dominated by the statistical error .
the size of the data sample also limits the determination of the systematic errors .
it is expected that both will decrease with larger data samples , allowing the fast experiment to reach its final goal of determining the fermi constant @xmath2 to 1 ppm precision .
we would like to acknowledge and thank several colleagues who were important in the design studies and early preparations of the experiment : p.de jong , d.della volpe , p.kammel , r.nahnhauer , f.navarria , g.passaleva , a.perrotta , r.stuart and g.valenti .
we also thank a.dijksman , d.george and t.rauber for their important contributions to the fast hardware .
finally we acknowledge the generous support of psi , and the psi accelerator groups for their efficient operation of the cyclotron and beamline .
fast collaboration ( f.r.cavallo _ et al .
_ ) , precision measurement of the @xmath21 lifetime ( @xmath2 ) with the fast detector , psi proposal r-99 - 06.1 ( 1999 ) ; fast collaboration ( t.case _ et al .
_ ) , study of systematic errors in the measurement of the @xmath21 lifetime ( @xmath2 ) with the fast detector , addendum to psi proposal r-99 - 06.1 ( 1999 ) .
|
an initial measurement of the lifetime of the positive muon to a precision of 16 parts per million ( ppm ) has been performed with the fast detector at the paul scherrer institute @xcite .
the result is @xmath0 = 2.197 083 ( 32 ) ( 15 ) @xmath1s , where the first error is statistical and the second is systematic .
the muon lifetime determines the fermi constant , @xmath2 @xmath3 gev@xmath4 ( 8 ppm ) .
a. barczyk , j. kirkby , l. malgeri + _ cern , geneva , switzerland _ + j. berdugo , j. casaus , c. ma , j. marin , g. martinez , e. snchez , c. willmott + _ ciemat , madrid , spain _ + c. casella , m. pohl + _ universit de genve , geneva , switzerland _ + k. deiters , p. dick , c. petitjean + _ paul scherrer institute , villigen , switzerland _ + * fast collaboration * + _ submitted to physics letters b _
|
in a companion paper@xcite , we have presented a conformal mapping technique that allows us to map any 2d medium bounded by a rough boundary onto a half - plane .
this method is based on the iterative use of fft transforms and is extremely fast and efficient provided that the local slope of the interface remains lower than one .
when the maximum slope exceeds one this algorithm , similar in spirit to a direct iteration technique well suited to circular geometries@xcite , can no longer be used in its original form .
underrelaxation@xcite suffices however to make it convergent for boundaries having large slopes . beyond the determination of a conformal mapping for a given rough interface we have also shown in ref.@xcite how to generate directly mappings onto self - affine rough interfaces of chosen roughness exponent .
self - affine formalism is an anisotropic scaling invariance known to give a good description of real surfaces such as fracture surfaces@xcite .
this statistical property of fracture surfaces is of great interest in the study of friction or transport processes in geological faults@xcite .
building a conformal mapping is entirely equivalent to solving a harmonic problem with a uniform potential ( or field ) condition on the boundary .
we used extensively this property in ref.@xcite to study stationary heat flows in the vicinity of a rough boundary and we focussed on the case of self - affine surfaces where we were able to compute the exact correlation between local surface field and height profile .
we gave also a special emphasis to the problem of the location of the plane interface equivalent to the rough one at infinity .
it turned out that the conformal mapping technique provides a very direct means of computing the shift between the plane equivalent interface and the mean plane of the rough interface .
the range of applications of this first study naturally covers fields where laplace equation appears : electrostatics , concentration diffusion , antiplane elasticity ... in this second paper , we are specifically concerned with the case of bi - harmonic problems .
the method we propose leads to the solution of the bi - harmonic field through the inversion of a linear system , as most other alternative numerical approaches ( e.g. boundary elements method , spectral method ... ) but in contrast to the latters , the linear system to invert is naturally well conditionned and of rather modest size ( n equations for n fourier modes in the conformal mapping method ) in contrast with direct spectral methods ( 4n unknowns ) .
moreover , following the first step of our algorithm analytically allows to obtain systematic perturbation expansion results . after recalling our main results about conformal mapping in the first section
, we deal successively with two important examples of bi - harmonic problems : stokes flows and plane elasticity .
the second section is thus devoted to the study of a stationary stokes ( _ i.e. _ low reynolds number ) flow close to a rough boundary .
we shall also develop in this section the paradigm of the equivalent `` no - slip '' plane interface . in the third section we point out the problem of the stress distribution in a two - dimensional material bounded by a rough boundary and submitted to a uni - axial tension .
the study of such situations is of particular interest for computing the influence of the roughness on the rupture probability law of brittle materials .
we show that this problem is formally identical to that previously solved for the stokes flow .
we pay further a special attention to the case of self - affine surfaces and we present in this section numerical results that suggest that for large stresses , the surface stress statistical distribution law presents a power law behavior .
we recall here the essential results described in ref . @xcite .
the aim of this section is to conformally map a half - plane onto a two - dimensional domain bounded by a rough interface .
we first recall briefly how to build a conformal mapping well - suited to a given rough interface .
this problem can be written under form similar as that of a `` theodorsen problem '' @xcite ; in the semi - infinite geometry we deal with , it can be solved with an iterative algorithm using fast fourier transforms @xcite .
the second part of this section is focussed on the specific case of self - affine interfaces .
it turns out , indeed , that very simple constraints on a conformal mapping allow to generate directly a two - dimensional domain bounded by a self - affine interface of chosen roughness exponent .
this property is of particular interest in statistical studies .
we could thus establish in ref.@xcite the correlation between the norm of a harmonic field at a self - affine interface and the height of this latter . as illustrated in fig .
[ fi : method ] , we are seeking for a mapping from a half - plane onto a two - dimensional domain bounded by a rough interface . in the following we place our study in the framework of the complex analysis .
we consider then the lower half - plane @xmath0 whose complex coordinate @xmath1 is such that @xmath2 and the two - dimensional domain @xmath3 bounded by the rough interface @xmath4 , we call @xmath5 the complex coordinate in @xmath6
. we are seeking for a mapping @xmath7 from @xmath0 onto @xmath6 .
we restrict our study to mappings @xmath7 that are bijective holomorphic functions , _
i.e. _ @xmath7 only depends on the variable @xmath8 and not on its conjugate @xmath9 .
the transformations associated with such functions are said to be conformal in the sense that they preserve locally the angles .
let us now take advantage of the semi - infinite geometry we have to deal with .
the two domains we consider are very similar apart from the region close to the boundary . far from this one
, @xmath7 is essentially the identity and we can write : @xmath10 where the perturbation @xmath11 decreases with depth and takes non negligible values only in the close vicinity of the interface . in the following ,
we consider periodic interfaces in order to minimize edge - effects . without loss of generality ,
let us choose @xmath12 to be the lateral period .
a natural form of @xmath7 is then : @xmath13 where @xmath11 is expanded on a basis of evanescent modes .
we consider a single valued interface @xmath14 .
let @xmath15 be the real function giving the interface geometry , for all point @xmath5 of @xmath14 , @xmath16 the mapping function @xmath7 is such that @xmath17 , _
i.e _ @xmath18 the first equation is here very close to a fourier transform except that we have @xmath19 instead of just @xmath20 in the first term .
this formal proximity can be used to build an iterative algorithm . for sufficiently small roughness
, we can see from the second equation that @xmath21 is an approximation of @xmath22 at zeroth order .
a direct fourier transform of the profile @xmath20 allows then a first approximation @xmath23 of the coefficients @xmath24 .
the latter can be used to correct the previous approximation of @xmath25 , using ( [ para]b ) gives then a following estimation of the non - uniform sampling @xmath25 and of the coefficients @xmath24 via the fourier transform of @xmath26 $ ] . in appendix
a , we give a more formal presentation of this algorithm .
it turns out that this iterative technique converges provided that the maximum slope of the profile remains below unity .
the technique can be used for any rough single - valued interface . for profiles whose maximum slope exceeds one ,
the algorithm can be made convergent with slight modifications such that the use of under - relaxation techniques .
we refer the reader to ref .
@xcite for more details on the convergence and the stability analysis in this specific framework .
extensive studies of this technique in case of cicular geometry are available in ref .
@xcite .
as pointed out above , the algorithm we have just described is suited to any rough interface .
it is in particular possible to build conformal mappings associated with self - affine interfaces .
the latter are defined by their scaling invariance properties : an interface described by the equation @xmath27 is said to be self - affine if it remains ( statistically ) invariant under the transformations @xmath28 for all values of the real parameter @xmath29 .
the exponent @xmath30 is called the `` hurst '' or roughness exponent .
it is characteristic of the scaling invariance . from this property , we derive easily that @xmath31 where @xmath32 is a prefactor .
a simple fourier transform gives then the power density spectrum of the rough self - affine profile .
@xmath33 when using the algorithm previously described to map a self - affine interface , the first guess for the mapping function coefficients @xmath24 is thus such that : @xmath34 where the @xmath35 are the coefficients of the fourier transform of the profile and @xmath36 are @xmath37 independent random variables .
it turns out that this power law behaviour is not altered by the following steps of the algorithm . in a symmetric way
, we can impose , without any further restriction , the @xmath24 to follow a power law and have a look at the interface generated .
we choose thus @xmath38 where @xmath39 are random gaussian variable with 0 mean and unit variance for the real and imaginary part independently .
we have however to note that nothing prescribes _ a priori _ that the function obtained is bijective . from the parametrical expression of the interface @xmath4 ,
we can write @xmath40 = 1 + a. \mathop{\im m } \left [ \sum_k \epsilon_k k^{1/2 - \zeta } e^{-ikx } \right ] \;,\ ] ] and to guarantee that @xmath4 remains single - valued we have to choose amplitudes lower than the threshold value @xmath41 where @xmath42}}}\;.\ ] ] we checked numerically ( see ref .
@xcite that the power spectrum of such synthetic profiles was indeed a power law with the expected exponent .
the stokes equation describes fluid flows at low reynolds numbers . we address in this section the problem of a stationary stokes flow in the vicinity of a rough boundary .
the study of such flows can be of great technological interest in the case of convective transport processes @xcite : one can think of problems of surface deposition or erosion . in the same spirit
, the occurrence of recirculating eddies can render very difficult the decontamination of a polluted surface ; contaminants particles can be captured by diffusion in a cavity and remain trapped in it for an arbitrary long time . from a more fundamantal point of view , in experiments consisting of tracing particles passively advected in a flow , the same process can lead to non gaussian statistics of the arrival time of tracer particles .
we consider a semi - infinite 2d geometry with periodic lateral boundary conditions and a unit shear rate at infinity . far from being specific
, the results obtained in this context can easily be extended for any case of stationary shear flow in the framework of a double scale analysis . considering a shear flow of shear rate @xmath43 and an interface of typical boundary @xmath44
, where the the velocity and pressure fields @xmath45 and @xmath46 obey the usual navier stokes equation , we can , following richardson @xcite , define an inner problem where the reduced non dimensional variables obey , at first order in @xmath44 , a simple stokes equation in a semi - infinite geometry . in the following
we present a solution of this stokes equation that can be rewritten as a biharmonic equation for the stream function .
this solution only requires the knowledge of a conformal mapping @xmath47 from the lower half plane @xmath0 onto the actual space @xmath3 bounded by the rough interface @xmath4 and the inversion of a well conditioned linear system ; it can thus be applied to any single - valued rough interface .
we focus this brief study on the problem of the determination of the location of a plane boundary equivalent to the rough one at infinity .
this problem is equivalent to the one of the replacement of the no - slip condition on a rough interface by a back flow condition ( to be determined ) on the mean plane .
we compare our results with those of tuck and kouzoubov @xcite who developped in the _ actual space _ a method similar to ours in spirit .
recent results about stokes flows near rough boundaries can also be found in references @xcite ; in most of them the stokes equation is solved using boundary elements methods ( see for instance the review of pozrikidis @xcite ) .
we adress here the problem of a unit shear stokes flow in the vicinity of a rough boundary .
let us call @xmath48 the stream - function associated to the veolociy field @xmath49 in the actual space @xmath6 .
we have by definition @xmath50 in a stream function formalism , the stokes equation is reduced to a simple bi - harmonic equation .
taking into account the boundary conditions _
i.e. _ no slip on the interface and unit shear rate at infinity , @xmath48 has to be solution of the following problem : @xmath51 the essential difficulty obviously lies in the no slip condition on the interface ; the use of a conformal mapping allows us to build an equivalent problem with a much easier boundary condition , the new interface being plane instead of rough .
let us associate to the stream - function @xmath52 in the actual space @xmath3 the real potential @xmath53 in the half - plane @xmath0 : @xmath54 where @xmath7 maps @xmath0 onto @xmath3 .
we can thus define the following equivalent problem in the new geometry : @xmath55 = 0 & \ \\ { \bf \nabla}_z \phi ( z)=0 & { \rm on \ } \partial { \cal d } \\
\phi(z ) \sim \frac{\displaystyle 1}{\displaystyle 2 } y^2 & { \rm as \ } y \to - \infty \end{array } \right .
\end{array}\ ] ] in the case of a simple harmonic equation , building the conformal mapping gives immediatly the complete solution ; this is unfortunately no longer true in the case of a bi - harmonic equation .
one can see that the original equation is changed into a linear equation with non - constant coefficients .
the latter equation is directly related to the mapping function @xmath7 .
we show in the following that this difficulty can be circumverted .
let us recall that in addition to the above described conditions , the new potential @xmath53 has to be real and @xmath56-periodic in @xmath21 .
besides , the boundary condition can be made simpler taking into account that the interface is now plane .
@xmath53 being defined apart from an additive constant , we can write that it obeys : @xmath57 the bi - harmonic potential @xmath53 can always be written in terms of two holomorphic functions @xmath58 and @xmath59 such that : @xmath60 in the following we split both functions into a purely periodic part ( denoted by the index @xmath46 ) and a non - periodic part . taking into account the desired behaviors at infinity
provides : @xmath61 with @xmath62 and @xmath63 is @xmath8 times a @xmath21-periodic function .
the lateral periodicity of @xmath53 forbids the occurrence of terms proportional to polynoms of the real variable @xmath21 , hence : @xmath64 and @xmath53 becomes then : @xmath65 we can deduce from the latter expression the partial derivative of @xmath53 with @xmath66 : @xmath67 the holomorphic functions @xmath68 , @xmath69 and @xmath70 being @xmath12-periodical , they can be developed using the basis of the functions @xmath71 . besides , the interface @xmath72 being simply the @xmath21-axis , the restriction of these holomorphic functions to the boundary can be written in @xmath73 .
the boundary condition problem can then be written by cancelling out the projections of @xmath53 and @xmath74 on the functions - vectors @xmath73 . keeping only the first @xmath75 modes
, we obtain @xmath76 equations which allow to obtain the @xmath76 components @xmath77 and @xmath78 .
the projection of the boundary condition on the function - vector @xmath79 ( @xmath80 ) gives first : @xmath81 we have then for the function - vectors @xmath73 with @xmath82 : @xmath83 we can thus write the following linear system : @xmath84 the coefficients @xmath85 are solutions of the first equation , a @xmath86 linear system
. the coefficients @xmath87 are easily deduced from the @xmath85 .
once the conformal mapping is known , the numerical resolution of the stokes equation just requires the inversion of the linear system .
let us note that the latter system is easy to invert , which would not have been the case if we have written the boundary condition in the @xmath88-space .
the latter method was recently used by tuck et kouzoubov @xcite . in the cited reference ,
they use expansions in a basis of @xmath89 and write a linear system discretizing the rough interface in @xmath75 points .
if this method is efficient in the small slope limit , it becomes however unpracticable for large slopes .
the procedure requires thus the numerical inversion of a matrix consisting of terms of order @xmath90 where @xmath91 is the roughness amplitude and this becomes practically difficult or unprecise as soon as the product @xmath92 increases .
the conformal mapping avoids such numerical difficulties since the boundary is plane in the equivalent domain .
one can see on figures [ stokes1 ] and [ stokes2 ] maps of the stream function for a stationary stokes flow for two rough surfaces identical up to a dilation of factor @xmath93 in the vertical direction .
the stream lines closely follow the smooth interface ( fig .
[ stokes1 ] ) while an eddy appears in the largest depression of the roughest interface ( fig .
[ stokes2 ] ) .
we now introduce the notion of an equivalent plane no - slip boundary in the framework of a stationary stokes flow .
our aim is here to replace the no - slip condition on the rough boundary by a no slip condition on an equivalent plane boundary , the stream function remaining unchanged at infinity .
let us recall that like in the case of a rough electrode which was studied in ref .
@xcite , nothing prescribes the plane equivalent boundary to lie at the average height of the rough one . in harmonic problems ,
the dissymetry between peaks and cavities due to point effects leads the equivalent plane boundary to be shifted towards the peaks .
we shall see in the following that the same conclusion holds in the case of the stationary stokes flow .
an illustration of this disymmetry naturally emerges with the occurrence of little eddies in pronounced depressions of the rough boundary .
we have chosen in this text to develop the paradigm of no - slip plane equivalent boundary but it is naturally also possible to consider a plane boundary fixed at the average plane with a roughness dependent slip boundary .
the conclusion previously mentioned about the place of the equivalent boundary ( nearer from the peaks than from the cavities ) is then expressed by a reversal flow condition .
we show in the following that the conformal mapping method gives a natural way to compute the vertical shift of the plane equivalent boundary .
we compare these results with a perturbative solution that can be directly computed from the fourier coefficients of the interface .
as already mentioned , the stream function @xmath94 of the stokes flow is entirely determined by the following conditions : @xmath95 if we now return to the solution obtained via conformal mapping , we have : @xmath96 by construction , @xmath97 is such that @xmath48 is bi - harmonic in @xmath3 and fulfills the no - slip boundary condition at the rough interface .
let us now build the stream function @xmath98 associated with a plane interface located at @xmath99 ; we have immediatly : @xmath100 and its associated function in the half - plane @xmath0 is @xmath101 ^ 2 \;,\ ] ] that becomes at infinity : @xmath102+{\cal o}(1 ) } & { \rm as \ } y \to -\infty \;. \end{array}\ ] ] taking into account eq.([f0 ] ) which specifies the expression of @xmath103 , the identity between @xmath53 and @xmath104 defines the value of @xmath59 @xmath105 let us consider an interface of amplitude , say @xmath91 , with characteristic length @xmath29 such that the profile is statistically symmetrical .
when we seek for the location of the plane equivalent electrode , we expect two different behaviors i ) a dependence on @xmath106 in case of small amplitude or low spatial frequency ii ) a linear dependence on @xmath91 in case of large amplitude or high spatial frequency .
the latter behavior comes directly from the fact that the equivalent plane reaches at maximum the level of the highest peaks . in case of small slopes ,
it is easy to show that the correction from the average plane is of order @xmath106 .
the deviation @xmath59 is naturally normalized by the amplitude of roughness @xmath91 .
the ratio @xmath107 has then to be a function of the two only characteristic lengths of the system , @xmath91 and @xmath29 and can be expanded in the limit of small slopes : @xmath108 a simple symmetry about the mean plane has to leave @xmath59 unchanged , since this symmetry is equivalent to a transformation of @xmath91 into @xmath109 , @xmath110 and @xmath111 have to be zero and : @xmath112 a detailed perturbative analysis can be built in the case of a simple sine interface @xcite . writing a perturbative solution in the conformal mapping formalism allows us to deal with any rough interface . following ( [ phi(z ) ] )
we have : @xmath113 by construction , @xmath11 , @xmath69 and @xmath70 are of order @xmath91 , @xmath91 being the roughness amplitude . at first order
the no - slip boundary condition becomes : @xmath114 the holomorphic functions being bounded , the resolution of these hilbert problems gives immediatly : @xmath115 using the iterative algorithm briefly presented in section ii ( see appendix for details ) , a first order approximation of the coefficients @xmath24 is @xmath116 where @xmath117 is the @xmath76 discrete fourier transform of the real function @xmath15 associated with the rough interface . using the expression of @xmath118 derived in section v - b of ref .
@xcite , we can write : @xmath119 this result is consistent with the one proposed in ref .
@xcite for a backflow slip condition on the mean plane of the interface . in the particular case of a pure sine profile of amplitude @xmath91 and wavelength @xmath29
, we recover : @xmath120 it has to be noted these first order results are exactly identical to those obtained in the case of a rough electrode ( up to a factor 2 ) despite the fact we had here to solve a biharmonic equation instead of a simple harmonic one . on figures [ stokesequivsin ] and [ stokesequiv ] we have plotted both results of the perturbative solution and of the conformal mapping calculation in case of a sine interface and of a self affine interface of roughness exponent 0.8 built with 64 fourier modes .
we check that the perturbative calculations correctly fit the results for small slopes ( up to 0.5 ) . for larger slopes ,
the perturbative expression overestimates the deviation whose behavior becomes progressively linear .
note that our numerical method allowed us to reach local slope values up to 2.5 .
this maximum slope can be increased by using more fourier modes in the mapping function ( we used 256 modes in the present calculation ) .
in this section , we analyse the effect of a slight surface roughness on the stress distribution in an elastic medium .
we emphasize here the case of a semi - infinite material submitted to a uni - axial tension .
although very elementary , this simple model illustrates an effect which has been suggested to be responsible for the mechanical strength of glass fibers .
recent experimental results @xcite suggest that the nanometric roughness at the surface of glass fibers of diameter a few micrometers could be responsible for the decrease of the tensile resistance by a factor of about 5 .
in uniaxial tension , the resistance of a fiber is directly related to the distribution of maximum positive principal stresses . using a simple perturbative expansion ,
we show that in the limit of small slopes , the surface stress can be directly computed from the hilbert transform of the local slope .
the weibull law @xcite usually gives a correct description of failure statistics in a wide range of brittle materials .
this phenomenological law is partly based on the identification of the weakest link in the material ( whose size is supposed to be that of the smallest defects ) .
we will see in the following that for a self - affine surface ( such as measured for glass fibers by afm techniques @xcite ) the statistical distribution of tensile stresses at the boundary displays a power law behavior , which naturally implies the validity of weibull statistical failure distribution .
moreover the exponent of this power law , _
i.e. _ the weibull modulus , continuously depends on the roughness amplitude . in plane stress or plane strain conditions , the stress tensor @xmath121 $ ]
can be completely represented by a unique real function , named the airy function : @xmath122 = \left[\matrix{\sigma_{uu } & \sigma_{uv } \cr \sigma_{uv } & \sigma_{vv } } \right ] \quad { \rm where } \quad \left\ { \begin{array}{lc } \vspace{2pt } \sigma_{uu}=&\frac{{\displaystyle}\partial^2 \psi}{{\displaystyle}\partial v^2}\\\vspace{2pt } \sigma_{uv}=&-\frac{{\displaystyle}\partial^2 \psi}{{\displaystyle}\partial u\partial v}\\ \sigma_{vv}=&\frac{{\displaystyle}\partial^2 \psi}{{\displaystyle}\partial u^2 } \end{array } \right.\ ] ] this form directly comes from the stress balance in absence of external forces _
i.e. _ @xmath123=0 $ ] . in the framework of 2d elasticity in an isotropic medium , the airy function obeys : @xmath124 the stress tensor being computed from two successive derivations of the bi - harmonic function @xmath52 , the latter is only defined apart from a linear function in @xmath22 and @xmath125 . in the following , we consider free boundary conditions @xmath121{\bf n}={\bf 0}$ ] and we impose a uni - axial tension at infinity .
let @xmath126 denote a unit vector normal to the interface , we have : @xmath127 { \bf n } & = & { \bf 0 } & { \rm on \ } \partial { \cal e}\;,\\ \left [ \sigma \right ] & \rightarrow & \left [ \begin{array}{cc } 1&0\\ 0&0 \end{array } \right ] & { \rm as \ } y \rightarrow -\infty \;. \end{array}\right.\ ] ] the airy function @xmath52 is thus such that : @xmath128 at any point of the interface @xmath4 , it is possible to give a parametric representation of the tangential and normal vectors @xmath129 and @xmath126 : @xmath130 the boundary conditions at the interface can be rewritten : @xmath131 { \bf n } = \enskip 0 \quad & { \rm on}\enskip \partial { \cal e}\\ \vspace{4pt } \leftrightarrow & \left({\bf t } \cdot \nabla_w \right)\nabla_w \psi = \enskip 0 \quad & { \rm on}\enskip \partial { \cal e } \\
\vspace{4pt } \leftrightarrow & \nabla_w \psi = cst .
\enskip=\enskip 0 & { \rm on}\enskip \partial { \cal e}\;. \end{array}\ ] ] we can choose the constant to be zero since the airy function is only defined apart from an affine function .
this leads to : @xmath132 it turns out that the boundary condition at the interface is exactly the same as the one we have encountered for the no - slip condition in a stokes flow .
the airy function is thus identical to the above derived stream function for stokes flow .
the normal stress being zero at the interface , the first order expression derived in the previous section gives us the following result for the principal ( tangential ) stress at the interface : @xmath133}\\ \vspace{6pt } & { { \displaystyle}= 1 -2 \sum_{k =- n+1}^{k = n}i \frac{k}{|k| } ( -ik ) \frac{\widetilde{h_k}}{2n } e^{-ikx } } \\ & { { \displaystyle}= 1 -2 { \cal h}[\varphi'](x ) } \ ; , \end{array}\ ] ] where @xmath134 $ ] stands for the hilbert transform oprerator on the real axis . at first order
, we find that the tangential surface stress deviation from its mean value is proportional to _ the hilbert transform of the local slope_. fig .
[ cont01 ] and [ cont04 ] give two examples of stress distribution on surfaces of repectively maximum slope 0.1 and 0.4 .
in both cases , we can see that the stress fluctuations are much greater than the height fluctuations ( which have been dilated by a factor 5 in the figures ) .
note that ( taking into account the dilation of the height profile ) the stress fluctuations are very large compared with those of the height . in the context of rupture , the relevant parameter being the maximum stress , one can understand that a very modest roughness can be responsible for a dramatic decrease of the material resistance .
we observe a good agreeement between the stress profile computed by conformal mapping and the perturbative result ( related to the hilbert transform of the slope ) for the smoothest interface but this is no longer the case for the roughest one especially in the area where the curvature is large .
higher order perturbative terms are thus necessary to recover the actual stress .
let us now turn to the study of stress distributions on self - affine surfaces .
the latter are designed to present a gaussian height distribution _
i.e. _ their fourier coefficients are @xmath135 where @xmath39 is a gaussian random variable of zero mean and unit standard deviation . in case of small slopes ,
the validity of the first order perturbative result suggests thus that the stress distribution is gaussian . on fig .
[ lognormal ] we have plotted in log - log scale the surface stress distribution obtained for 1500 self - affine surfaces of roughness exponent @xmath136 and of maximum slope 0.05 , one can check that , as expected , the distribution is well fitted by a parabola . in case of larger slopes
we have seen that the agreement between the first order perturbative results and the complete computation becomes poorer .
we can then wonder if the log - normal behavior of the distribution is preserved . on fig .
[ powerlaw ] , we have plotted the surface stress distribution for four self - affine surfaces of roughness exponent @xmath136 and of respective maximum slopes 0.2 , 0.4 , 0.6 and 0.8 . these results were obtained by averaging the data obtained from 1000 different surfaces each defined with 64 fourier modes .
we can see a clear power law like behaviour for large stress amplitudes .
the slopes we can measure are very dependent on the roughness amplitude .
the interpretation of these new numerical results requires a perturbative analysis that we have not developed yet for this bi - harmonic problem .
we have however performed a similar analysis in the case of a harmonic field on self - affine interfaces @xcite .
it turned out that in a similar fashion as the one presented above , the field distribution law present a power law tail with an exponent @xmath137 where @xmath91 is the roughness amplitude , @xmath138 the spatial lower cut - off of the self - affine domain and @xmath30 the roughness exponent . calling @xmath139 the logarithm of the field ,
the latter result was derived showing that the reduced variable @xmath140 follows a gaussian distribution ( @xmath141 for harmonic problems ) .
using @xmath142 , we can check indeed ( see fig .
[ collapse ] ) that all data obtained from our calculations collapse on a same parabola in log - log scale .
these numerical results indicate that a same scaling applies for both harmonic and bi - harmonic fields on self - affine surfaces . a detailed analysis of these statistical properties will be addressed elsewhere .
after introducing a conformal mapping technique that allowed for a detailed study of harmonic field in the vicinity of rough boundaries @xcite , we have extended in this paper the use of this method to the study of bi - harmonic fields .
we have given a general solution to problems such that the stokes flow over a rough surface and the stress distribution in a medium ( bounded by a rough interface ) in uni - axial tension .
besides the knowledge of the mapping function ( obtained using a simple iterative algorithm ) , this solution only requires the linear inversion of a well conditioned matrix .
the determination of the mapping function being only limited by the maximum value of the local slope at the interface , the method is well suited to any kind of single - valued interface . as an illustration
, we have thus presented results of stokes flow over self - affine boundaries whose maximum slope reach 2.5 . in the same context of a stokes flow over a rough boundary , we pay a special attention to the determination of the location of an equivalent no - slip plane interface .
the conformal mapping method gives way to a very natural determination of this quantity .
a simple perturbative solution allowed us to retrieve for it a result proposed by tuck and kouzoubov @xcite . in the context of the plane elasticity , the same perturbative result has allowed us to show that , in the limit of small slopes , the surface stress distribution was directly related to the hilbert transform of the slope of the interface .
this very simple result could be used e.g. for the evolution of a stress - corrosion front .
the analysis of statistical results for the principal stress on self - affine surfaces has shown moreover that the large stress distribution presents a power law tail whose exponent continuously depends on the roughness amplitude .
such results could be applied in the context of glass fiber rupture statistics to provide a fundamental basis for the weibull law that is known to describe accurately the rupture statistics .
a realistic description of these stress distributions require a second order perturbative analysis which will be presented in a futher study .
we have , however , recently proposed such an approach in case of harmonic fields @xcite where distributions of the same kind have also been found and justified in two and three dimensions .
it is a pleasure to acknowledge enlightning discussions with f. creuzet , e.j .
hinch and j.r .
we are also grateful to b. forget and a. tanguy for useful help and comments .
in this appendix we present the iterative scheme allowing us to build the conformal mapping @xmath143 for any given single valued rough interface . as exposed in ii.b , we are seeeking for a mapping function @xmath7 such that @xmath17 , where @xmath144 is the rough interface and @xmath145 is the x - axis . the single valued interface @xmath14 being associated to the real function @xmath15
, this condition is rewritten : @xmath146 we build the algorithm in the following way : the intermediate quantities appearing at the @xmath37th iteration are labelled with a superscript @xmath147 , all functions are decomposed over a set of @xmath148 discrete values , the number of fourier modes is thus limited to @xmath148 .
we first introduce a series of sampling points @xmath149 with @xmath150 which is initially set to an arithmetic series @xmath151 .
the sampling of @xmath19 by the @xmath149 gives the array : @xmath152 the discrete fourier transform of this array is the complex valued array @xmath153 for @xmath154 .
the latter is shortly written as : @xmath155 \;,\ ] ] where @xmath156 denotes the fourier transform , which will be chosen as the fast fourier transform ( fft ) algorithm , thus imposing that @xmath157 is an integer power of 2 .
the intermediate mapping @xmath158 is computed from the @xmath159 as : @xmath160 the latter form is obtained from the identification of eq.([eqboom]b ) and the definition of @xmath159 , taking care of the fact that one sum is over positive index , while the other extends over the interval @xmath161 $ ] . then , one computes the series : @xmath162 this linear transformation is shortly noted as : @xmath163 \;,\ ] ] where @xmath164 is the above detailed transformation .
the form of @xmath165 is dictated by eq.([eqboom]a ) for positive index , and from the fact that the inverse fourier transform of @xmath166 ( see below ) is real .
the new sampling series is finally obtained from : @xmath167\;. \label{eqalgend}\ ] ] the equations ( [ eqalgbeg]-[eqalgend ] ) define one step in the algorithm relating @xmath168 to @xmath158 .
we note this step @xmath169 .
the searched function @xmath7 is clearly a fixed point of the transformation @xmath170 defined above in a discretized version .
the uniqueness of the transformation @xmath7 results from that of the harmonic field in the domain @xmath3 with an equipotential condition on the boundary and a constant gradient perpendicular to the boundary at infinite distance from it . therefore
, the only condition to consider is the stability of the fixed point . as it can be seen in ref .
@xcite , this fixed point is attractive at the only condition that : @xmath171 the procedure described above gives a very fast and efficient way of building conformal mappings .
it has to be noted indeed that this two - dimensional problem is solved here by the only use of a few one - dimensional fast fourier transforms . in cases of slopes locally overcoming one , it is possible to use under - relaxation schemes ( see details in ref .
@xcite ) or to decompose the mapping in several substeps such that the convergence criterium is always fulfilled .
|
we use a conformal mapping method introduced in a companion paper @xcite to study the properties of bi - harmonic fields in the vicinity of rough boundaries .
we focus our analysis on two different situations where such bi - harmonic problems are encountered : a stokes flow near a rough wall and the stress distribution on the rough interface of a material in uni - axial tension .
we perform a complete numerical solution of these two - dimensional problems for any univalued rough surfaces .
we present results for sinusoidal and self - affine surface whose slope can locally reach 2.5 . beyond the numerical solution we present perturbative solutions of these problems .
we show in particular that at first order in roughness amplitude , the surface stress of a material in uni - axial tension can be directly obtained from the hilbert transform of the local slope . in case of self - affine surfaces ,
we show that the stress distribution presents , for large stresses , a power law tail whose exponent continuously depends on the roughness amplitude .
|
load disaggregation or non - intrusive load monitoring ( nilm ) is the process of finding out how much each appliance within a household is consuming when only the aggregate current or power reading is available @xcite .
such readings are now available through smart meters , which have been , or are being , installed by most power utilities .
in addition to determining appliance consumption patterns , nilm could help balance different loads within a power network @xcite by predicting demand without the use of additional sensors .
recent disaggregation methods make use of machine learning approaches such as clustering @xcite , fuzzy systems @xcite , and hidden markov models @xcite .
such methods might lead to practical solutions when large and sufficiently representative datasets become available for training , which is still not the case .
the ultimate goal for nilm is to enable disaggregation without the need for supervised learning .
there has been some recent progress in this area @xcite , although the accuracy of such methods is still low .
alternative approaches such as combinatorial optimization or integer programming ( ip ) have been much less explored .
two notable earlier works on ip - based disaggregation include egarter _ et al . _
@xcite and suzuki _ et al .
_ @xcite .
egarter _ et al . _
formulated disaggregation as a modified knapsack problem and proposed a solution using an evolutionary algorithm . from a practical standpoint ,
the drawback here is that evolutionary algorithms potentially have a long execution time , and their stochastic nature may lead to different solutions in different runs , even on the same data .
our goal here is to develop a simpler and more principled approach that gives repeatable results .
the other ip - based disaggregation proposal was by suzuki _ et al . _
@xcite in 2008 using high - frequency sampled current readings .
we extract the ip portion of @xcite ( without the load signature part ) and enhance it in multiple ways to improve disaggregation accuracy .
the enhancements include additional constraints , correction based on a state diagram , median filtering , and linear programming - based refinement.our method works on low - frequency data , which is a more realistic solution for current smart meters that usually report power readings at intervals of 815 seconds .
the remainder of the paper is organized as follows .
we mathematically define the problem of load disaggregation as a mixed - integer linear ip problem in section [ sec : sec2 ] .
we propose several enhancements to ip - based disaggregation in section [ sec : sec4 ] , which is our main contribution .
a number of experimental results with a comparison to previous works are reported in section [ sec : sec5 ] , followed by conclusions in section [ sec : sec6 ] .
consider a household with @xmath0 appliances , where the @xmath1-th appliance ( @xmath2 ) has @xmath3 non - off states .
for example , a conventional light bulb would have one non - off state .
vector @xmath4 contains the voltampere ( va ) ratings of the @xmath3 non - off states of the @xmath1-th appliance .
let @xmath5 . with this we construct vector @xmath6 $ ] that contains all va ( or watt ) ratings of all @xmath0 appliances .
for the @xmath7-th time instant , the indicator of each non - off state is stored in a vector @xmath8 as @xmath9\in\{1,\ 0\ } \mathrm{\ for\ } i=1,2,\ldots , m,\ ] ] with @xmath10 denoting that the particular state is active and @xmath11 denoting the state is inactive . at the @xmath7-th time instant , the smart meter yields the total va reading @xmath12 , which is the sum of all power drawn by @xmath0 appliances at that time .
this can be expressed as @xmath13 where @xmath14\in\mathbb{r}^{n}$ ] and vector @xmath15 contains va draws of each appliance that is turned on .
the binary matrix @xmath16 in ( [ eqn : eqn2 ] ) is a block diagonal matrix given as @xmath17 where @xmath18 is a unity vector for @xmath19 .
the objective of load disaggregation is to find which appliance states are active at the @xmath7-th time instant . specifically , the goal is to find @xmath8 in ( [ eqn : eqn2 ] ) by using the known quantities @xmath12 , @xmath20 , and @xmath21 .
similarly to @xcite , load disaggregation can be formulated as an integer programming ( ip ) problem @xmath22 like any ip , this problem can be solved by exhaustive search over all possibilities for @xmath8 , however this approach can be prohibitive even for a moderate number of appliances .
the alternative is to explore more efficient ip solvers @xcite . before proceeding to enhancements , we reformulate ( [ eqn : eqn4 ] ) as a mixed - integer linear ip .
let @xmath23 .
since any appliance must be in exactly one state at any given time and vector @xmath8 is an indicator for non - off states , we can formulate this constraint as @xmath24+\mathbf{b}_k[q_j+2]+\cdots+\mathbf{b}_k[q_j+l_j]\le 1\ ] ] for @xmath25 .
let @xmath26 where @xmath27 denotes the element - wise product .
further , let @xmath28 @xmath29 @xmath30 $ ] , @xmath31 @xmath29 @xmath32 $ ] , @xmath33 $ ] , and @xmath34 $ ] , where @xmath35 is an auxiliary real variable
. we can now rewrite the quadratic ip in ( [ eqn : eqn4 ] ) as a mixed - integer _ linear _
ip @xmath36 subject to @xmath37&\in&\{1 , 0\ } \textrm{\ for\ } i=2,3,\ldots , m+1\end{aligned}\ ] ] where matrix @xmath38 and vector @xmath39 in ( [ eqn : eqn10 ] ) are given by @xmath40 @xmath29 @xmath41 @xmath42 @xmath43 $ ] and @xmath39 @xmath29 @xmath44 @xmath45 @xmath46 $ ] and the rows of the binary matrix @xmath47 are copies of the rows of matrix @xmath20 that have more than one nonzero element .
the solution of ( [ eqn : eqn9])([eqn : eqn11 ] ) leads to correct disaggregation only if the elements in @xmath21 are not binary combinations ( linear combinations with coefficients @xmath11 or @xmath10 ) of each other , and the deviations from the steady - state power / current draw during transients do not overlap with the steady - state draws of other appliances , or their combinations . in practice , these criteria are rarely met , so disaggregation based on ( [ eqn : eqn9])([eqn : eqn11 ] ) would yield unsatisfactory results .
this can also be true for an appliance with an infrequently occurring state with a high rating , since an undetected state with high rating would significantly reduce the accuracy score ( section [ sec : sec5 ] ) . in the next section
we introduce several enhancements that are meant to overcome the aforementioned limitations of the above ip disaggregation .
the proposed aided linear ip ( alip ) for load disaggregation incorporates several enhancements to the ip given in ( [ eqn : eqn9])([eqn : eqn11 ] ) , each of which is discussed next .
the first enhancement involves additional constraints that help resolve ambiguities related to the possible non - uniqueness of the solution to the ip .
first , consider appliances like refrigerator , surveillance camera , smoke detector , heat pump , etc .
, that happen to switch between the `` sleep mode '' and one or more higher - power states . in other words , these appliances always draw some power
. we can incorporate this additional information as an equality constraint to be added to ( [ eqn : eqn9])([eqn : eqn11 ] ) , @xmath48 where each row in the binary matrix @xmath49 has unity elements only in those positions that correspond to the states of those appliances in vector @xmath50 that remain turned on at all time .
next , consider the scenario where the rating of a given state of an appliance is equal to the sum of the ratings of some of the states of other appliances .
for example , an appliance h1 could have a va rating @xmath51 in one of its states , and appliances h2 and h3 could have va ratings of @xmath52 and @xmath53 , respectively , in some of their states .
then a reading @xmath54 could be interpreted in two ways - either h1 is on , or h2 and h3 are simultaneously on . to break such ties , we assume the minimum number of appliances is on at any given time - a heuristic that does not always hold , but turns out to be surprisingly good based on empirical evidence in existing datasets . this can be incorporated into the ip by using an additional row in the binary @xmath38 and an additional @xmath10 in @xmath39 in ( [ eqn : eqn10 ] ) . the additional row in @xmath38 would have unity elements only in those positions that correspond to the states where the ratings become binary combinations of each other . finally , consider the scenario where the rating of a given appliance ( say h1 ) is equal to the amount by which the steady - state rating of another appliance ( say h2 ) differs from its transient va measurement .
the transient reading may cause the ip solver to declare both h1 and h2 as on , even though only h2 is in the on transient .
such situation can also be avoided by using an additional row in the binary matrix @xmath38 and an additional @xmath10 in @xmath39 in ( [ eqn : eqn10 ] ) , where the additional row in @xmath38 has unity elements only in those positions that correspond to the states whose combinations produce a transient rating of another appliance .
many appliances operate as finite state machines and their possible state transitions can be described by a state transition diagram ( std ) .
for example , the fridge ( frg ) appliance from the dataset in @xcite has the std shown in fig .
[ fig : fig1 ] .
this offers the possibility to correct the output of an ip solver if it happens to violate the std .
for example , if frg was in state @xmath55 at time @xmath56 , then at time @xmath7 it can only be in @xmath55 or @xmath57 .
if the ip solver output suggests otherwise , we know there must be an error , and can therefore select either @xmath55 or @xmath57 , depending on which of the two options yields lower cost @xmath58 in ( [ eqn : eqn9 ] ) .
the same type of correction can also be applied backwards ( for example , the only way to get to @xmath59 is either from @xmath57 or @xmath59 ) , although we did not incorporate such processing in our experiments .
median filtering can help filter out implausible events such as frequent switching between states , which may occur in the ip solver output if the ratings of a particular appliance are much smaller than the total reading @xmath12 .
consider the appliance b1e from @xcite , which has a fully - connected 2-state diagram ( fig .
[ fig : fig1 ] with states @xmath59 and @xmath60 deleted ) .
although any transition between these two states is possible , it is implausible that the appliance changes state at each sampling instant ; we expect it to stay in a state for at least a few sampling instants . to enforce this , we apply the correction rule @xmath61 for @xmath62 in the solution obtained by the ip solver . in other words ,
the state estimate at time @xmath63 ( i.e. , @xmath64 ) is corrected based on the current state estimate @xmath65 and the corrected state does not affect the subsequent median filter outputs .
analogous corrections are applied to all states of all appliances . as mentioned before
, vector @xmath21 contains steady - state appliance ratings , which could be obtained from appliance data sheets or measurements .
however , va ( or watt ) values for transients between states are usually much more difficult to obtain , and even if this were possible , incorporating transient state ratings into the model would tremendously increase the model size ( i.e. , number of states ) @xmath66 . yet ,
if the sampling instance happens to catch a transient of one or more appliances , it could lead to incorrect solution of the ip .
for this reason , we develop a method to refine the ip solution using only the minimum and maximum transient rating of each appliance , which is easier to obtain , either from the data sheet or measurement .
let @xmath67 and @xmath68 be the vectors of the same size as @xmath21 that contain , respectively , the minimum and maximum transient ratings for each state of each appliance .
let vector @xmath69 contain indices of @xmath21 for which @xmath70 @xmath29
@xmath71 @xmath29 @xmath72 .
such states do not exhibit transient behavior .
let the vector @xmath73 contain the indices of other , potentially transient , states , that have been declared active by the ip solver ( i.e. , the corresponding value in @xmath8 is @xmath10 ) .
if @xmath73 is non - empty , then the current measurement may contain transient states and the solution given by the ip solver needs to be refined .
let @xmath74 @xmath29 @xmath75 and @xmath76 . to refine the solution
, we solve the following problem @xmath77 subject to @xmath78 constraints ( [ eqn : eqn19])([eqn : eqn20 ] ) force the non - transient states to match the steady - state ratings in @xmath21 while requiring potentially transient states to be between the corresponding minimum and maximum
. the cost function can be simplified by subtracting out the steady - state portion of the measurement , @xmath79
@xmath29 @xmath12 @xmath80 @xmath81 , and focusing on the transient portion of @xmath82 , i.e. , @xmath83 . then , applying a similar procedure as in section [ sec : sec2 ] , the problem can be converted to a linear programming problem . with @xmath84 $ ] , @xmath85 @xmath29 @xmath86 $ ] , @xmath31 @xmath29 @xmath32 $ ] , and @xmath33 $ ] , the problem becomes @xmath87 subject to @xmath88 where matrix @xmath38 and vector @xmath39 in ( [ eqn : eqn24 ] ) are given by @xmath40 @xmath29 @xmath89 @xmath90 @xmath91 @xmath92;$ ] @xmath93 @xmath94 $ ] and @xmath95 respectively .
the flowchart of alip is shown in fig .
[ fig : fig2 ] .
further steps such as time - of - day probabilities can be incorporated in order to further improve disaggregation accuracy .
we compare the performance of our alip method with the ip - based disaggregation from @xcite in terms of two accuracy measures @xcite : per appliance accuracy @xmath96|}\ ] ] and overall accuracy @xmath97|}\ ] ] where @xmath98 $ ] is the ground - truth rating of the @xmath1-th appliance at time index @xmath7 from the dataset and @xmath99 $ ] is its estimate obtained by disaggregation . in all experiments , the steady - state ratings of each appliance
were determined empirically from the datasets from the power consumption of each appliance separately . the maximum and minimum transient va or watt values for the alip disaggregator
were also determined empirically .
the aggregate va or watt totals for the appliances used in the experiments were computed from aggregated individual appliance readings . for alip
, the enhancements were applied in the order depicted in the flowchart in fig .
[ fig : fig2 ] .
all ratings and parameters used in the experiments can be found in the matlab code , which can be used to reproduce all the results . in experiment 1
we used @xmath100 appliances ( cde , frg , hpe , and b1e ) from the dataset in @xcite .
the number of states considered for the cde , frg , hpe , and b1e appliances were @xmath101 , @xmath102 , @xmath102 , and @xmath103 , respectively .
the total number of samples considered was @xmath104 , which we partitioned into 72 blocks of data each containing @xmath105 samples .
this number of samples covers @xmath106 months worth of readings .
we computed ac and acc for each block .
the ac curves obtained by using the ip and proposed alip disaggregators for cde , frg , and b1e appliances are illustrated in fig .
[ fig : fig3 ] , along with the acc curves .
we have not illustrated ac curves for hpe , since both disaggregators produced very similar curves for this appliance .
it is seen from the ac , cde plots ( top - left in fig .
[ fig : fig3 ] ) that alip performs better than ip consistently in all blocks . the same is true for b1e ( bottom - left in fig .
[ fig : fig3 ] ) . from the ac , frg plots it is seen that the alip performs considerably better than ip in many blocks , while ip performs marginally better in some blocks .
this is because frg has occasional impulsive va readings which get filtered out by alip but not by ip .
overall , however , alip disaggregates frg better than ip . from the acc plots ( bottom - right of fig . [
fig : fig3 ] ) , it can be seen the alip disaggregator is overall more accurate than the ip disaggregator - usually by 58% , and in some cases by up to 20% .
the ac values obtained over the whole 9 months worth of data are given in table [ tbl : allres ] , while the overall acc measure for this and other experiments is given in table [ tbl : overallres ] .
as seen in these tables , alip outperforms ip on each appliance , as well as overall .
.results for exp .
17 [ cols="<,<,^,<,>,>",options="header " , ] [ tbl : exetime ] in experiment 6 we used @xmath107 appliances ( mic , lte , ukn , sbp , hea , and dif ) from house 5 in @xcite , with @xmath108 states .
the data were downsampled by a factor 10 to obtain the samples at 30-second intervals . here
, alip outperforms ip on all appliances , as well as in overall accuracy , as seen in tables [ tbl : allres ] and [ tbl : overallres ] , respectively .
finally , in experiment 7 we used @xmath109 appliances ( ele , bth , frg , ukn , lte , arc , and dif ) from house 6 in @xcite , with @xmath110 states .
the data were downsampled by a factor 10 to obtain the samples at 30-second intervals . again
, alip outperforms ip consistently on all appliances , as well as in overall accuracy , as seen in tables [ tbl : allres ] and [ tbl : overallres ] , respectively . in table
[ tbl : overallres ] we also include the published acc results of two state - of - the - art machine learning - based disaggregation approaches , @xcite and @xcite , on the redd dataset .
note that @xcite and @xcite did not report the results for all the houses .
although these methods used different downsampling rates and there is some uncertainty about the processing of data prior to testing , the comparison gives us a feeling for how the proposed alip would compare against machine learning - based disaggregation .
we note that against @xcite , alip scores on average 0.32 better . against @xcite
, it scores on average 0.03 better on houses 3 and 6 , and on average 0.04 lower on houses 1 and 2 .
based on these results , we conclude that the proposed alip is competitive in terms of accuracy with the state - of - the - art machine learning - based disaggregation approaches . a final word on complexity .
the proposed alip approach incorporates several additional processing steps compared to the conventional ip - based disaggregation .
hence , its computational complexity is slightly higher . in table
[ tbl : exetime ] we list the average execution time ( in milliseconds ) per sample of ip and alip disaggregators for experiments 17 , which were obtained in matlab 2015a using ` intlinprog ` and ` linprog ` ( with default settings ) on an intel(r ) core(tm ) i7 - 4770 [email protected] ghz processor with 16 gb ram .
as seen in the table , alip is only slightly slower than ip , and both disaggregators take less than 20 ms per data sample .
integer programming ( ip ) provides a natural way to solve the load disaggregation problem , by trying to determine which appliance states are active at any given time .
however , previous ip - based disaggregation is shown to run into problems on real data due to issues related to transient readings and in cases when some states are binary combinations of other states .
we proposed an aided linear ip ( alip ) approach for disaggregation that overcomes many of the shortcomings of the previous ip - based approach .
experimental results demonstrate significant accuracy advantage of alip over the previous ip - based disaggregation method , as well as competitive performance against two state - of - the art machine learning - based disaggregation approaches .
v. zdraveski , m. todorovski , d. trajanov , and l. kocarev , `` dynamic load balancing and reactive power compensation switch embedded in power meters , '' _ ieee transactions on circuits and systems ii : express briefs _ , 2016 .
ducange , f. marcelloni , and m. antonelli , `` a novel approach based on finite - state machines with fuzzy transitions for nonintrusive home appliance monitoring , '' _ ieee trans .
_ , vol . 10 , pp .
11851197 , may 2014 .
s. makonin , f. popowich , i. v. baji , b. gill , and l. bartram , `` exploiting hmm sparsity to perform online real - time nonintrusive load monitoring , '' _ ieee trans . smart grid _ ,
pp , no .
99 , pp . 111 , 2015 .
d. egarter and w. elmenreich , `` autonomous load disaggregation approach based on active power measurements , '' in _ pervasive computing and communication workshops ( percom workshops ) , 2015 ieee international conference on _ , pp . 293298 , ieee , 2015 .
d. egarter , a. sobe , and w. elmenreich , `` evolving non - intrusive load monitoring , '' in _ proc .
16th european conference on applications of evolutionary computation : evoapplications13 _ , ( vienna , austria ) , pp . 182191 , apr .
2013 .
s. makonin , b. ellert , i. v. baji , and f. popowich , `` electricity , water , and natural gas consumption of a residential house in canada from 2012 to 2014 , '' _ scientific data _ , vol . 3 , no .
160037 , pp . 112 , 2016 .
|
load disaggregation based on aided linear integer programming ( alip ) is proposed .
we start with a conventional linear integer programming ( ip ) based disaggregation and enhance it in several ways .
the enhancements include additional constraints , correction based on a state diagram , median filtering , and linear programming - based refinement . with the aid of these enhancements ,
the performance of ip - based disaggregation is significantly improved .
the proposed alip system relies only on the instantaneous load samples instead of waveform signatures , and hence works well on low - frequency data .
experimental results show that the proposed alip system performs better than conventional ip - based load disaggregation .
integer programming , combinatorial optimization , linear programming , load disaggregation , nilm
|
within the framework of grand unified theories ( gut ) , supermassive magnetic monopoles ( @xmath56gev ) would have been produced in the early universe as intrinsically stable topological defects when the symmetry of the unified fundamental interactions was spontaneously broken @xcite . at our epoch they should be searched for in the cosmic radiation as remnants of primordial phase transition(s ) .
the velocity range in which gut monopoles should be sought spreads over several decades @xcite .
if sufficiently heavy ( @xmath57gev ) , gut monopoles will be gravitationally bound to the galaxy with a velocity distribution peaked at @xmath58 . lighter monopoles ( @xmath59gev ) would be accelerated in one or more regions of coherent galactic magnetic field up to velocities of @xmath60 while other acceleration mechanisms ( e.g. in the neighborhood of a neutron star ) could bring them to relativistic velocities .
macro was a multipurpose underground detector ( located in the hall b of the laboratori nazionali del gran sasso , italy ) optimized for the search for gut monopoles with velocity @xmath61 and with a sensitivity below the parker bound ( i.e. @xmath62@xmath63s@xmath64sr@xmath64 @xcite ) .
the apparatus was arranged in a modular structure with overall dimensions of @xmath65m@xmath66 and was made up by three subdetectors : liquid scintillation counters , limited streamer tubes and nuclear track detectors ( cr39 and lexan ) @xcite . in this work
we describe an analysis technique which uses all of the macro subdetectors simultaneously in the search for fast moving magnetic monopoles ( i.e. @xmath67 ) in the cosmic radiation .
the results on a first data sample is also reported .
the analysis procedure uses the data coming from the scintillator and the streamer tube subdetectors to identify candidate events .
this is done by reconstructing the energy release ( using scintillators energy reconstruction processor erp as well as streamer tubes charge and time processor , qtp ) and the particle s trajectory ( using the streamer tubes digital hit information ) .
the selected tracks are then searched for in the track - etch layers as a final tool for their rejection or confirmation .
since the three techniques are independent and complementary , we obtain a good rejection power against the background due to high energy muons and a high reliability of the possible candidates . in the following sections we will describe the signatures of a fast monopole in the macro detector and the analysis procedures .
analysis efficiencies and acceptance evaluations will then be given together with the results of a search performed on a first sample of more than two years of data .
a complete description of the macro detector , with particular emphasis on magnetic monopole detection techniques , has been given elsewhere @xcite .
here we will concentrate on the aspects which concern this particular analysis .
the energy loss suffered by a fast magnetic monopole in matter is due to atomic excitation and ionization .
thus the average rate of energy loss follows immediately from an extension of the bethe - bloch formula for a moving magnetic charge .
this allows a straightforward and well - grounded calculation of the detector response .
the peculiar characteristic of the fast magnetic monopoles is their large ionizing power compared either to considerably slower monopoles or to minimum ionizing electrically charged particles .
the energy released in a macro streamer tube by a @xmath68 monopole is at least one hundred times larger than that due to a minimum ionizing particle ( m.i.p . )
since the charge of the streamer pulse has been shown to have a logarithmic dependence on the ionizing power @xcite , it provides a good rejection of the muon background .
the response of the scintillation counters to a magnetic monopole of a given velocity is reported in @xcite .
the expected signal for a monopole is about a factor 30 greater than the m.i.p .
level for @xmath69 and is even larger for greater velocities . in the case of the track - etch detectors ,
the calculated value of restricted energy loss ( rel , defined as the fraction of the particle s energy loss contained in a 10@xmath70 nm diameter around its trajectory ) for a magnetic monopole with @xmath71 is @xmath72 60 mev @xmath73g@xmath64 , and it increases rapidly at higher velocities @xcite .
this value is well above the detection threshold of the macro cr39 nuclear track detector , rel@xmath74 = 26 mev @xmath73g@xmath64 .
the lexan has a much higher threshold compared to that of cr39 , making it sensitive to relativistic monopoles only .
the particular signatures of a fast monopole in macro are thus : @xmath39 ) a single track in the detector s spatial and temporal views corresponding to a particle with a velocity greater than @xmath75 , and @xmath40 ) a large energy deposition along the particle trajectory .
the background for this search is almost entirely due to the flux of high energy single muons . at the depth of macro , about @xmath76 of these have energies greater than the threshold energy above which the discrete energy loss processes of nuclear interactions , bremsstrahlung , and pair production start to dominate , viz . ,
these discrete mechanisms produce showering events in the apparatus .
this background is generally reduced by a combination of geometrical and energy cuts imposed on each of the subdetectors .
an analysis based on all three subsystems in macro allows the use of these cuts in a rather conservative way .
in particular , the energy cut remains safely within the linear response of the system .
moreover , any systematic errors are greatly reduced by the combination of measurements from the three subsystems .
energetic muons ( @xmath78 ) showering in the apparatus are efficiently rejected by requiring a clean single track in the streamer tube system . however
, this cut will also reject monopole events in which electromagnetic or hadronic showers are produced . in particular , a monopole with @xmath79 ( i.e. @xmath80 ) can produce a @xmath81-ray in a single electron collision with enough energy to produce an electromagnetic shower , giving rise to secondary tracks .
furthermore , processes such as hadronic interactions and @xmath82 pair production are effective for @xmath83 @xcite and will produce secondary tracks
. finally , another possible source of non - clean monopole - induced single tracks in the apparatus might be due to the nucleon decay catalyzed by monopoles following the rubakov - callan effect @xcite . due to these effects
the present analysis might lose efficiency in identifying monopole events if the catalysis cross section @xmath84 or @xmath80 .
this search begins with event selection performed by both the erp and the streamer tube horizontal monopole trigger ( shmt ) .
event reconstruction starts from the procedures used in the search for slow monopoles with the streamer system @xcite .
once event tracking has been performed , the energy loss observed in the scintillators and the analog response of the streamer tubes are used to achieve the necessary rejection power against the muon background .
the details of the event tracking are fully described in @xcite .
the strategy adopted to eliminate the radioactivity background is briefly summarized in the following .
the shmt selects events with at least 7 horizontal planes with a rough temporal alignment ( a `` @xmath85 slice '' ) . for each of the triggered @xmath85-slices ,
wire hits are selected on the basis of time measurements made by the qtps , while strip hits are identified by requiring a spatial match with the wire view .
if a track is found in each of the two spatial views , a more refined time track is searched for by using the qtp timing information from the selected spatial hits .
furthermore we require a minimum of 7,6 and 7 points along the track in the wire , strip and time views respectively .
only events satisfying these requirements and consisting of a single track reconstructed in each of the wire , strip , and time views are analyzed further .
the erp , a system measuring the energy deposition in each macro single scintillator and time of flight across the apparatus , has been extensively described elsewhere @xcite . in order to obtain @xmath86 ,
the energy lost by a particle in a scintillation counter , the raw adc values from each of the two tank ends are corrected for light attenuation in the scintillator and a normalized weighted mean of the two is taken .
the normalization constant for each scintillator is updated on a weekly basis by fitting the landau - like signal size distribution produced by a selected set of cosmic muon events .
correcting for the light attenuation requires knowledge of the longitudinal position of an event in a scintillator .
this can be obtained either from the streamer tube tracking system or from the difference of arrival times of the light at the two scintillator ends . in this analysis
, we use the former since it yields better energy resolution than the time - difference method . in fig.[fig : dpos ] the difference between the positions reconstructed with these two different methods is shown for a sample of cosmic muons .
the width of this distribution is primarily due to the time resolution of the scintillators and it yields @xmath87 . in fig.[fig : energy ] the value of the reconstructed energy deposition of a sample of muons is shown as a function of the longitudinal position of the event in the scintillator . as can be seen the correction for varying sensitivity as a function of this position is performed correctly in almost all the sensitive volume .
near the scintillator ends this procedure fails by underestimating the energy lost by the particle . for this reason
a fiducial volume is defined by cutting the last @xmath88 cm close to each end .
a study of the response of the erp to large signals has been fully described in @xcite .
the response of the whole chain from the photomultiplier tubes ( pmt ) to the recording analog - to - digital converting ( adc ) electronics was measured as a function of light level by pulsing each individual counter with uv light coming from a nitrogen laser ( @xmath89 nm ) after passing through a precision light attenuator .
the system has been shown to have a linear response up to a level corresponding to roughly 5 times the amount of light produced in a pathlength of @xmath90 cm by a minimum ionizing muon . in this search the synergetic use of the digital ( tracking ) and analog ( pulse charge ) information coming from the streamer tube system allowed putting analysis cuts below the onset of erp non - linearity . since in this search
we are interested in monopoles with velocity down to @xmath69 , the scintillator crossing time might exceed the width of the erp integration gate . in this case
only a fraction of the pmt pulse would be integrated , thus underestimating the energy lost in the scintillator .
the erp adc integration gate is @xmath91ns wide and the time spent by the digital electronics to produce it is @xmath92ns . as
a consequence the pmt signal to be integrated is delayed by @xmath93ns .
moreover the decay time of the slow component of the scintillation light , the electronic chain bandwidth , and the geometry of the detector actually set a minimum pulse duration of about @xmath94ns , which can be estimated by observing the muon pulses .
this implies that saturation effects might start for pulses produced by particles which traverse the scintillator in more than @xmath95 .
however in @xmath95 , a @xmath69 monopole travels about @xmath96 and loses at least @xmath97 @xcite .
since the analysis cut will be set at a much lower level , the width of the integration gate does not affect the detection efficiency .
the selection criteria adopted in this analysis for the erp hits are then summarized as follows : * the values given by the erp tdc s at the two different system thresholds ( _ high _ and _ low _ ) can not differ , at each scintillator end , by more than @xmath98ns .
this is essentially a check that the system is working properly .
* the measurements of the longitudinal position of a hit as performed by the streamer system tracking and by the erp tdc s , can not differ by more than @xmath99 cm ( see fig.[fig : dpos ] ) . *
a fiducial volume is defined to guarantee a proper energy reconstruction .
this is done by excluding @xmath88 cm close to the scintillator ends ( see fig.[fig : energy ] ) .
* the pathlength of the particle in the scintillator volume , as measured by using the streamer system track parameters , must be greater than @xmath100 cm .
for all the events in which at least two different scintillators are selected , the time of flight information is also available .
experimental studies of the dependence of the streamer charge on the particle s trajectory and ionizing power are reported in @xcite . in this section
we show the results of a study of the macro streamer tubes charge , as measured by the qtp system , using cosmic ray muons .
this is done to better understand the detector characteristics as well as the dependence of the charge on the geometry of the ionizing track , the gas mixture parameters , etc .
the understanding of these parameters allowed us to set up a procedure for distinguishing a m.i.p . from a highly ionizing particles such as a fast gut monopole . for each selected event
, qtp hits along the track are identified in order to measure the streamer pulse charge .
they must : * spatially match with the wire and strips hits ; * be in the time view within @xmath101 from the temporal track ; * not be preceded in the previous @xmath102 by another hit in the same qtp channel ( this is required in order to both reject afterpulses and avoid systematic errors in the charge measurements due to the adc s recovery time ) . in the analysis we considered
only wire hits with cluster size cls=1 or 2 in a given qtp channel . in the following
we will refer to the streamer pulse charge as measured by the qtp receiving the signals from the horizontal tubes , once the above mentioned selection criteria are applied . for slanted particle trajectories with respect to the direction of the anode wire , the probability to produce more than one streamer in the gas volume increases .
as a consequence , after a region ( which we refer to as the dead zone ) in which the multi - streamer production is strongly reduced by space charge effects due to the first avalanche , the pulse charge linearly increases for increasing @xmath103 , the length of the projection of the particle track ( in the sensitive volume ) along the direction of the wire . this dependence has been studied in reference @xcite using a muon beam . here
we report also the results of a study on the macro streamer tubes performed with cosmic muons crossing the detector . in particular , for each @xmath104 cm interval in @xmath103 , the charge distributions produced by cosmic muons have been fit .
gaussian fits have been separately applied for cls=1 and cls=2 hits .
the mean value of the fits are given as a function of @xmath103 in fig.[fig : q_lp ] .
these behaviors are in good agreement with the aforementioned results obtained in @xcite .
the dependence of the charge on @xmath103 comes out to be : @xmath105 where @xmath106 , @xmath107 and @xmath108 are the values of the charge produced by vertical muons , the slope of the linear part of the plot , and the width of the dead zone , respectively .
identical behaviors are obtained with vertical streamer tubes . only the results for @xmath109
cm are reported .
this is due to the angular distribution of the cosmic muons that limits the statistics for @xmath110 cm .
it must be noticed that , given the trigger requirements of the shmt , the restriction to the analysis of events with @xmath109 cm results only in a @xmath111 reduction of the geometric acceptance . the procedure used to obtain the plots shown in fig.[fig : q_lp ]
has been applied to the whole data sample . for calibration purposes
the values of @xmath106 , @xmath107 and @xmath108 have been measured on a run - by - run basis .
this allowed us to study and monitor the macro streamer tube response , its dependence on the gas mixture parameters and the correlation between the cls=1 and cls=2 values .
the behaviors of @xmath106 , @xmath107 and @xmath108 , for cls=1 and cls=2 , are reported in fig.[fig : calibrast ] as a function of the solar time .
the fluctuations in the value of the charge produced by vertical muons are due to small variations of the parameters of the gas mixture . in march 1994 , during hardware work for detector upgrade , there was an abnormal streamer charge response ( see fig.[fig : calibrast ] ) , due to a strong air contamination of the gas mixture .
this caused the tubes to operate in geiger mode .
small variations in the actual values of @xmath106 , @xmath107 and @xmath108 are very useful in order to study their mutual dependencies and to compare them with the results obtained in the literature @xcite .
one example is the dependence of @xmath107 on @xmath106 , which is shown in fig.[fig : slo_q ] for cls=1 hits .
since the increase of the pulse charge for larger values of @xmath103 is due to the production of more than one streamer avalanche along the particle trajectory , one should expect @xmath112 .
however , in agreement with what is reported in @xcite , this approximation fails for large values of @xmath106 , when it is not easy to take into account space charge effects in a simple phenomenological model .
the important feature of the first plot of fig.[fig : slo_q ] is that for a highly ionizing particle ( i.e. large @xmath106 ) the value of @xmath107 has to be larger than the corresponding value for a m.i.p . in the same detector configuration
this is particularly important for this analysis since it allows a better discrimination between fast monopoles and cosmic muons .
the results of the study of the macro streamer tube response and the aforementioned calibration procedure allowed us to correct the pulse charge measurements for both geometrical effects and gas mixture variations .
the charge measured by the qtp system for the selected hits has been normalized by dividing it by the value of the average charge produced by a muon in the same operational and geometrical conditions .
the muon charge is calculated by using eq.[eq : normaq ] with the value of @xmath103 for the event under consideration and the values of @xmath106 , @xmath107 and @xmath108 from the calibrations for a given cls and run . as can be seen in fig.[fig : calibrast ] , the values of @xmath106 , @xmath107 and @xmath108 for cls=2 have large spreads around their mean values .
this is a purely statistical effect in the fit to the various charge distributions .
however the ratios of the cls=2 constants to the cls=1 ones are very stable in the whole data set . as a consequence
, we avoided small effects due to the aforementioned spreads by using the values obtained for cls=1 multiplied by the measured scaling factor . in order to measure the streamer tube response to the passage of a particle
, we defined a variable @xmath113 , where @xmath114 is the charge actually measured for a given event and @xmath115 is that produced by a muon and calculated as described above . as was reported in @xcite @xmath116 has a logarithmic dependence on the particle ionizing power .
in particular , for a monopole with @xmath117 , @xmath118 is expected .
the value of @xmath116 has been measured in each selected event by using the qtp information coming from the horizontal streamer tubes for cls=1 and cls=2 hits.2 the poor statistics ( due to the muon angular distribution ) does not allow setting up a calibration procedure on a run by run basis .
] in this way @xmath50 different measurements for cls=1 ( @xmath119 ) and @xmath120 different measurements for cls=2 ( @xmath121 ) have been obtained along the particle trajectory .
the value of @xmath116 associated with the event is then obtained by averaging these @xmath122 measurements , weighted by the inverse of the variances of the charge distributions , @xmath123 and @xmath124 , as obtained from the data sample under consideration . in order to reduce the fluctuations of the value of @xmath116
, we do not include in the average the hit with the largest value of @xmath125 ( @xmath55=1 or 2 ) .
therefore @xmath126 which is calculated on @xmath127 different points .
given the considerations concerning the efficiency and the acceptance of the analysis ( see next sections ) , we required an event to have at least @xmath128 .
the distribution of @xmath116 for a data sample is shown in fig.[fig : gamma ] . as expected , the shape is almost gaussian with a tail towards the largest values that is due to both fluctuations and @xmath81 ray production by high energy muons .
as anticipated , a magnetic monopole with @xmath117 would have @xmath118
. however , detailed studies of the streamer formation process and of the effects of the read - out electronics on the pulse shape show that the actual value of the cut must be a suitable function @xmath129 as discussed in the following .
the analysis method can be summarized as a sequence of three different phases : run selection , event selection and reconstruction , and candidate selection . in the first phase all runs longer than 0.5 hours
are selected ( the average time length of a macro run is 6.5 hours ) .
evident electronic problems are ruled out by discarding runs with abnormal trigger rates .
this selection results in a total live time of more than @xmath130 of the solar time .
event selection requires that both the shmt and the erp trigger have fired .
then single tracks must be identified in the spatial and temporal views with the requirements listed in sec.[par : strea ] . given a space track ,
at least one scintillator must satisfy the selection criteria discussed in sec.[par : scint ] . for each of the selected erp hits ,
the energy reconstruction procedure is applied .
if at least two different scintillators fired , a precise time of flight measurement ( at the level of less than 1.ns ) is also possible .
the qtp hits for the streamer pulse charge measurements are then selected ( see sec.[par : qtpsel ] ) .
this allows the measurement of @xmath131 for all cls=1 and cls=2 wire hits in the horizontal streamer tubes , by using the streamer charge calibration constants previously produced for each run .
the calculation of the @xmath116 of the event is then made by making an average on at least @xmath132 qtp hits ( see sec.[par : gamma ] ) .
the average amount of events reconstructed following these requirements is @xmath133evt / yr . monopole candidate events are then selected .
the energy reconstructed in each selected scintillator counter must be @xmath134mev ( see sec.[par : erpcut ] ) . on the @xmath135evt / yr which are left a cut on @xmath116
is imposed : @xmath136 ( see sec.[par : gamcut ] ) .
finally the analysis of the possible survived candidates ( @xmath137evt / yr ) is made by the scanning of the track - etch sheets identified by the streamer track .
as can be envisaged from the previous sections , muon events pass the first two analysis steps but not the last , which is based on energy loss measurements in the scintillator , streamer and track - etch subdetectors .
this allowed the use of the cosmic muons as an important tool for an evaluation of the analysis efficiency .
the data from @xcite and some characteristics of the streamer tube read - out card have been used to compute the efficiency of the cut on the analog streamer response .
the shmt sensitivity to muons allows a simple and direct measure of the overall reconstruction efficiency @xmath138 of the analysis steps based on the streamer information .
it has been estimated according to the procedure fully discussed in @xcite .
basically @xmath138 is expressed as the product of the efficiency of the reconstruction algorithm @xmath139 times the efficiency @xmath140 due to the trigger and all the electronics used to produce the streamer data : @xmath141 .
actually the two contributions are not independent , leading to a conservative estimate of the overall efficiency .
the reconstruction efficiency @xmath139 for particle with @xmath142 is measured by the ratio between the rate of muons reconstructed by the analysis and the expected value .
then a check is performed of the independence of @xmath139 on the velocity of the particle .
this is done by allowing the natural radioactivity hits to simulate time alignements corresponding to slow particles .
electronic failures that can be responsible for trigger inefficiency , i.e. @xmath140 , are monitored by computing the ratio between the measured trigger rate and the expected one on the basis of the knowledge of the trigger circuitry and of the radioactivity background rate .
the scintillator counters considered in a given event are identified by using the tracking information provided by the streamer tubes .
the single scintillator efficiency @xmath143 , with respect to the event reconstruction procedure with the streamer tubes , is measured for each run and each counter by checking if the erp trigger associated with the tanks crossed by the reconstructed streamer track actually fired .
this single scintillator efficiency , @xmath143 , is very close to one except for a few counters with evident hardware problems .
the erp efficiency @xmath144 , has then been evaluated for each run by measuring the fraction of scintillators with @xmath145 .
this procedure is very sensitive to the overall erp efficiency and it automatically takes into account all the inefficiencies of the tracking algorithm . as calculated in @xcite , a magnetic monopole with @xmath146 produces a light yield in a scintillator which is more than 30 times larger than that due to the passage of a m.i.p . if we take into account that this estimate has a factor of two uncertainty @xcite , the minimum light yield produced by a monopole with the shortest accepted pathlength in the fiducial scintillator volume ( @xmath100 cm ) is @xmath147 where @xmath148 is the mass density of the scintillating liquid . for this value of @xmath86 ,
the non linearity effects ( due to the pmt and/or adc saturation ) are below the @xmath149 level and the energy resolution is about @xmath76 @xcite . therefore a cut which imposes @xmath150mev has an efficiency greater than @xmath151 .
as shown before , a monopole with @xmath152 would produce a track in the streamer system with @xmath153 .
the streamer charge would be even larger for faster monopoles .
the analysis requires @xmath154 .
the suitable value of @xmath129 is chosen to ensure an efficiency of at least @xmath1 for recognizing a monopole . to evaluate @xmath129 ,
a complete and detailed simulation has been performed of the full analysis chain starting from the production of the electronic signal . to ensure that the input waveforms for the simulations were correct , we collected @xmath155mhz wfd data , at different @xmath103 values , from a streamer tube operating in a muon beam @xcite under conditions very similar to those in macro .
since @xmath156 a monopole with @xmath152 is expected to produce streamer pulses with a charge four times higher than that produced by a muon , @xmath157 the dead region length @xmath108 is essentially independent on @xmath106 , and @xmath158 the slope @xmath107 increases with @xmath106 , the monopole charge parameters are conservatively set at : @xmath159 in order to simulate the waveform of the signal produced by a monopole , the muon streamer response was multiplied by the ratio of the charge produced by a monopole and that by a muon as computed with eq.[eq : normaq ] and eq.[eq : qmono ] . in the simulation , the effect of the protection diodes of the read - out cards is also considered .
they actually introduce an amplitude saturation of wire signal outputs corresponding to more than three times the signal produced by a m.i.p .
( see fig.[fig : diodo ] ) . for each @xmath103 value both muon and monopole waveforms
have been conservatively saturated at a level corresponding to two times the amplitude of the single streamer pulse .
the values of @xmath116 have been determined , separately for muon and monopole events , by using the charges of the simulated pulses at different @xmath103 . for this purpose a detailed monte carlo simulation of the macro detector
provided the number of hits with cls=1 and cls=2 for muons and monopoles , taking into account the different angular distributions and the different effects of the trigger requirements for monopoles and muons .
a result of this procedure for @xmath160 monopoles is shown in fig.[fig : mumono ] .
the @xmath116 distributions of muons and monopoles are well separated for small @xmath103 while the effect of the saturation due to diodes is clearly visible for larger @xmath103 . starting from these data ,
a function @xmath129 was extracted for which at least @xmath1 of monopoles have @xmath161 . for monopoles with @xmath68 , a larger value of @xmath116
is expected . however , in the whole @xmath85 range covered by this analysis , we conservatively used the @xmath129 estimated for @xmath160 as shown above .
the analysis procedure described above has been performed on a first data sample collected since december 1992 to september 1996 , with the exclusion of the period june 1993-december 1994 , when the efficiency of the streamer and scintillator subdetectors were affected by the works due to the upgrading of the apparatus with the construction of the upper part . in particular , by considering the erp and shmt rates , the erp efficiency and the calibration of the streamer charge response we have selected three data taking periods with a total live time of more than 16,000 hours .
the main characteristics of the considered data set , as well as the values of the analysis efficiencies discussed above , are listed in tab.[tab : period ] .
once all the selected events were fully reconstructed , the 150 mev energy cut mentioned above was applied .
then the streamer charge on the surviving events was analyzed .
the result is shown in fig.[fig : lpcut123 ] where the dependence of @xmath116 on @xmath103 is reported together with the function @xmath162 previuosly described .
._[tab : period ] main characteristics of the analyzed data sample .
the values of the analysis efficiencies ( see sec.[par : effi ] ) and the integrated live times are also reported . _ [ cols="^,^,^,^,^",options="header " , ]
|
we describe a search method for fast moving ( @xmath0 ) magnetic monopoles using simultaneously the scintillator , streamer tube and track - etch subdetectors of the macro apparatus .
the first two subdetectors are used primarily for the identification of candidates while the track - etch one is used as the final tool for their rejection or confirmation . using this technique ,
a first sample of more than two years of data has been analyzed without any evidence of a magnetic monopole .
we set a @xmath1 cl upper limit to the local monopole flux of @xmath2 in the velocity range @xmath3 and for nucleon decay catalysis cross section smaller than @xmath4 .
= 100 the macro collaboration + = 10000 m. ambrosio@xmath5 , r. antolini@xmath6 , g. auriemma@xmath7 , d. bakari@xmath8 , a. baldini@xmath9 , g. c. barbarino@xmath5 , b. c. barish@xmath10 , g. battistoni@xmath11 , y. becherini@xmath12 , r. bellotti@xmath13 , c. bemporad@xmath9 , p. bernardini@xmath14 , h. bilokon@xmath15 , c. bloise@xmath15 , c. bower@xmath16 , m. brigida@xmath13 , s. bussino@xmath17 , f. cafagna@xmath13 , m. calicchio@xmath13 , d. campana@xmath5 , m. carboni@xmath15 , r. caruso@xmath18 , s. cecchini@xmath19 , f. cei@xmath9 , v. chiarella@xmath15 , b. c. choudhary@xmath10 , s. coutu@xmath20 , g. de cataldo@xmath13 , h. dekhissi@xmath8 , c. de marzo@xmath13 , i. de mitri@xmath14 , j. derkaoui@xmath8 , m. de vincenzi@xmath17 , a. di credico@xmath6 , o. erriquez@xmath13 , c. favuzzi@xmath13 , c. forti@xmath15 , p. fusco@xmath13 , g. giacomelli@xmath12 , g. giannini@xmath21 , n. giglietto@xmath13 , m. giorgini@xmath12 , m. grassi@xmath9 , a. grillo@xmath6 , f. guarino@xmath5 , c. gustavino@xmath6 , a. habig@xmath22 , r. heinz@xmath16 , e. iarocci@xmath23 , e. katsavounidis@xmath24 , i. katsavounidis@xmath25 , e. kearns@xmath26 , h. kim@xmath10 , s. kyriazopoulou@xmath10 , e. lamanna@xmath27 , c. lane@xmath28 , d. s. levin@xmath29 , p. lipari@xmath30 , n. p. longley@xmath31 , m. j. longo@xmath29 , f. loparco@xmath13 , f. maaroufi@xmath8 , g. mancarella@xmath14 , g. mandrioli@xmath12 , s. manzoor@xmath32 , a. margiotta@xmath12 , a. marini@xmath15 , d. martello@xmath14 , a. marzari - chiesa@xmath33 , m. n. mazziotta@xmath13 , d. g. michael@xmath10 , p. monacelli@xmath18 , t. montaruli@xmath13 , m. monteno@xmath33 , s. mufson@xmath16 , j. musser@xmath16 , d. nicol@xmath9 , r. nolty@xmath10 , c. orth@xmath26 , g. osteria@xmath5 , o. palamara@xmath6 , v. patera@xmath23 , l. patrizii@xmath12 , r. pazzi@xmath9 , c. w. peck@xmath10 , l. perrone@xmath14 , s. petrera@xmath18 , v. popa@xmath34 , j. reynoldson@xmath6 , f. ronga@xmath15 , a. rrhioua@xmath8 , c. satriano@xmath7 , e. scapparone@xmath6 , k. scholberg@xmath35 , a. sciubba@xmath23 , p. serra@xmath12 , m. sioli@xmath12 , g. sirri@xmath12 , m. sitta@xmath36 , p. spinelli@xmath13 , m. spinetti@xmath15 , m. spurio@xmath12 , r. steinberg@xmath28 , j. l. stone@xmath26 , l. r. sulak@xmath26 , a. surdo@xmath14 , g. tarl@xmath29 , v. togo@xmath12 , m. vakili@xmath37 , c. w. walter@xmath26 and r. webb@xmath38 .
+ 1 .
dipartimento di fisica delluniversit di bari and infn , 70126 bari , italy + 2 .
dipartimento di fisica delluniversit di bologna and infn , 40126 bologna , italy + 3 .
physics department , boston university , boston , ma 02215 , usa + 4 .
california institute of technology , pasadena , ca 91125 , usa + 5 .
department of physics , drexel university , philadelphia , pa 19104 , usa + 6 .
laboratori nazionali di frascati dellinfn , 00044 frascati ( roma ) , italy + 7 .
laboratori nazionali del gran sasso dellinfn , 67010 assergi ( laquila ) , italy + 8 . depts . of physics and of astronomy , indiana university , bloomington , in 47405 , usa + 9 .
dipartimento di fisica delluniversit dellaquila and infn , 67100 laquila , italy + 10 .
dipartimento di fisica delluniversit di lecce and infn , 73100 lecce , italy + 11 .
department of physics , university of michigan , ann arbor , mi 48109 , usa + 12 .
dipartimento di fisica delluniversit di napoli and infn , 80125 napoli , italy + 13 . dipartimento di fisica delluniversit di pisa and infn , 56010 pisa , italy + 14 .
dipartimento di fisica delluniversit di roma `` la sapienza '' and infn , 00185 roma , italy + 15 .
physics department , texas a&m university , college station , tx 77843 , usa + 16 .
dipartimento di fisica sperimentale delluniversit di torino and infn , 10125 torino , italy + 17 .
l.p.t.p , faculty of sciences , university mohamed i , b.p .
524 oujda , morocco + 18 .
dipartimento di fisica delluniversit di roma tre and infn sezione roma tre , 00146 roma , italy + @xmath39 also universit della basilicata , 85100 potenza , italy + @xmath40 also infn milano , 20133 milano , italy + @xmath41 also istituto tesre / cnr , 40129 bologna , italy + @xmath42 also universit di trieste and infn , 34100 trieste , italy + @xmath43 also dipartimento di energetica , universit di roma , 00185 roma , italy + @xmath44 also institute for nuclear research , russian academy of science , 117312 moscow , russia + @xmath45 also institute for space sciences , 76900 bucharest , romania + @xmath46 macalester college , dept . of physics and astr .
, st .
paul , mn 55105 + @xmath47 also department of physics , pennsylvania state university , university park , pa 16801 , usa + @xmath48also dipartimento di
fisica delluniversit della calabria , rende ( cosenza ) , italy + @xmath49 also department of physics , james madison university , harrisonburg , va 22807 , usa + @xmath50 also rpd , pinstech , p.o .
nilore , islamabad , pakistan + @xmath51 also dipartimento di scienze e tecnologie avanzate , universit del piemonte orientale , alessandria , italy + @xmath52 also u. minn .
duluth physics dept . ,
duluth , mn 55812 + @xmath53 also dept . of physics , mit , cambridge ,
ma 02139 + @xmath54 also intervideo inc . , torrance ca 90505 usa + @xmath55 also resonance photonics , markham , ontario , canada + = 100
|
upon cooling below the melting point , liquids may either crystallize or enter a super cooled regime . in the latter case , as the glass transition temperature @xmath0 is approached , molecular motion gets slower and slower and the viscosity increases enormously .
the relaxation time increases by several orders of magnitude and for all practical purposes the system remains out of equilibrium .
although mechanically responding as a solid , structural relaxation is still present , slowing down as the system ages .
the response to a perturbation applied at a particular time @xmath1 will persist for very long times ( long term memory ) , preventing the system from reaching equilibrium .
while the system ages , one time quantities asymptotically tend to their equilibrium values while two times quantities depend explicitly both on the observation time and on the time when the perturbation was applied : time translation invariance ( tti ) is broken , which is a manifestation of history dependence . upon cooling the system gets trapped on long lived meta stable states which depend on the cooling rate , eventually escaping as a result of activated processes .
this complex glassy dynamics is reflected in several characteristic features , as non - exponential relaxation , breakdown of fluctuation - dissipation relations , etc , and there is a widespread believe that the underlying rugged energy landscape plays a fundamental role on the dynamics@xcite .
many aspects related to the structural glass transition and the nature of the glassy phase are still poorly understood @xcite , as a consequence of the inherent complexity of the underlying physics and the corresponding lack of a simple but non - trivial model .
only recently some analytical progress has been made with classic models based on lennard - jones or soft spheres potentials @xcite which , on the other hand , have been extensively studied via molecular dynamics simulations @xcite . meanwhile , in the field of spin glasses , disordered magnetic systems which share many physical properties with structural glasses , a reasonable theoretical understanding of the basic physics has been achieved at least at the mean field level @xcite .
recently it has been found that the equations describing dynamical correlation functions of a kind of mean field spin glasses simplify , above the transition , to the single equation of the mode coupling theory of super cooled liquids @xcite . in particular ,
several results point out that these models ( e.g. the @xmath2-spin model ) with one step of replica symmetry breaking and structural glasses are in the same universality class @xcite , indicating a deeper analogy between the physics of spin glasses and structural glasses than previously thought
. nevertheless , it is still matter of debate to what extent this analogy can be pushed forward . some important differences between these two kinds of systems
are evident : while a defining feature of spin glasses is the quenched disorder and frustration , no obvious quench disorder is present in structural glasses .
several models with frustration but no disorder have been studied recently @xcite .
another important difference is the fact that the dynamical variables in a spin glass are localized in space and do not diffuse as the molecules in a super cooled liquid .
we would like to have at hand a simple , yet non trivial , microscopic model of super cooled liquids in which the fruitful techniques developed in spin glass theory could be applied and possibly be extended to finite dimension .
moreover , mode coupling theory and mean field models in general do not describe activated processes , which are responsible for some important characteristics of the glass phase , as e.g. cooling rate dependence of the asymptotic state .
there are some other very simple models with kinetic constraints which reproduce quite well the glassy phenomenology @xcite , but being essentially dynamical they are hard to be studied with analytic techniques . in this paper
we describe salient dynamical features of a hamiltonian lattice model of structural glasses , the frustrated ising lattice gas ( filg ) .
the interesting properties revealed by the model rest on an interplay between two different kinds of degrees of freedom : diffusive particles or translational degrees of freedom and internal or orientational degrees of freedom .
we present results of monte carlo simulations of one and two times quantities , like density relaxation , density - density correlations and responses , correlations and responses of internal degrees of freedom , mean square displacements of the particles and compression experiments .
this three dimensional model may serve as a numerical laboratory to perform a systematic study of the role played by activated processes on the long time dynamics of glass formers , a feature that is missing from mean field approximations .
the paper is organized as follows . in section [ section.filg ]
we review the essential results both from equilibrium and out of equilibrium dynamics for the filg . in section [ section.cycle ] results for slowly varying chemical potentials
are presented and compared with the case where a sudden quench is performed , while in section [ section.fdt ] the breakdown of the fluctuation - dissipation theorem in this model is analyzed . finally , in section [ section.conclusions ]
, we present some conclusions and discuss some open questions .
the filg @xcite is defined by the hamiltonian : @xmath3 the dynamical variables @xmath4 ( @xmath5 ) are local densities or site occupations . the @xmath6
represent internal degrees of freedom , e.g. rotational ones .
although molecules in glass forming liquids may assume several spatial orientations , here we take the simpler case of only two possibilities , @xmath7 .
the usually complex spatial structure of these molecules is in part responsible for the geometric constraints , imposed by the neighborhood , on their translational and rotational dynamics .
this hindrance effect is mimicked by the quenched random variables @xmath8 . when @xmath9 , in order to minimize @xmath10 , either the spins should satisfy the bond @xmath11 or at least one of the sites @xmath12 or @xmath13 must be empty .
in this limit , the ( site ) frustrated percolation @xcite model is recovered , where no frustrated link can be fully occupied , implying that any frustrated loop in the lattice will have a hole and then @xmath14 , preventing the system from reaching the close packed configuration .
the equilibrium properties of the @xmath15 model have been studied in @xcite for @xmath16 and @xmath17 , which corresponds in practice to the frustrated percolation limit . in the low density regime ( @xmath18 )
the behavior is liquid like , time correlation functions decay exponentially , equilibration is quickly achieved and the particles mean squared displacement grows linearly with time , a simple diffusion scenario since particles hardly feel any constraint in their mobility . at @xmath19 there is a percolation transition , the corresponding density being @xmath20 .
this first transition manifests dynamically in the onset of two different relaxation regimes in the correlation functions , a fast exponential relaxation at short times and a slow , stretched exponential relaxation at longer times . while for intermediate times the diffusion is anomalous , for longer times it is linear with the diffusion coefficient becoming smaller as the density grows with growing chemical potential ( or equivalently , with lowering temperature ) .
a second transition appears for @xmath21 , corresponding to a density @xmath22 .
this is a glass transition at which the relaxation times diverge and the diffusion constant goes to zero . in the filg
the structural manifestation of this transition is the presence of a _ frozen _ percolating cluster .
this transition corresponds to the dynamical one in mean field @xmath2-spin or potts glasses or the ideal glass transition in the mode coupling theory .
@xcite studied an equivalent , non local version of the filg ( the site frustrated percolation , sfp ) in 2d , finding a behavior similar to the 3d filg , the main difference being the ( arrhenius ) dynamical singularity occurring at zero temperature .
it was also found in @xcite that the spin glass susceptibility @xmath23 , associated with the internal degrees of freedom of occupied sites , diverges at the same value of @xmath24 where a glass transition takes place .
this is a thermodynamic spin glass transition associated with the frozen - in of the internal degrees of freedom . on the other hand , the compressibility associated with the density variables does not present critical behavior but only a maximum as observed in many glass formers . in view of this behavior
one could ask whether a suitably defined nonlinear compressibility should diverge at the glass transition .
this should indicate the thermodynamic al character of the glass transition in this model .
a nonlinear compressibility , @xmath25 , can be introduced with @xmath26 where 1 and 2 label two real replicas of the system .
one could think that in this quantity , associated with fluctuations of an edwards - anderson like order parameter involving only density variables , a divergence analogous to the one present in the @xmath23 might be detected .
this quantity has been measured @xcite but no evidence of critical behavior has been found near @xmath27 : @xmath28 attains a plateau that seems to be roughly sample and size independent .
the fact that the spin variables present a thermodynamic transition while the particles do not points once more to the purely dynamical character of the glass transition .
of course , a thermodynamic transition may be present at a lower temperature ( or a higher chemical potential ) corresponding to the kauzmann @xmath29 and associated with a configurational entropy collapse , but this remains an open question in this model which deserves further study . at a mean field level ,
equilibrium properties have been studied in closely related models @xcite . for large @xmath24 all sites
are occupied and the behavior corresponds to the sherrington - kirkpatrick model , with a continuous spin glass transition . upon decreasing @xmath24 one reaches a tricritical point , below which the transition is first order . in this region , analogously to other first order transition models like the @xmath2-spin and the potts glass @xcite , above the critical temperature there is a dynamical transition where meta stable states first appear@xcite .
we still do not know to what extension this scenario remains valid in finite dimensions .
a characteristic feature of the out of equilibrium dynamics of glassy systems is aging @xcite .
slow aging dynamics is present in the filg both in density - density two times correlations as well as in correlations of the internal degrees of freedom @xcite after performing sudden quenches in @xmath24 from a very small value characteristic of the liquid phase ( low density ) to a high value corresponding to the glassy phase .
performing a quench in @xmath24 is similar to the application of a sudden compression , we raise the density of the system keeping the ratio @xmath30 in order to prevent close packing .
connected two point correlations can be defined as : @xmath31 where the global density at time @xmath32 is given by @xmath33 .
we now define the density autocorrelations as @xmath34 . in fig .
[ densitycorrelation ] the behavior of @xmath35 is shown as a function of @xmath32 in a semi log plot after a quench in chemical potential to a value @xmath36 , for waiting times between @xmath37 and @xmath38 .
a typical aging scenario is present signalling the slowing down of the dynamics as the waiting time grows .
for the longest waiting times the correlation presents a rather fast relaxation to a plateau in which the system evolves in quasi equilibrium : the dynamics is stationary and the fluctuation - dissipation relations hold .
the plateau separates two time scales typical of glassy systems : a @xmath39 ( fast ) relaxation for short times and an @xmath40 ( slow ) relaxation at longer times , corresponding respectively to the fast movements of the particles inside the dynamical cages and the large scale , cooperative process that takes much more time in order to rearrange the cages . moreover , in this very long time regime ( @xmath41 ) , the system falls out of equilibrium , the correlations decay to zero asymptotically and time translational invariance ( tti ) no longer holds with the corresponding violation of the fluctuation - dissipation theorem ( fdt ) , as will be shown in the following sections .
a good scaling of the correlation functions is obtained assuming a time dependence of the form @xmath42 with @xmath43 given by @xmath44 $ ] where @xmath45 and @xmath46 is a microscopic time scale @xcite .
this form is quite general as one recovers the cases of a simple @xmath47 dependence ( full aging ) when @xmath48 and stationary dynamics when @xmath49 @xcite .
a similar aging scenario has been found associated with the relaxation of internal degrees of freedom @xcite .
when the system is gently compressed the glass transition depends on the rate of compression . for a fixed temperature
we have done simulations of slow compression experiments , where at each monte carlo step ( mcs ) @xmath24 is increased by a quantity @xmath50 .
[ annealing ] shows the evolution of the density for several compression rates , ranging from @xmath51 to @xmath52 .
it can be noted that for slower compressions , the final density is higher . up to @xmath53 the density remains small enough and the system is able to equilibrate sufficiently fast , regardless of @xmath50 , and all curves collapse . as the density increases , the equilibration time also increases since the system cooperative rearrangement involves a larger number of particles in a smaller region . in analogy with cooling experiments ,
for each compression rate the system gets out of equilibrium at a different value of @xmath24 and the asymptotic density increases with decreasing rates .
the points where the curves for each rate depart from the master curve define the glass transition point @xmath54 .
the dependence of @xmath54 on the compression rate is characteristic of glass formers and is thought to be consequence of activated processes : no cooling ( compression ) rate dependence is observed in mean field models where the glass transition appears at a fixed temperature ( chemical potential ) . for a fixed @xmath24 , by decreasing the compression rate , the density @xmath55 tends to the equilibrium density @xmath56 as a power law @xmath57 , as can be seen in the inset of fig .
[ annealing ] , the parameters depending on the value of @xmath24 .
the extrapolated equilibrium value is slightly higher than the one obtained upon fast compressions , as will be shown later . in analogy with temperature cycling experiments in spin glasses which show the onset of irreversible processes , here , by increasing and decreasing the chemical potential at a given rate
, hysteresis cycles can be produced whose internal area increases with @xmath50 .
moreover , the chemical potential range in which the system behaves in an irreversible way also increases with @xmath50 . in fig .
[ cycle ] the hysteresis curves are presented for several cycling rates .
notice that the lower left part of the cycle coincides with the equilibrium curve and for very slow rates the cycle tends to disappear and we recover the above annealing equilibrium curve . as before ,
the maximum achieved density increases for slower rates . instead of slowly compressing the sample , high densities
can also be obtained by a fast compression analogous to temperature quenching experiments .
here we consider an initial state with an empty lattice , corresponding to @xmath58 .
then the chemical potential is suddenly increased to a value above the critical one . in fig .
[ dens ] we show how the density increases after a sudden quench to different @xmath24 values , ranging from 1 to 10 .
for all @xmath59 the density converges more or less rapidly to an asymptotic value , which depends on @xmath24 .
for @xmath60 the density increases with a law @xmath61 which seems to be independent of the chemical potential and that converges to the asymptotic value with a power law : @xmath62 .
the best value for the parameters are @xmath63 and @xmath64 , slightly lower than the value obtained in the slow compressions .
in contrast to a sudden compression , by gently annealing the system can approach equilibrium more efficiently .
nevertheless one can expect that the equilibrium density will be reached after a quench in a much longer time scale driven by activated processes . in fig .
[ density ] we show the relaxation of the density after a sudden compression to @xmath36 , deep in the glass phase .
if we assume that the density asymptotic value is essentially determined by the amount of frustration present in a sample , then it is possible to calculate the range of allowed values for @xmath65 . in a sample with no frustration obviously @xmath66 , while in a fully frustrated sample we have that @xmath67 .
a fully frustrated sample has the interactions @xmath11 such that every plaquette is frustrated .
the value we find for the random model , @xmath68 , falls in this interval , as it should , and is not very far from the one found in the fully frustrated case .
this would suggest that the filg defined on a fully frustrated lattice may be an interesting non random model with features similar to those presented in this paper .
in fact we believe that the glassy behavior of the filg is based much more on its frustration rather than on its randomness . as long as activated processes do not enter the scene we would expect a behavior reminiscent of mean field models . in the thermodynamic limit ,
mean field theory for the @xmath2-spin spin glass predicts that the internal energy will relax with a power law to a threshold value @xmath69 greater than the equilibrium one @xcite .
the absence of activated processes will prevent the system from escaping the threshold states and it will be kept out of equilibrium forever . in a true glass
the relaxation of the density at sufficiently low temperatures should behave this way . in the case of the filg
we do not know what the equilibrium density is , so in order to test the previous hypothesis we have measured autocorrelations at constant density .
considering again the connected two times autocorrelations ( eq . [ density - density ] ) , we present in fig .
[ constant ] results of simulations of quenches at fixed densities .
the upper figure shows results for a density that is slightly below the critical one ( @xmath70 ) for several waiting times @xmath1 .
there exists a characteristic waiting time after which all the curves collapse on a master one , meaning that aging is interrupted and time translation invariance ( tti ) is recovered .
the relaxation preserves the characteristic two steps with equilibrium dynamics for large @xmath1 . on the other hand , for a density @xmath71 , slightly above the critical one , the system seems to age forever , as can be seen in the lower figure . in this case , even for the largest waiting times , the quasi - equilibrium regime is followed by an out of equilibrium relaxation where tti no longer holds and fluctuation - dissipation relations are violated .
the behavior in this two cases is qualitatively different and is reminiscent of mean field behavior with a threshold or critical value for the density .
this will be reflected also in the form in which the fluctuation - dissipation theorem is violated , as will be discussed below . in real space , one way of seeing aging effects is looking at the diffusion properties of the particles .
the averaged mean square displacement ( msd ) in the time interval @xmath72 is defined by @xmath73 ^ 2\ ] ] where the function @xmath74 is 1 if the @xmath12-@xmath75 particle is present at both times @xmath32 and @xmath1 and is 0 otherwise . in this way
only particles that do not leave the system enter the measures . obviously , if the density is kept constant , @xmath76 . for the smallest waiting times , the density is low and diffusion is like in a simple fluid , @xmath77 being linear in @xmath32 ( see fig .
[ diffusion ] ) . only for large times
the density attains a large value , cages are formed trapping the particles , and the diffusion is arrested , as signaled by the appearance of a plateau . for larger @xmath1 and @xmath78 , the particles are rattling inside the cages formed by their neighbors and the diffusion is still normal , @xmath77 being again linear in @xmath32 . the fact that the first linear regime is longer the smaller is
@xmath1 is simply because for larger @xmath1 the density at @xmath79 is larger .
as soon as @xmath77 is comparable to the cage linear size , the diffusion halts and a plateau is attained . to escape from the cage ( hence from the plateau ) , a large , cooperative rearrangement must occur among neighbor cages .
this slow phenomenon is reflected in the extension of the plateau .
moreover , for larger times we are essentially measuring the diffusion of the cages , what can be seen as the second linear regime present in fig .
[ diffusion ] . from this figure
it is clear that aging effects with characteristic dynamical regimes are also present in real space properties of the system .
when glasses evolve slowly out of equilibrium fluctuation - dissipation relations are no longer valid . instead , in computer simulations and mean field approximations a generalized relation
is obtained @xcite : @xmath80}{t}\frac{\partial c(t , t_w)}{\partial t_w}\ , \label{gfdt}\ ] ] where @xmath81 is a two times correlation function and @xmath82 with @xmath83 is the associated response .
@xmath84 is the heat bath temperature and @xmath85 , the `` fluctuation - dissipation ratio''(fdr ) , is a function that measures the departure from fdt : at equilibrium @xmath86 and the usual fdt is recovered , while in the out of equilibrium regime , @xmath87 .
the function @xmath88 turns out to be a constant in mean field models of glasses , signalling that the asymptotic dynamics of the system is dominated by only one time scale ( apart from the fdt scale ) . in this time scale , corresponding to the aging regime , the slow degrees of freedom are at an effective temperature different from the bath one and that depends on the fdr as @xmath89 @xcite .
recently the first attempt at an experimental measurement of this effective temperature has been done in glycerol glass @xcite . in a finite system aging
will eventually be interrupted due to activated processes leading the system to equilibration at the temperature of the bath .
nevertheless the true role of activated processes in the long time dynamics of glassy systems still remains to be understood .
while an analytic approach to the problem is very difficult one may gain insight from simulations on finite systems .
we have measured the fdr on the filg @xcite .
the main result is shown in fig .
[ fdt_1 ] , where @xmath88 is given by minus the derivative of the data curves . in this case
we have measured density correlations and associated compressibilities ( for details see @xcite ) .
this result was obtained in a lattice of linear size @xmath90 .
the result is the one expected for a model having only two dominating time scales : the fdt one , when @xmath91 and another one , when the system is relaxing very slowly out of equilibrium and @xmath92 , being the edwards - anderson ( ea ) order parameter @xcite .
the constancy of @xmath88 in the aging regime is exactly what is predicted by mean field theory .
what is very remarkable is that this constancy seems to be valid not only for the asymptotic curve ( @xmath93 ) , but also for finite ( although large ) @xmath1 curves
. then we can express the fdr as function only of the smaller time , @xmath1 , and we observe that it changes very slowly with @xmath1 .
we also note that the fact that @xmath88 seems to depend only on the smaller time is in agreement with some recent analytical results . in @xcite it has been shown that slow glassy dynamics can be viewed as an evolution between `` quasi states '' , where the system ( almost ) equilibrates before relaxing further .
when the field is switched on at time @xmath1 it will select a particular set of quasi - states , those with the `` right '' magnetization .
if this is the case then the externally induced small perturbation @xcite will regress in the same way a spontaneous fluctuation does and one can show @xcite that the fdr may depend on the smaller time , but not on the larger one . and
this is exactly what we have observed numerically in the filg model .
nevertheless , in contrast with mean field predictions , the curves for different temperatures and chemical potentials in the glass phase ( for a fixed large @xmath1 ) remain parallel , simply shifting upwards with a corresponding decrease in the value of the ea parameter @xcite .
this is reasonable as one expects that the fdt regime will be the only one present as @xmath94 .
this behavior is not observed in mean field approximation , where the ea parameter has a discontinuous jump at @xmath0 and @xmath88 changes continuously for @xmath95 .
other studies in long range models seem to point out a continuous approach of @xmath96 to the bath one @xcite .
a systematic study of these effects is necessary in order to understand the underlying mechanisms responsible for them .
it is not clear whether the system will undergo also a static transition with one step of replica symmetry breaking , as the off - equilibrium dynamics would suggest .
thermalization at @xmath97 is really hard to attain and so we did not succeed in answering this question .
however , even in the case that all the observed dynamical features would not correspond to an underlying static transition and they would eventually disappear as the system equilibrates , the results are interesting for at least two reasons .
first , real glassy systems can be in the off - equilibrium regime all along their life and so it is more relevant the study of their out of equilibrium behavior than the static one .
second , even if what we are observing is due to finite time effects ( although very large ) , it is very remarkable that it is well described by mean field approximations . maybe is the case that on time scales where activated processes are very rare mean field theories ( that completely neglect activated processes ) are really a good approximation . we have also measured the fdr associated with the relaxation of the spin or internal degrees of freedom .
one can consider the spin - spin autocorrelation @xmath98 or , only taking spins from occupied sites , the spin - density - spin - density one @xmath99 by applying suitable perturbations , like small random magnetic fields , one can define in the standard way the integrated responses corresponding to the correlations defined above .
the parametric plot of the integrated responses versus correlations is shown in fig .
[ fdt_2 ] .
the interesting result which shows up here is that , qualitatively , the fdr is the same in the three cases .
furthermore , fits to the data show approximately the same asymptotic value of @xmath88 .
while the data for spin - density - spin - density and density - density are exactly on the same straight curve the data for spin - spin variables depart from the fdt line at a smaller value of the edwards - anderson parameter .
this is due to the fraction of spins placed on empty sites .
the fact that we have measured the same effective temperature @xmath96 perturbing different degrees of freedom is remarkable .
it is known that when different degrees of freedom are coupled they must reach the same temperature .
however to our knowledge this is the first time that such behavior has been measured in the aging regime of a finite dimensional system .
it is worth noting that there is strong evidence @xcite that the internal degrees of freedom are responsible for a thermodynamic spin glass transition at a temperature near @xmath0 , the structural glass transition temperature .
the constant value of the fdr for the spin variables indicates that the transition should be typical of systems with one replica symmetry breaking .
if this is confirmed we have at hand a finite model , with short range interactions , which presents a discontinuous transition , like that present in mean field models as the @xmath2-spin .
in fact , at a mean field level , a line of discontinuous transitions have been found in models closely related to the filg @xcite .
more work is needed in order to test this possibility in the short range model . for the sake of comparison , in fig .
[ 3corr ] results for all three correlations studied are shown for @xmath100 . notice that the correlation involving both occupation and spin variables is always smaller than the other two : @xmath101 . for small times
, @xmath102 is dominated by the particles dynamics , while , for large times , by the internal degrees of freedom dynamics .
we have described the main characteristics of the out of equilibrium dynamics of the three dimensional frustrated ising lattice gas .
it shows a very complex and interesting phenomenology similar to that observed in real glass formers .
the glass transition seems to have a purely dynamical origin as no evidence of singular behavior was found neither in the linear nor in nonlinear compressibilities .
aging dynamics is present in both density and internal degrees of freedom and the decay of autocorrelations is similar to that observed in molecular dynamics simulations of lennard - jones glasses .
cage effects are evident both in density relaxation and on diffusion properties where a clear characterization of different regimes is possible . from slow compression experiments different densities
are reached depending on the compression rate and hysteresis is observed from cycling experiments .
after a sudden compression the density slowly approaches a limiting value .
aging experiments at fixed densities slightly above and below this limiting value show clear qualitative differences showing that this density may separate two different dynamical regimes .
the overall scenario is very reminiscent of mean field scenarios for the glass transition , except for the presence of compression rate dependence which is absent in mean field .
interestingly , the form of the fluctuation - dissipation theorem violations is exactly that predicted in mean field approximations , i.e. a constant fluctuation - dissipation ratio .
this means that this model presents two very separated time scales .
the effective temperature in the out of equilibrium scale is nearly temperature independent in the glassy phase .
this raises the question whether activated processes play or not an important role in its dynamics .
the same constant effective temperature is observed associated with the internal degrees of freedom . at mean field level ,
a constant effective temperature has been obtained in systems characterized by one step replica symmetry braking ( 1rsb ) .
thermodynamically , a phase transition with a divergent susceptibility associated with internal degrees of freedom is observed .
this may be the first realization of a glassy phase characterized by 1rsb in a finite dimensional model .
the presence of a static transition associated with density variables at a low `` kauzmann temperature '' @xmath103 remains an open problem in this model .
it would be interesting to study in more detail how the restoration of fdt relations is achieved as @xmath0 is approached from below in order to test the strength of the time scales separation . also activated processes may show up for bath temperatures larger than those explored in these simulations as long as frustration begins to be lifted by thermal activation .
thanks s. franz for useful discussions on the subject .
this work was partly supported by the brazilian agency cnpq .
jja acknowledges the abdus salam international centre of theoretical physics for support during his stay , where part of this work was done .
r. bhmer , k.l .
ngai , c.a .
angell , and d.j .
plazek , j. chem . phys .
* 99 * , 4201 ( 1993 ) .
angell , science * 267 * , 1924 ( 1995 ) .
ediger , c.a . angell and s.r .
nagel , j. chem .
phys . * 100 * , 13200 ( 1996 ) .
w. gtze and l. sjgren , rep .
* 55 * , 241 ( 1992 ) .
w. gtze in _ liquids , freezing and the glass transition _ , les houches 1989 , eds .
hansen , d. levesque and j. zinn - justin ( north holland , amsterdam , 1991 ) .
bouchaud and m. mzard , j. phys .
i ( france ) * 4 * , 1109 ( 1994 ) .
e. marinari , g. parisi , and f. ritort , j. phys .
a * 27 * , 7647 ( 1994 ) .
cugliandolo , j. kurchan , g. parisi and f. ritort , phys .
lett . * 74 * , 1012 ( 1995 ) .
fredrickson and h.c .
andersen , phys .
lett . * 53 * , 1244 ( 1985 ) ; j. chem . phys .
* 83 * , 5822 ( 1985 ) . w. kob and h.c .
andersen , phys .
e * 48 * , 4364 ( 1993 ) .
j. kurchan , l. peliti , and m. sellitto , europhys . lett . * 39 * , 365 ( 1997 ) . l. peliti and m. sellitto , in _ disorder and chaos _ , ed .
a. vulpiani , ( editions de physique 1998 ) ; m. sellitto , preprint cond - mat/9912257 .
e. follana and f. ritort , phys .
e * 54 * , 930 ( 1996 ) .
i.s . graham , l. pich and m. grant , phys .
e * 55 * , 2132 ( 1997 ) .
arenzon , m. nicodemi , and m. sellitto , j. de physique * 6 * , 1143 ( 1996 ) .
m. sellitto , m. nicodemi , and j.j .
arenzon , j. de physique * 7 * , 45 ( 1997 ) .
schreiber , eur .
j. b * 9 * 479 ( 1999 ) .
a. coniglio , a. de candia , a. fierro and m. nicodemi , j. phys .
c * 11 * , a167 ( 1999 ) .
see e.g. j.p .
bouchaud , l.f .
cugliandolo , j. kurchan and m. mzard in @xcite and e. vincent , j. hammann , m. ocio , j.p .
bouchaud and l.f .
cugliandolo , in proceedings of the sitges conference on glassy dynamics , ed .
e. rubi ( springer , berlin , 1996 ) .
s. franz and h. rieger , j. stat .
phys . * 79 * , 749 ( 1995 ) .
e. marinari , g. parisi , f. ricci - tersenghi and j.j .
ruiz - lorenzo , j. phys .
a * 31 * , 2611 ( 1998 ) .
a. barrat , phys .
e * 57 * , 3629 ( 1998 ) .
g. parisi , f. ricci - tersenghi and j.j .
ruiz - lorenzo , eur .
j. b * 11 * , 317 ( 1999 ) .
g. parisi , phys .
lett . * 79 * , 3660 ( 1997 ) .
barrat and w. kob , europhys . lett . *
46 * , 637 ( 1999 ) .
m. sellitto , eur .
j. b * 4 * , 135 ( 1998 ) .
r. di leonardo , l. angelani , g. parisi and g. ruocco , preprint cond - mat/0001311 .
|
the dynamical properties of a three dimensional model glass , the frustrated ising lattice gas ( filg ) are studied by monte carlo simulations .
we present results of compression experiments , where the chemical potential is either slowly or abruptly changed , as well as simulations at constant density .
one time quantities like density and two times ones as correlations , responses and mean square displacements are measured , and the departure from equilibrium clearly characterized .
the aging scenario , particularly in the case of the density autocorrelations , is reminiscent of spin glass phenomenology with violations of the fluctuation - dissipation theorem , typical of systems with one replica symmetry breaking .
the filg , as a valid on - lattice model of structural glasses , can be described with tools developed in spin glass theory and , being a finite dimensional model , can open the way for a systematic study of activated processes in glasses .
= -1.5 cm 2
|
the presence of power - law behaviour in nature is such an extremely common phenomenon that considerable lore has now grown up concerning its genesis .
one of the most common situations in which it occurs is in the presence of scale - invariant systems .
such behaviour occurs , for instance , in any sort of thermodynamic system undergoing a second - order phase transition , and the behaviour of physical quantities ( such as susceptibilities ) in terms of the distance from criticality is typically given by a power - law : ( t - t_t_)^. second - order phase transitions are extremely well - described by statistical field theories .
the associated technical machinery ( the renormalization group ) , is now so well developed that it is sometimes difficult to remember that second - order phase transitions are not the _ only _ route to power - law behaviour .
the onset of power - law behaviour actually occurs at a much more primitive level , and can be analyzed directly in terms of the underlying differential equations .
scale invariance is commonly thought to be synonymous with power - law behaviour .
this is not quite correct : scale invariance is not enough to guarantee power - law behaviour , while power - law behaviour implies scale invariance only in the trivial sense that it is always possible to write down _ some _ scale - invariant differential equation .
( but such a de does not necessarily have to describe a physical theory . ) in the present paper , we examine scale invariant theories and find one direct route to power - law behaviour .
this formalism consists of transforming a scale invariant theory into its associated autonomous form@xcite and then determining the location of its fixed points .
these fixed points will lead inexorably to a background solution that exhibits power - law behaviour . by expanding the system about its fixed point ,
a linearized autonomous differential equation emerges , whose solution will exhibit _ different _ power - law behaviour .
this naturally leads to two independent power laws : one for the background fixed point and a second one for the linearized deviations from the fixed point . combining both results
, our formalism indicates that the complete general solution to the system will , over some limited region , have a frobenius - like power - series expansion .
the usual frobenius series is one of the form y(x ) = x^p _ i=0^a_i x^i . with @xmath0 a natural number , and is suitable for exploring linear odes in the region of regular singular points . in the present context ,
involving nonlinear des , the generalized frobenius - like series we encounter are generically of the form y(x ) = x^p _ i=0^a_i x^i .
the exponent in the prefactor of this frobenius - like series ( @xmath1 ) is governed by the exponent associated with the fixed - point power - law background behaviour , while the power series is given in terms of exponents ( @xmath2 ) that are integer multiples of those arising in the linearized problem ( @xmath3 ) .
this type of expansion is essentially a modification of that given by liapunov s expansion theorem for autonomous des@xcite , wherein we have now back - tracked to give the expansion in terms of the original scale - invariant system . as specific examples we apply this formalism to both newtonian and relativistic isothermal stars , where the relevant differential equations are equivalent to the exponential [ isothermal ] lane emden equation and its relativistic generalization .
the resultant frobenius - like power series generalize the approximate solutions found , for instance , by chandrasekhar@xcite and by harrison _ _ et .
al.__@xcite , and provides significantly greater analytic information concerning their structure .
this formalism is also applied to a more general star , where the equation of state need not be linear except in the core .
both newtonian and relativistic stars are verified to undergo damped oscillations in the total mass , pressure and radius as the central density goes to infinity .
these oscillations can accurately be described by a frobenius - like power series with complex exponents . the relevant exponents ( though not the numerical values of the coefficients ) drop out naturally from the linearized problem .
this power law solution matches the behaviour observed in numerical calculations to great accuracy .
in this section we lay down the basic foundations of the formalism .
first , we define `` scale invariant '' , `` equidimensional '' , and `` autonomous '' , and present the appropriate substitutions to go from one form to another .
a single @xmath4th - order differential equation is chosen as a model because of its simplicity , and then the formalism is extended to a system of @xmath5 coupled @xmath4th - order differential equations . after determining the existence of fixed points in the equation , the background solution is obtained .
the differential equation is then expanded about its fixed points to obtain a second power - series solution near the fixed point ( linearized solution ) . combining both results ,
a frobenius - like series solution is obtained in a neighborhood close enough to the fixed point .
let us define scale invariant differential equations in the usual manner , as those which remain unchanged under the substitutions @xmath6 and @xmath7 , where @xmath8 is a fixed but arbitrary parameter.@xcite equations that remain unchanged under the substitutions @xmath9 and @xmath10 , where @xmath11 is another fixed but arbitrary parameter@xcite , will be defined as equidimensional - in-@xmath12 .
thus equidimensional equations can always be viewed as scale invariant equations corresponding to the particular value @xmath13 for the exponent .
finally , autonomous equations will be defined as those equations which remain unchanged under the substitutions @xmath14 and @xmath10 , where @xmath15 is a third fixed but arbitrary parameter@xcite .
this last condition is that of translation invariance in the independent variable @xmath12 .
it is a standard result that it is always possible to transform scale invariant equations into equidimensional - in-@xmath12 equations through the following substitution@xcite [ eq:1 ] y(x ) = x^p w(x ) .
it is also a standard result that we can transform equidimensional - in-@xmath12 equations into autonomous equations by substituting@xcite [ eq:2 ] x = e^t z(t ) = w(x ) , where @xmath16 is the new variable .
let us first consider a @xmath4th - order scale invariant differential equation in a single dependent variable , [ scale ] f ( x , y(x ) , y(x ) , y(x ) , , ^n y(x)x^n ) = 0 , where the primes stand for differentiation with respect to @xmath12 .
we will write this as @xmath17 , but the presence of an appropriate number of derivatives should be inferred . now assuming this equation to be scale invariant ,
let s apply equation ( [ eq:1 ] ) to make equation ( [ scale ] ) equidimensional - in-@xmath12 .
the derivatives of @xmath18 become = ( x)^n = _ j=0^n n
x^p - j w^(n - j)(x ) .
note that the @xmath19th derivative of @xmath18 involves only the @xmath19th and lower - order derivatives of @xmath20 .
rewriting @xmath21 , we can define a new function @xmath22 as follows , @xmath23 since @xmath21 depends only on @xmath12 , @xmath20 and its derivatives .
the new differential equation @xmath24 is now equidimensional - in-@xmath12 .
let s now apply the substitution in equation ( [ eq:2 ] ) to transform it into an autonomous equation .
the derivatives of @xmath25 are [ eq:3 ] ^n x^n = ( e^-t t)^n = e^-n t _ j=0^n-1 ( t - j ) . as before , note here that the product in equation ( [ eq:3 ] ) involves an @xmath4th order term plus lower - order derivatives ( and no higher - order derivatives ) . rewriting @xmath22
we can define a new function @xmath26 via & & |f ( z(t ) , ( t ) ,
^n t^n z(t ) ) f(t ) = + & & ( e^t , z(t ) , e^-t(t ) , , e^-n t
_ j=0^n-1 ( t - j ) z(t ) ) , where the dots denote differentiation with respect to @xmath16 and @xmath27 is some arbitrary function of @xmath16 .
the function @xmath28 is now translation invariant , since the substitution @xmath29 as @xmath30 leaves the equation @xmath24 ( based on setting the rhs @xmath31 zero ) unchanged .
the only way this can be true is if we can factorize @xmath32 into an autonomous @xmath26 times some @xmath27 , yielding the lhs .
our original scale invariant equation @xmath17 , that had become equivalent to the equidimensional - in-@xmath12 equation @xmath33 , has now become equivalent to the autonomous equation @xmath34 .
we can of course also work backward and prove that for any autonomous - in-@xmath16 equation for the dependent variable @xmath35 the substitution @xmath36 ( that is , @xmath37 ) produces an equation that is equidimensional - in-@xmath12 with @xmath38 .
furthermore for any value of @xmath1 the substitution @xmath39 results in a differential equation that is scale invariant . however , this scale invariant equation need not be equal to the physical differential equation that describes the system we are concerned with . without some additional information beyond the mere existence of a power law , there is simply not much one can say about the physics . in order to obtain the background solution of the ode , we must look at the fixed points . let s assume that @xmath26 has a fixed point @xmath40 such that all derivatives of @xmath35 vanish at that fixed point , , |f ( z(t ) = z _ * , ^i z(t)t^i = 0 ) = 0 , where @xmath41 $ ] denotes any positive integer in that range .
hence , a solution to the differential equation @xmath42 is @xmath43 .
tracing back all the substitutions we made , we obtain [ eq:4 ] y(x ) = x^p z _ * ( x , y(x ) ) = 0 . from this equation ,
the power law behaviour is obvious .
however , not all scale invariant equations possess fixed points . a scale invariant differential equation , whose associated autonomous equation _ does not _ have any fixed points , _ will not _ possess power - law solutions . refer to [ appe:1 ] for an explicit counter - example . in light of this
, we can formulate the following theorem : any first - order differential equation that is scale invariant _ and _ whose associated autonomous equation possesses a fixed point will have a _ background _ solution in the form of a power law [ equation ( [ eq:4 ] ) ] with exponent given by the scale invariant condition .
let us linearize the autonomous equation by setting @xmath44 $ ] , where @xmath40 is the fixed point and @xmath45 is a small perturbation about @xmath40 .
@xmath34 reduces to |f(z(t ) ) = |f(z _ * ) + linear approximation , we must have |f_1(z_1(t ) ) = 0 . but
@xmath46 is an ode which is both autonomous and linear , and still retains its character as a @xmath4th - order differential equation .
autonomous plus linear is a very powerful constraint , and in fact implies a differential equation with constant coefficients .
that is , there is some polynomial @xmath47 such that the function @xmath48 satisfies |f_1(z_1(t ) ) p(t ) z_1(t ) .
the standard technique@xcite for solving differential equations of this type is to insert a trial solution of the form @xmath49 .
then the differential equation reduces to the simple @xmath4th - order polynomial p ( ) = 0 , which has at most @xmath4 distinct solutions . assuming for the time being that all roots are distinct
, the general solution to the linearized autonomous equation is z_1(t ) = _
i=1^n a_i ( _ i t ) . that is z(t ) = z _ * + _ i=1^n a_i ( _ i t ) + o(a^2 ) . in terms of the original variable @xmath18 [ theo:1 ] y(x ) = x^p .
near the fixed point , we now see the presence of multiple power laws .
the prefactor @xmath50 is governed directly by the scale invariant condition , while the subsidiary exponents arise from solving a polynomial equation based on the linearized approximation .
we are now in a position to formulate the following theorem : [ theo:5 ] any @xmath4th - order ordinary differential equation that is scale invariant _ and _ whose associated autonomous equation possesses a fixed point will have a linearized solution which is generically expressible in the form of a power law ( equation [ theo:1 ] ) in a neighborhood surrounding the fixed point .
we can also enunciate a general rule of thumb : because the exponent @xmath1 ultimately arises from algebraic manipulations involving the physical dimensionalities of the various quantities present in the ode it is likely to be a rational number .
in contrast the exponents @xmath51 , being solutions of a polynomial equation , will generically be irrational . in the exceptional case where the roots of the polynomial @xmath52 are not distinct , then there are additional technical complications .
let @xmath53 of the roots be distinct , and let @xmath54 with @xmath55 $ ] be the multiplicity of the @xmath0th distinct root .
then we have @xmath56 and z_1(t ) = _ i=1^m _ j=1^g_i-1 a_ij t^i-1 ( _ i t ) . that is z(t ) = z _ * + _ i=1^m _ j=1^g_i-1 a_ij t^i-1 ( _ i t ) + o(a^2 ) . in terms of the original variable @xmath18 [ theo:6 ] [ e : logs ]
y(x ) = x^p .
thus logarithmic deviations from power law behaviour , while not generic , are certainly relatively easy to get despite the lore based in statistical field theory and second - order phase transitions a logarithm is not necessarily the result of one - loop physics .
logarithms can arise from causes as mundane as a repeated root in a linearized autonomous ode .
going beyond the linearized approximation , in the vicinity of the fixed point , one can write the expansion more systematically as @xmath57 $ ] , where @xmath58 , in which case @xmath34 reduces to |f(z(t ) ) = |f(z _ * ) + represents a second - order functional derivative of @xmath59 around the fixed point , which will be some messy quadratic differential operator with constant coefficients .
this implies both |f_1(z_1(t ) ) = 0 , and |f_1(z_2(t ) ) = - |f_2(z_1(t),z_1(t ) ) .
that is z_2(t ) = - |f_1 ^ -1 .
but we already know that @xmath60 , so @xmath61 consists of pieces of the form @xmath62 t)$ ] .
indeed z_2(t ) = _ i=1^n _
j=1^n b(a_i , a_j ) ( [ _ i+_j ] t ) . with @xmath63 being some complicated function of the @xmath64 that is deducible from the precise form of @xmath65 and @xmath66 . collecting terms z(t ) = z _ * + _ i=1^n a_i ( _ i t ) + _ i=1^n _ j=1^n
b(a_i , a_j ) ( [ _ i+_j ] t ) + o(^3 ) . in terms of the original variable @xmath18 y(x ) = x^p .
the general pattern is now clear .
let us take @xmath67 and let @xmath68 .
then formally z(t ) = _ m ^n a(m ) ( m t ) .
this series may be asymptotic , in general we make no claims about convergence , though in specific cases it often does converge in an open region surrounding the fixed point . the coefficients @xmath69 are certainly not independent but are inter - related via the underlying differential equation .
this is basically liapunov s expansion theorem , as discussed for instance by lefschetz@xcite .
readers interested in additional mathematical rigour are directed to that reference . near the fixed point
our equation becomes , in terms of the original variables @xmath18 , [ theo:2 ] y(x ) = x^p _ m ^n a(m ) x^m .
this type of series is frobenius - like in the sense that the prefactor @xmath50 will be governed by the background solution , while the exponents of the power series will be given by the linearized problem .
hence , we can formulate the following theorem : any @xmath4th - order ordinary differential equation that is scale invariant _ and _ whose associated autonomous equation possesses a fixed point will generically have a formal solution in the form of a frobenius - like power series ( equation ( [ theo:2 ] ) ) in a neighborhood close to the fixed point .
as per the previous discussion , this form holds only if the roots of the linearized problem are distinct . for coincident roots one should expect the frobenius - like series above to be modified by various logarithmic terms . indeed based on liapunov s expansion theorem@xcite the coefficients
@xmath69 become polynomial in @xmath16 , and so polynomial in @xmath70 [ _ cf .
_ equation ( [ e : logs ] ) above ] .
the resulting series is too clumsy to be of immediate use and fortunately none of the specific examples we encounter exhibit this particular behaviour .
all of the previous analysis can be generalized to systems of @xmath5 coupled @xmath4th - order ordinary differential equations .
there are three possible routes : 1 .
develop a suitable `` vector '' and `` matrix '' notation .
2 . use the fact that a system of @xmath5 coupled @xmath4th - order differential equations can always be reduced to a system of @xmath71 coupled first - order differential equations@xcite , but you would still need to develop a suitable `` vector '' and `` matrix '' notation .
3 . use the fact that a system of @xmath5 coupled @xmath4th - order differential equations can generically be `` decoupled '' by being reduced to a single de of order @xmath72 in one of the dependent variables , though there seems to be no constructive algorithm for doing so.@xcite on balance , we have found it most useful to directly develop a suitable `` vector '' and `` matrix '' notation .
let us begin by considering the equation = 0 where @xmath73 $ ] is the column vector of odes = ( c f_1(x , y_1(x),y_1(x ) , , y_2(x),y_2(x ) , , y_i(x ) ,
, ^n y_i(x)x^n ) + f_2(x , y_1(x),y_1(x ) , , y_2(x),y_2(x ) , , y_i(x ) , , ^n y_i(x)x^n ) + + f_i(x , y_1(x),y_1(x ) , , y_2(x),y_2(x ) , , y_i(x ) , , ^n y_i(x)x^n ) + ) . when we say that this system is scale invariant , we mean that as @xmath74 and , where = and @xmath75 $ ] is a column vector of dependent variables , the equation @xmath73 = 0 $ ] remains invariant .
note that this means that we have chosen our dependent variables @xmath76 in some convenient manner to make their scale transformation properties both simple and independent of each other this will almost always occur automatically in systems of odes derived from an underlying physical problem .
previously , we also showed that any @xmath4th - order differential equation in a single dependent variable that is scale invariant can be put into its associated autonomous form by applying the substitutions described in equations ( [ eq:1 ] ) and ( [ eq:2 ] ) .
therefore , it carries over that we can make the same kind of substitution in each equation of the system . after doing so
, we can define a new column vector @xmath77 $ ] in the following manner : = ( c |f_1(z_1(t),(t ) , , z_2(t),(t ) ,
, z_i(t), ,^n z_i(t)t^n ) + + ) , where for each @xmath78 $ ] we have @xmath79 and furthermore @xmath80 . from the previous analysis we know that each equation in this new vector equation @xmath81=0 $ ] is now autonomous , hence , the whole system is autonomous . let s now assume that this autonomous system has a fixed point @xmath82 $ ] such that all derivatives of @xmath35 with respect to @xmath16 in @xmath81 $ ] vanish at the fixed point , where we mean = ( c z_1^ * + z_2^ * + + z_i^ * ) .
that is = ( c z_1(t ) + z_2(t ) + + z_i(t ) ) = = ( c z_1^ * + z_2^ * + + z_i^ * ) . is one specific solution to the system of odes . substituting back we obtain the following equations for @xmath75 $ ] : [ theo:3 ] ( c y_1(x ) + y_2(x ) + + y_i(x ) ) = ( c z_1^ * + z_2^ * + + z_i^ * ) = ( c z_1^ * x^p_1 + z_2^ * x^p_2 + + z_i^ * x^p_i ) . therefore , we can formulate the following theorem : [ thm:1 ] any system of @xmath5 @xmath4th - order differential equations that is scale invariant , _ and _ whose associated system of autonomous equations possesses a fixed point , will have solutions in the form of a collection of power laws ( equation [ theo:3 ] ) , with exponents given by the scale invariant condition .
if we linearize the @xmath4th order system , then for each @xmath83 we will obtain a linear differential equation that has the capability of mixing the various @xmath84 . specifically let z_i(t ) = z^*_i + z_1,i(t )
+ o(^2 ) , then |f_i(z_j(t ) ) = @xmath85 are linear differential operators that lead to a system of linear and autonomous differential equations _
j=1^n \{|f_1}_ij(z_1,j(t ) ) = 0 .
this implies that each of the @xmath85 is a constant coefficient polynomial in @xmath86 \{|f_1}_ij p_ij(t ) . inserting a trial solution of the form
@xmath87 now yields the vector equation _
j=1^n p_ij ( ) a_j = 0 .
there is a non - trivial solution if and only if the @xmath88 matrix @xmath89 is singular ( @xmath90 ) , in which case the @xmath64 correspond to the singular eigenvector .
since each element of the matrix @xmath89 is itself a polynomial of order @xmath4 , ( where @xmath4 is the order of the individual differential equations in the system ) , the determinant is a polynomial of order @xmath91 having up to @xmath92 separate roots .
let us assume these roots are distinct and call them @xmath93 with @xmath94 and denote the corresponding eigenvector by @xmath95 .
as long as the roots are distinct these eigenvectors , multiplied by the associated exponentials , will span the solution space and we can write the general solution to our autonomous system as z_1,i(t ) = _ = 1^n ( a_)_i ( _ t ) . in terms of our original equation , [ theo:4 ] y_i(x )
= x^p_i \ { z^*_i + _ = 1^n ( a_)_i x^ _
+ o(^2 ) } .
this now allows us to formulate the following theorem : any system of @xmath5 @xmath4th - order differential equations that is scale invariant _ and _ whose associated autonomous equation possesses a fixed point will generically have solutions in the form of a collection of power laws ( equation [ theo:4 ] ) in a neighborhood close to the fixed point .
the prefactors @xmath96 $ ] will be governed by the background solution , while the @xmath93 are given by the linearized problem . if the roots are not distinct then a much messier equation involving logarithms in @xmath12 can be constructed , but is beyond the scope of the applications we have in mind .
we can again go beyond the linearized approximation in the vicinity of the fixed point and write @xmath97 $ ] . then , & & |f_i(z_j(t ) ) = & & + ^2 + o(^3 ) , where , similarly to the preceding discussion for a single dependent variable , the quantity @xmath98 represents the second - order functional derivative of @xmath99 around the fixed point @xmath100 $ ] . using the same reasoning as before , we will obtain a solution of the form z_2,i(t ) = _
= 1^n _ = 1^n \{b(a_,a_)}_i ( [ _ + _ ] t ) .
collecting terms and substituting back to our original solutions of @xmath101 we obtain y_i(x ) = x^p_i , the general pattern is again clear [ theo:6 ] y_i(x ) = x^p_i _ m ^n \{a(m)}_i x^m , where as before , @xmath102 and @xmath103 .
we can now formulate the following theorem : any system of @xmath5 @xmath4th - order differential equations that is scale invariant _ and _ whose associated autonomous equation possesses fixed points @xmath104 will generically have a solution in the form of a frobenius - like power series ( equation [ theo:6 ] ) in a neighborhood close to the fixed point .
the prefactors @xmath105 will be governed by the background solution , while the exponents of the power series will be given by the linearized problem . as previously
, there is an explicit requirement that the exponents arising from the linearized problem be distinct , otherwise ( as per liapunov s expansion theorem ) messy expressions polynomial in @xmath106 will be encountered .
( see lefschetz@xcite for technical details . )
before we turn to several specific astrophysical examples to show how this formalism works in detail , let us summarize key points : * scale invariance does not automatically imply power - law behaviour ; the existence of one or more fixed points is an additional requirement .
an explicit example of a scale - invariant ode without any power - law solution is presented in [ appe:1 ] . *
if we perturb slightly away from the fixed points , there will no longer be an exact power - law behaviour , but we have something similar : generically there will be a frobenius - like expansion of the form y(x ) = x^p _ i=0^a_i x^i , where the exponent @xmath1 depends on the fixed point and the exponent @xmath3 depends on the linearized ode around the fixed point .
the coefficients @xmath107 are of course constrained by the original exact ode , and since it is @xmath4th - order we expect @xmath4 of these coefficients to be `` free '' ( that is , to depend on initial conditions ) with the other coefficients in principle being determinable from the first @xmath4 and the exact ode .
* we feel that this formalism , attributing as it does power - law behaviour to rather general features of differential equations , goes a long way towards explaining the apparent ubiquitous occurrence of power - laws in nature .
in the following section , we apply the general formalism developed above to a static isothermal newtonian star .
the relevant system of odes is scale invariant , and contains two fixed points , one corresponding to the center of the star , and the other corresponding to the point at infinity ( where the star develops an infinitely thin halo ) .
we formally solve the differential system associated with it and obtain power - series solutions for the total pressure , radius , compactness and mass of the star in terms of a frobenius - like series about either fixed point .
for each of these physical quantities , the radius of convergence of these series ( one for each fixed point ) overlap , so they constitute a complete power - series solution .
these results subsume and extend previous results found by chandrasekhar@xcite , and the classic results of lane and emden . in particular , these series are then used to discuss the behaviour of such stars near the onset of collapse .
the basic equations of ( static ) non - relativistic stellar structure@xcite are ( @xmath108 ) = - ( r ) ; m(r ) = _ 0^r 4 ^ 2(r ) .
let us now adopt the following particularly simple equation of state , which is appropriate , for instance , for an isothermal star : p(r ) = c_s^2
. then = - p(r)c_s^2 , [ mass ] m(r)r = 4p(r ) r^2c_s^2 , where @xmath109 stands for the speed of sound .
the relevant boundary conditions are @xmath110 and @xmath111 .
this system of differential equations is scale invariant under @xmath112 , @xmath113 and @xmath114 .
this means that from our previous analysis we have @xmath115 and @xmath116 . as stated previously
, we can make suitable transformations to convert this scale invariant system into an autonomous system .
for this purpose , let s substitute [ zeta:3 ] ( r)=2p(r ) r^2/c_s^4 , and [ chi:3 ] ( r)=m(r)/(2 c_s^2 r ) .
then , the system becomes [ zeta:1 ] r = 2 ( r ) ; [ chi:1 ] r
= ( r ) - ( r ) .
as expected this system is now equidimensional - in-@xmath117 .
define a new variable @xmath118 so @xmath119 , then = 2 ( t ) ; = ( t ) - ( t ) . as we can clearly see this system is now autonomous .
there is a fixed point at @xmath120 and @xmath121 .
for any initial condition this point is an attractor as @xmath122 .
the center of the star is at @xmath123 ( @xmath124 ) . for a regular star with finite @xmath125
we have @xmath126 and @xmath127 as @xmath128 ; which is the second fixed point .
so solving our special dimensionless differential equation is equivalent to looking for the unique integral curve that emerges from the fixed point at @xmath129 and terminates on the fixed point at @xmath130 . applying theorem [ thm:1 ]
we can easily check that a singular solution to the system is of the form p(r ) = c_1 r^-2 , and m(r ) = c_2 r^1 , where @xmath131 and @xmath132 are constants . solving for the constants we see that the @xmath130 fixed point corresponds to the power - law solution [ eq:5 ] m(r ) = 2 c_s^2 r , and [ eq:6 ] p(r ) = c_s^42r^2 , while the @xmath129 fixed point corresponds to the more prosaic @xmath133 . in order to obtain the linearized solution it is necessary to linearize about its critical points .
let us first make the substitution @xmath134 and @xmath135 . then
, = - 2 _ 1(t ) ; = _
1(t ) - _ 1(t ) .
close enough to the @xmath130 stagnation point , solutions are of the form = \ { ( t - t_i ) } .
the eigenvalues of this matrix are @xmath136 , and the approximate linearized solutions approach the critical point exponentially in @xmath16 as @xmath137 .
there is a second ( unstable ) stagnation point at @xmath129 a repeller .
if we linearize around @xmath129 we get = + 2 _ 1(t ) ; = _
1(t ) - _ 1(t ) .
the corresponding eigenvalues are @xmath138 , corresponding to regular @xmath139 behaviour for @xmath140 and @xmath141 at the center of the star , plus an otherwise possible but irregular and unphysical behaviour for @xmath141 that goes as @xmath142 .
this @xmath142 behaviour would correspond to a finite @xmath143 if such were possible .
based on theorem [ theo:5 ] and our general comments regarding frobenius - like series it is reasonable to assume the following expansion to solve the system [ eq:8 ] ( z ) = _
m=0^_n=0^b_mn z^m_1+n_2 .
here we have introduced the dimensionless variable z= r
= ( /c_s^2 ) r , with the particular numerical coefficients being chosen for future convenience . in this case , since the @xmath144 index is unphysical , it will be excluded from the analysis .
hence , we restrict attention to an expansion of the following form , ( z ) = _
n=0^b_n z^n_1 _ n=0^b_n z^2n .
in order to solve for the correct coefficients @xmath145 we need to plug the previous equation into our differential equations .
first , by uncoupling the system ( eliminating @xmath140 to obtain a single second - order differential equation for @xmath141 ) we obtain - 2(z)z^2 + 2(z)^2z^2 + 2(z)z = 0 .
the relevant boundary conditions may be inferred from the fact that the pressure ( and so also the density ) is finite at the origin : ( 0)=(0)=0 ; ( 0 ) = 13 ;
the fact that only integer exponents arise in our series solution for @xmath146 is ultimately due to the fact that the second - order ode above , though nonlinear , has what amounts to a regular fixed point at @xmath147 .
so standard results in terms of ordinary frobenius series suffice to develop an appropriate ansatz . now multiplying by @xmath148 , we can write [ eq:7 ] ( zz)^2 - zdz - 2 = - 2 ^2 - 2 zz , or in operator form , [ operator ] = - ^2 .
now we plug in our assumption of the expansion in equation ( [ eq:8 ] ) and we get a recursion relation of the form b_n = -_i=0^n b_i b_n - i . because @xmath149 we can rewrite this as [ e : recur1 ] b_n = -(n+1)(n-1)(2n+1 ) _
i=1^n-1 b_i b_n - i .
this recursion relation , together with the initial condition @xmath150 , derived from the fact that @xmath151 , completely specifies the power series about the @xmath129 critical point .
it is straightforward to compute these coefficients numerically , though there does not seem to be any simple explicit closed form for the general term .
for @xmath140 we use the relation z z = - .
so if ( z ) = _ n=0^a_n z^2n , then a_n = ( 2n+1 ) b_n . keeping the first 250 terms in this series , and comparing with an explicit @xmath152th order runge kutta integration , gives very high accuracy [ @xmath153 out to @xmath154 . in terms of the physical variables @xmath155 and @xmath156 we have m(r ) = 2 c_s^2 r ( r ) = 2 c_s^2 r _ n=1^b_n ( 4_0 r^2c_s^2)^n = 2 c_s^2 r _ n=1^b_n ( 4p_0 r^2c_s^4)^n , and p(r ) = ( z ) c_s^42 r^2 = c_s^42 r^2 _ n=1^a_n ( 4_0 r^2c_s^2)^n = c_s^42 r^2 _ n=1^a_n ( 4p_0 r^2c_s^4)^n
. these are taylor series for the solution to the differential equations that describe an isothermal star , with the coefficients being specified by the recursion relations determined above . by bounding the coefficients using the recursion relation
, one can show that the series definitely converges for @xmath157 and that it definitely diverges for @xmath158 .
numerical methods seem to indicate a radius of convergence of @xmath159 .
the @xmath160 and @xmath161 coefficients do not have a simple explicit analytic form , but computing them numerically from the recursion relation is trivial .
note that in more classical language this analysis is equivalent to finding a power series solution of the isothermal lane
emden equation .
see [ a : lane - emden ] for details
. turning to the other critical point , from equation ( [ eq:7 ] ) we see that @xmath162 . now making a shift of variables appropriate to the @xmath130 fixed point , @xmath163 , where @xmath164 is small , we obtain [ eq:9 ] ( zz zdz + zdz + 2 ) _ 1 = - ( 2 + zz ) _
1 ^ 2 , with @xmath165 .
let s assume an expansion for @xmath164 of the form given in equation ( [ eq:8 ] ) with the substitution @xmath166 .
the real parts of @xmath167 must be negative so that the expansion has the right limit at infinity , where @xmath168 . by construction
we know that @xmath169 . in order for @xmath141 to be real ,
the matrix of coefficients @xmath170 must be hermitian .
now let s insert this ansatz into our shifted differential
equation [ equation ( [ eq:9 ] ) ] . for convenience define @xmath171 , since then z z z^m_1+n_2 = z z z^= z^ , and so equation ( [ eq:9 ] ) becomes ( ^2 + + 2 ) b_mn = -(+2 ) _ i=0^m _ j=0^n b_ij b_m - i;n - j .
this implies the recursion relation [ e : recur2 ] b_mn = -(+2 ) ( ^2 + + 2 ) _ i=0^m _
j=0^n b_ij b_m - i;n - j .
since @xmath172 , the right hand side does not involve @xmath170 itself .
thus @xmath170 can be calculated once the lower - order @xmath173 are known .
in particular once @xmath174 is specified , all other coefficients can be calculated recursively .
unfortunately the only known way of calculating @xmath174 is by performing a numerical integration ( _ e.g. _ , runge
kutta ) out to large radius and then numerically fitting the data . summing up ,
we obtain ( z ) = 1 + _
m=0^_n=0^b_mn z^m_1+n_2 .
a similar power series will exist for @xmath175 .
indeed suppose ( z ) = 1 + _
m=0^_n=0^a_mn z^m_1+n_2 . then a_mn = ( + 1 ) b_mn = ( m_1+n_2 + 1 ) b_mn . in terms of the physical variables
@xmath155 and @xmath156 m(r ) = 2 c_s^2 r ( r ) = 2 c_s^2 r \ { 1 + _ m=0^_n=0^b_mn ( 4p_0 r^2c_s^4)^(m_1+n_2 ) } , and p(r ) = ( z ) c_s^42 r^2 = c_s^42 r^2 \ { 1 + _ m=0^_n=0^a_mn ( 4p_0 r^2c_s^4)^(m_1+n_2 ) } .
these are frobenius - like power series with fractional complex exponents for the isothermal star [ recall @xmath176 , where the coefficients @xmath177 and @xmath170 are specified by the recursion relations determined above .
these expansions are considerably more general than the discussion presented in chandrasekhar @xcite , and the related results for the relativistic case implicit in harrison _
_ et al.__@xcite .
those results are tantamount to just keeping the first sub - asymptotic term , which is also contained in our solution , namely m(r ) 2 c_s^2 r \ { 1 + 2 } , and p(r ) \ { 1 + 2 } . using analytic techniques , the series can be shown to converge for @xmath178 .
this is a weak bound but at least establishes a non - zero radius of convergence .
numerically fitting @xmath179 by matching to a 4th - order runge
kutta integration implies guaranteed convergence for @xmath180 .
direct numerical evidence suggests that the power series is reasonably well behaved down to @xmath181 with the first three terms providing @xmath182 accuracy . by numerically applying the ratio test , we find that this power series seems to converge for @xmath183 . indeed if we keep the first @xmath184 `` levels '' ( @xmath185 ) then the power series reproduces a runge - kutta integration to within 6 significant figures for @xmath186 . the combination of both power series
, a taylor series around @xmath123 and a fractional power series around @xmath187 , provides a solution everywhere to arbitrarily good accuracy , since the two expansions have overlapping radii of convergence . instead of expanding @xmath141 and @xmath140 in terms of radius @xmath117 , we now want to develop an expansion in terms of pressure @xmath1
. this will be useful if we want to generalize beyond a simple linear equation of state .
we can rewrite the equation of hydrostatic equilibrium in the following manner : = 2 p(r ) ( r)r . now , by applying the chain rule to @xmath188 , differentiating equation ( [ zeta:1 ] ) with respect to @xmath117 , and plugging in our new expression for the equation of hydrostatic equilibrium , we obtain [ zeta:2 ] p = - . the same can be done to equation ( [ chi:1 ] ) to obtain [ chi:2 ] p = 1 2 ( 1- ) .
this new system of differential equations in terms of the pressure still has a stagnation point at @xmath189 = @xmath130 .
this is the same stagnation point that we encountered earlier .
this stagnation point is again an attractor , which means that for any initial condition as @xmath190 almost all solutions of the system approach this solution .
note that the second stagnation point ( the repeller ) at @xmath129 is still there but somewhat disguised , since the system in terms of @xmath1 is highly singular there .
we will not need the second stagnation point in the following discussion .
linearizing about the @xmath130 stagnation point we obtain p = _ 1 ; p = 1 2 ( _ 1-_1 ) .
close to the @xmath130 stagnation point , the solutions to the system are of the form = \ { ( p / p_i ) } .
the eigenvalues of this matrix are @xmath191 .
note that this is simply @xmath192 , where @xmath193 are the exponents as encountered when working with the variable @xmath117 .
it is essential to decouple the system of differential equations in order to obtain a power series expression for the mass and the radius in terms of the pressure that is , we want to replace the first - order system by a single second - order equation . in order to decouple equations ( [ zeta:2 ] ) and ( [ chi:2 ] ) , we solve for @xmath140 in terms of @xmath141 from equation ( [ zeta:2 ] ) : = ( 1- 2 p p ) .
now , if we insert this expression back into the first differential equation , we obtain -1 + + pp = 2 \ { pp - ( p ) ^2 - p p p p } .
we can see from the analysis in the previous section that @xmath194 .
it is convenient to write @xmath195 because we now have @xmath196 . after this shift in variables ,
our decoupled differential equation becomes [ shifted:1 ] \ { 1 - pp + 2 p p p p } = \ { pp - p p p p } ^2 .
now , in line with theorem [ theo:5 ] and our general discussion regarding frobenius - like series , we assume an expansion of the form ( p ) = _ m=0^_n=0^d_mn ( p p_0)^m_1+n_2 .
previously we used the dimensionless variable @xmath197 to create a power series expansion about the stagnation point @xmath130 for @xmath141 and @xmath140 .
this time we find it more convenient to expand in powers of @xmath198 .
it is interesting to note that our eigenvalues , @xmath199 , make physical sense because we need positive real parts in order to satisfy the boundary condition @xmath200 .
for @xmath201 we have @xmath202 and the hermiticity condition discussed previously still holds .
again , let us for simplicity define @xmath203 . we then have p p ( pp_0)^m_1+n_2 = p p ( pp_0)^= ( pp_0)^. and therefore equation ( [ shifted:1 ] ) implies d_mn = ( 1- ) _ i=0^m _ j=0^n d_ij d_m - i;n - j . note
that by construction @xmath204 , so once @xmath205 is specified all other coefficients can be calculated recursively .
now this recursion relation holds for the coefficients in @xmath201 only . for @xmath141 itself we have ( p / p_0 ) = 1 + _ m=0^_n=0^d_mn ( p p_0)^m_1+n_2 .
a similar power series will exist for @xmath206 .
let us assume the following expansion ( p/ p_0)= _ m=0^_n=0^c_mn ( p p_0 ) ^. solving for @xmath140 , from equation ( [ zeta:2 ] ) we obtain = - p ^2p , which , upon shifting to the fixed point , is equal to = - p p ( 2+^2 ) . comparing exponents we obtain c_mn = d_mn - ( 2 d_mn + _ i=0^m _ j=0^n d_ij d_m - i;n - j ) . using the recursion relation for the @xmath207 c_mn = d_mn - ( 2 d_mn + ( 2 ^2 - + 1 ) ( 1- ) d_mn ) = -21- d_mn .
note that by construction @xmath208 .
now this relation holds for the coefficients in @xmath209 only . for @xmath140 itself we have ( p / p_0 ) = 1 + _ m=0^_n=0^c_mn ( p p_0)^m_1+n_2 .
we could of course unwrap the definitions of @xmath140 and @xmath141 to obtain frobenius - like power series for the physical observables @xmath210 and @xmath211 .
let us now consider a more realistic equation of state : ( p ) = \ { l@:l _ c= p_c / c_s^2 & p < p_c + p / c_s^2 & p > p_c . , where @xmath213 and @xmath214 stand for the pressure and density at the surface of the core .
most of the analysis carried until now still holds , at least for the central region of the star .
but we now chose to replace the previous equation of state [ @xmath215 , which was linear out to arbitrarily low density ] with a new equation that is somewhat more realistic .
we have a dense core described by a linear equation of state surrounded by an envelope of constant density in which the pressure drops rapidly to zero .
since in the previous section we have established the existence of a power series for @xmath216 , we can formally solve equation ( [ zeta:3 ] ) for the radius of the core itself by setting @xmath217 ( at this stage making no comment about the envelope ) .
hence , r_(p_c , p_0 ) = = c_s^2 .
this last equation clearly shows the oscillatory behaviour of the core radius as a function of central pressure , due to the imaginary exponents in the power series .
if we now look at equation ( [ chi:3 ] ) and solve for the mass of the core in terms of the compactness and radius we obtain @xmath218 as we can see from this equation the mass of the core will also oscillate as a function of central pressure due to the imaginary part of the exponent of the power series . indeed there is a related power series such that m_(p_c , p_0 ) = 2c_s^4 \ { 1 + _ m=0^_n=0^ ( p_c p_0 ) ^ } , with the hermitian coefficients @xmath219 being in principle calculable in terms of the @xmath220 and @xmath207 .
the above expression gives us the mass of the core of the star , not of the star itself . in order to calculate the total mass of the star we need to first calculate the mass of the envelope of the star : m_(r)= _
r_^r4 r^2 _ c. hence ,
m_(r ) = 43 _ c ( r^3-r_^3 ) + m_. we calculate @xmath221 from our equation of state and equation of hydrostatic equilibrium : p_(r ) = p_c - _ r_^r _ c m_(r ) , where we have used the fact that @xmath222 is constant throughout the envelope . integrating p_(r ) =
p _ + ( _ c m _ - 4 3 _ c^2 r_^3 ) ( 1 r - 1 r _ ) + 2 3 _ c^2 ( r_^2 - r^2 ) .
we can solve for @xmath221 by solving for @xmath223 .
the resulting cubic equation is p_(r = r_)=0=c_1 r_^3 + c_2 r _ - c_3 , where we have chosen the constants @xmath131 , @xmath132 and @xmath224 to be c_1 = 2 3 _ c^2 , c_2 = _ c m _
r _ - 2 _ c^2 r_^2 - p _
c m_- 4 3 _ c^2 r_^3
. this cubic equation can be explicitly [ if tediously ] solved for @xmath221 as a function of @xmath214 , @xmath225 , and @xmath226 .
suffice it to say that we now have implicitly solved for the total mass of the star ( @xmath227 ) , where we have taken @xmath221 as a parameter ( which can ultimately be expressed , via @xmath225 , and @xmath226 , as an oscillating function of @xmath213 and @xmath125 ) . that is m_(p_c , p_0)= 43 _ 0 ( r_(p_c , p_0)^3 -r_(p_c , p_0)^3 ) + m_(p_c , p_0 ) .
ultimately , since the individual terms above exhibit oscillatory behaviour , there will be some frobenius - like power series such that m_(p_c , p_0 ) = m_(p_c , p_0 ) \ { 1 + _ m=0^ _
n=0^ ( p_c p_0 ) ^ } . from this expression for the total mass of the star we can clearly see the oscillatory behaviour mentioned earlier .
note that this behaviour is much more extensive than the scaling results discussed in chandrasekhar @xcite [ see also harrison _ _ et al__@xcite ] .
their result is equivalent to keeping only the first nontrivial term in the power series .
that is : m_(p_c , p_0 ) m_(p_c , p_0 ) \ { 1 + 2 } .
we now see that the scaling behaviour they discussed is only the first term in an infinite series .
it may be objected that our current analysis is only valid for the special case of a linear equation of state with a constant density envelope . however , from the way we have set up the general theory it is clear that this restriction is not particularly serious .
first , any reasonable equation of state will saturate to a linear at high enough densities , so our results can be applied without modification to the core of any reasonable stellar configuration as the central pressure @xmath228 .
furthermore , suppose one has some generic equation of state
. then the exact system will not be scale invariant .
nevertheless there will be a special critical configuration in which the central pressure is infinite , and since by hypothesis the equation of state becomes linear at sufficiently high pressure , the critical solution will exhibit approximate scale invariance in the high - pressure region and this will correspond to a fixed point in the autonomous system .
linearization around the fixed point will generate some critical exponents @xmath229 , which depend only on the high - pressure limit of the equation of state .
but deviations from the scale invariant critical configuration in the core of the star will now feed into the exact scale non - invariant equations , still leading to the total mass being given in terms of some frobenius - like series .
ultimately this frobenius - like power series behaviour is generic to many systems of differential equations and intrinsically does not have anything to do with gravity _ per se_.
in order to treat relativistic stars , we need to consider the relativistic equation of hydrostatic equilibrium ( tov equation ) [ tov ] pr = - ( + pc^2 ) ( m + 4 r^3 p / c^2 ) .
we again have a system of two first - order differential equations , namely equations ( [ tov ] ) and the mass equation ( [ mass ] ) , which must be supplemented by an algebraic equation of state . note that in the tov equation @xmath15 is the speed of light , and that as @xmath230 we formally recover newtonian physics .
we will consider an equation of state identical to that used in the previous chapter , _
i.e. _ @xmath231 , where @xmath109 stands for the speed of sound in the fluid . making the appropriate substitutions we obtain the closed system s \ { l = -p ( 1c_s^2 + 1c^2 ) ( m + 4 r^3 p/ c^2 ) , + + = 4 r^2 pc_s^2 . .
note that ( as in the newtonian case ) this system @xmath232 is invariant under the substitutions @xmath233 , @xmath113 and @xmath234 . revisiting our theoretical analysis
we also discover that , @xmath235 and @xmath236 , again as in the newtonian case .
it is possible to systematically transform the system @xmath232 into an equidimensional - in-@xmath117 system @xmath237 , and subsequently into an autonomous system @xmath238 .
let us first substitute @xmath239 and @xmath240 .
let us also define @xmath118 so @xmath119 .
then , the system becomes s_1 \
{ l = 2- 2 ( 1+^2 ) ( + ^2 ) , + + = - , . where @xmath241 .
as we can clearly see @xmath237 is now autonomous , and therefore , translation invariant .
there are two fixed points , namely @xmath242 and @xmath243 , where @xmath244/@xmath245 .
note that @xmath246 corresponds to the newtonian limit . as we found previously ,
the first fixed point is an attractor as @xmath122 , which corresponds to @xmath247 behaviour for the pressure .
the second fixed point at @xmath129 is a repeller , as we also found in the newtonian case .
this fixed point corresponds to regular @xmath139 behaviour for @xmath140 and @xmath141 at the center of the star , plus an irregular and unphysical behaviour for @xmath141 that goes as @xmath142 .
the center of the star is at @xmath123 ( @xmath124 ) . for a regular star with finite @xmath125
we have @xmath126 and @xmath127 as @xmath128 .
so solving @xmath237 is equivalent to looking for the unique integral curve that emerges from @xmath129 and terminates at @xmath248 .
since system @xmath232 is scale invariant and its associated autonomous system @xmath237 has fixed points , we can apply theorem [ thm:1 ] and easily check that the background solution to @xmath232 is p(r ) = c_1 r^-2 , and m(r ) = c_2 r^1 , where @xmath131 and @xmath132 are constants . inserting back into @xmath232 and solving for the constants we see that the @xmath248 fixed point corresponds to the power - law solution [ eq:5b ] m(r ) = 2c_s^2 r , and [ eq:6b ] p(r ) = c_s^4 r^2 .
note that if we take the limit @xmath249 , we recover the exact solutions obtained in the previous chapter for the newtonian case .
also note that the background solution for the @xmath129 fixed point corresponds to the trivial @xmath133 .
we suspect that , as in the newtonian case , the pressure and the mass will have a series expansion in the form of an iterated power law . in order to determine the exponents of the power , let s analyze each fixed point in more detail .
linearizing about the @xmath248 fixed point we obtain the following system s^linear_1 \ { l = a + b , + + = - , . where a = -2 ^2 ; b = -2 ( 1 + 5 ^2 ) , and @xmath209 and @xmath201 are small perturbations
. therefore , close to the stagnation point , solutions are of the form = \ { ( t - t_i ) } , the exponents of the power law are given by the eigenvalues of this matrix , which are _
= -1 - 3 ^2 , and the approximate linearized solutions approach the critical point exponentially in @xmath16 as @xmath250 .
these are the exact same power exponents found by harrison _ _ et .
at.__@xcite note that if we take the newtonian limit @xmath249 , , @xmath251 and @xmath252 , we recover the eigenvalues for the newtonian case .
let us now analyze the @xmath129 fixed point .
making the substitution @xmath253 and @xmath254 , where @xmath201 and @xmath209 are to be viewed as small perturbations , we obtain the following system s^linear_2 \
{ l = 2 , + + = - . .
close to the stagnation point , solutions to this system have the form = \ { ( t - t_i ) } .
this is the second unstable stagnation point , which was mentioned in the previous section .
this fixed point is a repeller with eigenvalues @xmath255 .
note that these are the same eigenvalues found for the fixed point @xmath129 in the newtonian analysis .
we have now determined the exponents of the critical power - law behaviour for both fixed points . based on theorem [ theo:5 ] and our general argument for the existence of frobenius - like series let us now assume an expansion for @xmath141 and @xmath140 of the form [ series ] ( r ) = _
m=0^_n=0^b_mn r^m_1+n_2 ; ( r ) = _ m=0^_n=0^a_mn r^m_1+n_2 .
we now need to calculate the actual coefficients @xmath145 and @xmath256 of the series . in order to find them
, we need to first decouple the system @xmath237 . solving for @xmath140 in the @xmath257 equation in @xmath237 and inserting this into the @xmath258 equation we obtain [ decoupled ] r r rr- 2- rr = - 2 ( 1+^2 ) ( + rr ) .
it is now convenient to use the dimensionless variable @xmath259 . rearranging the above ,
we obtain & & + & & = ^2 + & & + 2 ^ 2(3 + ^2 ) ( z z ) ^2 .
note here that @xmath260 and that if we take @xmath252 we recover the newtonian result @xmath261 .
we now have a second order decoupled differential equation for @xmath141 . before solving for the coefficients of the series , we remind ourselves that for this case the @xmath142 behaviour is unphysical
hence , let us suppress all occurrences of @xmath262 . and restrict our analysis to @xmath263 . instead of equation ( [ series ] )
we now have [ e : series2 ] & & ( z ) = _ n=1^b_n z^2n = _ n=1^b_n ( 4p_0 r^2c_s^4)^n ; + [ e : series22 ] & & ( z ) = _
n=1^a_n z^2n = _ n=1^a_n ( 4p_0 r^2c_s^4)^n . plugging this ansatz into our decoupled differential equation , we obtain the following recursion relation for the coefficients @xmath145 b_n & = & - ( 1 + 6 ^ 2+^4 ) + ( 1 + 5 ^ 2 + 2 ^ 4)n - 4 ^2 n^2 _ i=1^n-1 b_i b_n - i + & & - 4 ^ 2(3+^2 ) _
i=1^n-1 i ( n - i ) b_i
b_n - i . with this recursion relation
, we have an expression for @xmath146 equation ( [ e : series2 ] ) in the form of a power series close to the fixed point @xmath129 .
a similar power series will exist for @xmath140 . using the @xmath264 equation from @xmath237
, we observe that ( as in the newtonian case ) a_n = ( 2n+1 ) b_n . in order to completely specify the power series ,
we need to have boundary conditions .
the boundary conditions to begin the recursion relation are @xmath149 and @xmath150 , which are the same conditions used previously in the newtonian analysis of the system .
choosing @xmath150 is equivalent to the specific choice of @xmath25 made above .
hence , inserting these conditions into the @xmath160 equation we obtain @xmath265 and @xmath266 .
we have now a complete expression for @xmath141 and @xmath140 close to the @xmath129 stagnation point given by equation ( [ e : series2 ] ) and ( [ e : series22 ] ) .
the radius of convergence @xmath267 of the @xmath141 series will be the same as that for the @xmath140 series , and is now a function of the parameter @xmath268 .
we performed numerical calculations and discovered that there is a finite radius of convergence for almost all values of @xmath268 . in order to better understand the convergence of the series ,
let us define two regions : region i , ranging from @xmath269 and region ii , defined from @xmath270 .
the radius of convergence is well - behaved as a function of @xmath268 in region i , allowing us to recover the newtonian upper bound at @xmath271 ( @xmath272 . )
our results also showed that the radius of convergence strictly increases monotonically until it reaches @xmath273 at which @xmath274 . in region
ii , the global behaviour of @xmath268 is given by a power law ( approximately @xmath275 ) , decreasing from @xmath276 at @xmath273 to @xmath277 at @xmath278 .
the local behaviour as a function of @xmath268 in region ii , however , is more complex , since there are values of @xmath268 for which the radius of convergence drops to zero .
the initial value at which our series diverges is @xmath279 , and after that it will diverge at @xmath280 , where @xmath4 is an integer ( @xmath281 ) and @xmath282 .
it seems possible to approximate the behaviour of the radius of convergence in region ii with a damped oscillating function , _
e.g. _ @xmath283 . in order to determine the coefficients
@xmath170 and @xmath177 for the power series about the fixed point @xmath248 we need to make a shift to @xmath285 in our decoupled differential equation ( [ decoupled ] ) . by doing so , equation ( [ decoupled ] ) can be simplified and put in operator form to obtain @xmath286\right ) z{\d \over\d z } - \left(1 - 4\alpha\beta^2\right ) z { \d \over\d z } z{\d \over\d z } - 2 \right]\hat\chi \nonumber\\ & & \qquad\qquad = \left [ 2 \left(1 + 6\beta^2 + \beta^4\right ) + \left(1 + 5\beta^2 + 2\beta^4\right ) z{\d \over\d z } - 2\beta^2 z{\d \over\d z } z{\d \over\d z } \right]\hat\chi^2 \nonumber\\ & & \qquad\qquad\qquad + 2\beta^2\left(3 + \beta^2\right ) \left(z { \d \over\d z}\hat\chi\right)^2 . \end{aligned}\ ] ] again , we have a second order decoupled differential equation for @xmath201 .
note that for our series expansion we have the expression given in equation ( [ series ] ) , where we must take @xmath287 . plugging this ansatz into our decoupled differential equation
, we obtain a recursion relation for the coefficients @xmath170 @xmath288\right)\delta + 2 } } \;\ ; \sum_{j=0}^{m}\sum_{i=0}^{n } b_{ij } b_{m - i;n - j } \nonumber\\ & & -{2\beta^2\left(3+\beta^2\right ) \over { \left(1 - 4\beta^2\alpha\right)\delta^2 - \left(1 - 2\alpha \left[1 + 5\beta^2 + 2\beta^4\right]\right)\delta + 2 } } \nonumber\\ & & \qquad\qquad \times \sum_{j=0}^{m}\sum_{i=0}^{n } ij \left(m - j\right)\left(n - i\right ) b_{ij } b_{m - i;n - j } , \end{aligned}\ ] ] where @xmath289 . with this recursion relation
, we again have a complete expression for @xmath141 in the form of a power series close to the fixed point @xmath248 . @xmath140
will also have a power series expansion , with coefficients given ( as in the newtonian case ) by a_mn = ( + 1 ) b_mn .
once more , we need boundary conditions to begin the recursion relation . we know analytically that @xmath172 and @xmath290 .
if we knew @xmath291 [ or equivalently @xmath292 then we could recursively calculate the entire series . as in the newtonian case
, there is no analytic theory for these coefficients . though we do not have an analytic way of specifying @xmath293 or @xmath294
, we do have the powerful result that @xmath140 and @xmath141 are [ close to the fixed point @xmath248 ] given by power series of the form ( r ) = + _ m=0^_n=0^b_mn ( 4p_0 rc_s^2)^m_1+n_2 , and ( r ) = + _ m=0^_n=0^a_mn ( 4 p_0 rc_s^2)^m_1+n_2 . in terms of the dimensional physical variables @xmath156 and @xmath155 this can explicitly be put in frobenius - like form m(r ) = 2 c_s^2 r , and p(r ) = c_s^44r^2 .
though we will not go into any of the details , it is clear from the existence of the power series @xmath156 , that the series can be `` reverted '' to provide a power series for @xmath210 .
the key points are that @xmath156 is monotonic , so the inverse @xmath210 exists . then from the power series above we deduce the existence of a reverted frobenius - like power series r(p , p_0 ) = c_s^2 . from this
we deduce that for a cutoff equation of state ( linear in the core , constant density in the envelope ) even in the relativistic case the core mass and core radius will exhibit damped oscillations as a function of central pressure @xmath125 .
( p_c , p_0 ) = r(p = p_c , p_0 ) = c_s^2 .
similarly the total mass and total radius will exhibit damped oscillations .
it should also be clear that the details of the equation of state at low density do not matter .
as long as the equation of state at high densities is linear , then the analysis above implies that as the central pressure goes to infinity the core radius and core mass will undergo damped oscillations as a function of central pressure .
once the core undergoes these oscillations , the envelope serves only to communicate this internal behaviour out to the stellar surface .
thus the total mass and total radius will similarly undergo damped oscillations .
these damped oscillations will involve an infinite power series in terms of @xmath296 , which will contain the first - order term discussed in harrison _ _ et al.__@xcite , plus an infinite collection of higher - order terms .
we have presented a unified formalism for calculating the solution to scale invariant theories in the form of power laws , and deviations from pure power - law behaviour .
we have proved that some of the solutions to a scale - invariant differential system , whose associated autonomous equation possesses fixed points , will exhibit pure power - law behaviour .
we have also proved that after linearizing about these fixed points , the solution to the scale - invariant linearized system can also be written in the form of an independent collection of power laws .
generically , these collections of power - law solutions can be combined to develop a frobenius - like series in a neighborhood surrounding the fixed point .
we applied these ideas to several problems taken from astrophysics , such as static isothermal stars both in the newtonian and tov frameworks .
we showed , both numerically and analytically , that in all the static cases the solution to the scale - invariant differential system associated with each example exhibits power - law behaviour as the star approaches the point of collapse .
we also showed analytically that the damped oscillations in the pressure , mass and radius of a star , as the central density goes to infinity , can be represented as a frobenius - like power series , which is easily obtained by applying our formalism . more generally , it would be advantageous to consider truly dynamical systems involving pdes in both space and time .
work along these lines is progressing .
this research was partially supported by the us doe .
nicolas yunes wishes to acknowledge the support of washington university through the mesmer and washington university scholarship , as well as support from the washington university gravitation group .
we gratefully thank the washington university gravitation group for useful comments and support .
caution must be exercised since not all scale invariant equations possess fixed points .
a scale invariant differential equation , whose associated autonomous equation _ does not _ have any fixed points , will _ not _ possess power - law solutions . as an example
, let s analyze what happens to the following relatively simple differential equation : [ countereg ] f(x , y(x ) ) - p y(x)x - f x^p-1 = 0 , where @xmath297 is a non - zero constant . this
is easily verified to be a scale - invariant equation and the associated equidimensional - in-@xmath12 equation is f(x , w(x ) ) x^p ( w(x)x - fx ) = 0 .
this factorizes to give f(t ) and it does not possess a fixed point since we are assuming that @xmath298 . the general solution for this differential equation is easily verified to be y(x ) = x^p ( f + a ) .
note that the solution to our example will be a power law if and only if @xmath299 , regardless of the fact that our example is explicitly scale invariant for all values of @xmath297 . in conclusion ,
scale invariance by itself does not guarantee that the solution to a scale invariant differential equation will exhibit power law behaviour .
more complicated examples of this behaviour can easily be constructed , but this simple equation is already enough to make the point that scale invariance does not necessarily lead to power - law solutions . however , if we impose _ both _ scale invariance and the constraint that the associated autonomous equation has fixed points , then the solution will certainly exhibit power law behaviour .
in the theory of autonomous differential equations the existence of a fixed point is one of the simplest things one could wish for ; the next most complicated structure one might encounter is a limit cycle .
suppose we start with a scale invariant differential equation and that the associated autonomous equation has a limit cycle rather than a fixed point .
this means there is a special solution @xmath300 with a period @xmath301 such that z_*(t + n t ) = z_*(t ) n z. in terms of the equidimensional variables this implies w _ * ( ( n t ) x ) = w_*(x ) .
that is , a limit cycle in the variable @xmath16 becomes a discrete self - similarity in the variable @xmath12 .
similarly in terms of the original variable @xmath18 we obtain discrete power - law behaviour y(x ) = x^p w_*(x ) ; nz : y ( ( n t ) x ) = ( p n t ) y(x ) .
this is not the classic power - law behaviour discussed previously , but is , in a sense , the next least complicated thing . as an example , let s consider the second - order differential equation f(x , y(x ) ) x^2 y(x ) - ( 2p-1 ) x y(x ) + p^2 y + ^2 y(x ) = 0 this is easily verified to be scale invariant with index @xmath1 .
the associated equidimensional - in-@xmath12 equation is f(x , w(x ) ) x^2 w(x ) + x w(x ) + ^2 w(x ) = 0 .
the associated autonomous equation is |f(z(t ) ) z(t ) + ^2 z(t ) = 0 .
now there is a trivial fixed point at @xmath147 but the general solution to the autonomous equation is z(t ) = a(t ) + b(t ) . in terms of the equidimensional
variables w(x ) = .
finally , in terms of @xmath18 y(x ) = x^p .
this now is an explicit example of a discrete power law .
the solution is not scale invariant under arbitrary rescalings @xmath302 , but it is invariant under the specific rescaling @xmath303 . note the implication : discrete self - similarity and discrete power laws are not as peculiar as one might at first imagine ; instead they are simply the next most complicated thing ( after simple power - law behaviour ) that can happen in scale invariant systems .
( viewed in this light the occurrence of discrete self - similarity in choptuik s critical solution @xcite should not at all be considered surprising . )
the traditional [ isothermal ] lane emden equation is ( z^2 ( z)z ) = ( -(z ) ) .
( 0 ) = 0 ; .z|_0 = 0 .
this second - oder de is completely equivalent to the first - order system we derived for a newtonian relativistic star . indeed
( z ) = - , and the series we previously determined for @xmath175 in the newtonian case then carry over into series for the lane - emden function @xmath304 .
it is then straightforward to compute ( z ) = 16z^2-z^4+z^6 -z^8+z^10 + o(z^12 ) .
you can also derive this in a more direct way by putting a trial power series directly into the lane emden de and equating coefficients .
perhaps more surprising is the implication that at large @xmath25 the lane
emden function should have an expansion of the form ( z ) = + _ m=0^_n=0^q_mn z^(m_1+n_2 ) .
[ recall @xmath305 . ] the coefficients @xmath306 are in turn determined by the recursion relations for @xmath177 and @xmath170 previously discussed .
an alternative more direct route is to write @xmath307 + q(z),\ ] ] and substitute into the lane
emden equation yielding \ { z z zz + zz } q(z ) = 2 \ { - 1 }
. linearizing around the obvious solution @xmath308 recovers the critical indices @xmath305 , while inserting a trial series consisting of integer powers of these critical exponents will determine all the higher coefficients @xmath306 in terms of @xmath309 .
of course , as is usual , there is no analytic theory for this first coefficient since it corresponds to finding the unique boundary condition at infinity that leads to a regular star at the center @xmath309 has to be determined numerically from an outward integration of the lane
emden equation starting from the origin where we do know the physical boundary conditions .
when it comes to writing down explicit recursion relations for the coefficients , it is more convenient to work with the compactness @xmath146 in terms of which the [ isothermal ] lane emden equation is equivalent to equation ( [ operator ] ) and explicit recursion relations are given in equations ( [ e : recur1 ] ) and ( [ e : recur2 ] ) .
m. w. choptuik , `` universality and scaling in gravitational collapse of a massless scalar field , '' phys .
* 70 * ( 1993 ) 9
. + m. w. choptuik , t. chmaj and p. bizon , `` critical behaviour in gravitational collapse of a yang - mills field , '' phys .
* 77 * ( 1996 ) 424 [ arxiv : gr - qc/9603051 ] . c. gundlach ,
`` critical phenomena in gravitational collapse , '' adv .
* 2 * ( 1998 ) 424 [ arxiv : gr - qc/9603051 ] .
+ c. gundlach , `` critical phenomena in gravitational collapse , '' adv . theor .
* 2 * ( 1998 ) 1 [ arxiv : gr - qc/9712084 ] .
+ c. gundlach , `` critical phenomena in gravitational collapse - living reviews , '' living rev .
* 2 * ( 1999 ) 4 [ arxiv : gr - qc/0001046 ] .
|
we present a self - contained formalism for calculating the background solution , the linearized solutions , and a class of generalized frobenius solutions to a system of scale invariant differential equations .
we first cast the scale invariant model into its equidimensional and autonomous forms , find its fixed points , and then obtain power - law background solutions .
after linearizing about these fixed points , we find a second linearized solution , which provides a _ distinct _ collection of power laws characterizing the deviations from the fixed point .
we prove that generically there will be a region surrounding the fixed point in which the complete general solution can be represented as a generalized frobenius - like power series with exponents that are integer multiples of the exponents arising in the linearized problem .
this frobenius - like series can be viewed as a variant of liapunov s expansion theorem . as specific examples we apply these ideas to newtonian and relativistic isothermal stars and demonstrate ( both numerically and analytically ) that the solution exhibits oscillatory power - law behaviour as the star approaches the point of collapse .
these series solutions extend classical results ; as exemplified for instance by the work of lane , emden , and chandrasekhar in the newtonian case , and that of harrison , thorne , wakano , and wheeler in the relativistic case .
we also indicate how to extend these ideas to situations where fixed points may not exist either due to `` monotone '' flow or due to the presence of limit cycles .
monotone flow generically leads to logarithmic deviations from scaling , while limit cycles generally lead to discrete self - similar solutions .
|
has been discovered on 1999 october 17 during a search for millisecond pulsar ( msps ) in galactic globular clusters ( ggcs ) in progress at the parkes radiotelescope @xcite .
it is a binary millisecond pulsar with a spin period of 3.26 ms , an orbital period of 0.84 days and very low eccentricity ( @xmath5 ) .
precise celestial coordinates ( @xmath6 , @xmath7 ) have been recently obtained for this source from pulsar timing observations @xcite .
this position is far away ( @xmath2 ) from the cluster optical center : indeed is the more off - centered pulsar among the sample of 44 msps whose position in the respective cluster is known , and it suggests that this object might be the result of strong interactions occurred in the cluster core .
@xcite explored a number of possibilities for the peculiar location of + @xmath0 : a careful analysis led to discard the hypothesis of a primordial binary ( born either in the halo or in the core of the cluster ) and to reject also the possibility of a 3-body scattering or exchange event off core stars .
hence they conjectured that a more massive target ( either a binary or a single intermediate - mass black hole , see also * ? ? ?
* ; * ? ? ?
* ) could have provided the necessary thrust to propel into its current halo orbit at an acceptable event rate . moreover , beside @xmath0 , ngc6752 hosts also the second most displaced msp ever seen in a globular ( psr j1911@xmath86000c ) and at least two ( out the three ) msps located in the cluster center ( psr - b and psr - e , see figure 2 in * ? ? ?
* ) experience strong acceleration , implying an unusually high central mass to light ratio @xmath9 @xcite .
@xcite showed that ngc6752 is a dynamically evolved cluster probably undergoing a post - core - collapse bounce and investigated scenarios for simultaneously explaining both the anomalous acceleration of psr - b and psr - e and the ejection of and psr j1911@xmath86000c , concluding that the existence of a binary black hole of intermediate mass @xcite could be a viable possibility .
optical detection of the companion to a millisecond pulsar in a globular cluster proved particularly helpful in assessing the origin and the evolution of the binary , besides supporting its cluster membership ( see , e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
in fact , unlike the systems in the galactic field , the age , metallicity , extinction , distance and hence intrinsic luminosity and radius can be estimated from the parent cluster parameters .
hence , we undertook a systematic program @xcite devoted to the optical identification of msps companion in ggcs ; as a part of this project in this letter we present the identification of the optical counterpart to the psr j1911 - 5958a companion .
the photometric data presented here consist of a set of high resolution images obtained by using the camera fors1 mounted at the _ antu _ unit telescope 1 ( ut1 ) of the very large telescope ( vlt ) at eso on cerro paranal ( chile ) on three nigths in march , april and may 2003 .
all the observations have been performed in the high resolution ( hr ) mode of fors1 . in this configuration
the plate - scale is @xmath10 and the fors1 @xmath11 pixel@xmath12 array has a global field of view of @xmath13 .
the data comprise eight 220s @xmath14-band exposures , five 360s @xmath15-band exposures and three 1500s @xmath16-band exposures , roughly centered on the ( hereafter msp - a ) nominal position .
all the observations were performed in service mode under good seeing conditions ( fwhm@xmath17 ) . a sub - image of @xmath18 pixels centered on the nominal position of the msp - a has been extracted from the original frames and carefully analized .
all the reductions have been carried out using romafot @xcite , a package specifically developed to perform accurate photometry in crowded fields , allowing for a visual inspection of the quality of the psf - fitting procedure . in order to optimize
the search for faint objects a _ median_-combined image in each band has been obtained and the search procedure has been performed on the deepest combined image .
then the masks with the star positions have been adapted to each image and the psf - fitting procedure performed on each individual frame separately .
the resulting instrumental magnitudes have been transformed to a common photometric system and then averaged .
thus , a final catalogue with the coordinates and the average instrumental magnitude in each filter has been compiled for all the stars identified in the considered sub - image .
photometric calibration of the instrumental magnitudes ( in the @xmath15 and @xmath16 band ) was obtained using four photometric standard stars @xcite secured under photometric conditions .
since no accurate calibration of the v filter was possible with the data secured in service , we calibrated the @xmath14 magnitude by using @xmath19 stars in common with the @xmath20 catalog recently published by @xcite .
the stars in common between the two catalogs permitted also an independent check of the calibration obtained in the @xmath15 band , displaying agreement within a few hundredths of magnitude .
the stars identified in the considered sub - image have been reported to an absolute astrometric system by using the 200 stars in common with the @xcite catalog ( already astrometized ) .
the details of the procedure adopted to derive the astrometric solution are described in other papers @xcite . in short ,
the new astrometric _ guide star catalog _ ( @xmath21 ) was used to search for astrometric standard stars lying in the considered field of view ( fov ) . at the end of the procedure ,
the rms residuals were of the order of @xmath22 both in ra and dec and we assume this value as representative of the astrometric accuracy .
figure 1 shows the @xmath23 and @xmath24 color magnitude diagram ( cmd ) for the stars ( _ large filled circles _ ) identified in the @xmath25 fors1 sub - image considered here .
most of the stars trace a clean and well defined main sequence ( ms ) spanning almost 8 mags in the @xmath16 band reaching @xmath26 .
only a few sparce objects , showing a blue excess , are visible on the left side of the ms .
the most extreme blue objects in the cmds are co white dwarfs ( wds ) : once the ms is matched , they nicely overlaps the position of a co - wds population ( shown as _ small empty circles _ in fig .
1 ) observed in this cluster by @xcite and ( * ? ? ?
* , in preparation ) . in particular , the colors and the luminosities of the three co - wds found here and of the previously observed population agree with the theoretical cooling sequence for @xmath27 hydrogen rich wds @xcite ( drawn as
_ heavy dashed line _ in fig . 1 ) .
on the basis of the accurate astrometric positions obtained from the photometric catalog , we identified a blue object ( hereafter @xmath0 ) lying at only @xmath28 from the nominal position of @xmath0 .
finding chart ( in the @xmath15-band filter ) for + is shown in fig . 2 .
only a few objects ( @xmath29 ) are lying in the region of the cmd to the left of the ms : considering the fov of the sub - image , there is a very small probability ( @xmath30 ) of detecting one of the blue objects just by chance in a circular aperture of radius @xmath31 ( corresponding to the uncertainty in the relative radio - optical astrometry ) centered on the pulsar position .
the position of is marked as a _
large empty square _ in the cmds in fig . 1 , whereas absolute coordinates and magnitudes are listed in table 1 .
taking into account the expected variability of this object and the global uncertainty in the calibration of the photometric zero - point , we conservatively adopted an overall uncertainty in the reported magnitudes of 0.15 mag . while we were writing this letter a not yet refereed paper by bassa et al .
appeared on the web ( astro - ph/0307340 ) also presenting the optical identification of this object .
though the identification is consistent with that presented here , it can be considered , at most , a preliminary detection since the object is nearly at the detection limit in the exposures utilized in that paper .
the deep vlt observations presented here allow us to measure the object with a quite high s / n ratio ( @xmath32 ) and to derive the photometric properties of the star with much higher accuracy .
the location of in the cmds ( @xmath33 mag bluer than the ms in the diagram ) excludes it is a co - wd whilst it resembles @xmath34 , the companion to psr j00247203u in 47 tuc @xcite , which was suggested @xcite to be a low mass helium core wd ( he - wd ) from the comparison with theoretical models by @xcite .
however , the he - wd tracks used by @xcite were computed for progenitors with high ( solar ) metallicity ( @xmath35 ) . here
we have applied the same procedure taking advantage of a new set of tracks specifically computed @xcite for globular cluster applications , i.e. assuming progenitors with low metallicities . in particular
, we used the cooling tracks at @xmath36 which are the closest to the cluster metallicity ( @xmath37=-1.43\pm0.04 $ ] ) as recently derived by @xcite .
the cooling sequences for two he - wd masses ( 0.172 and 0.197 , respectively ) are over - plotted in fig . 1 ( _ light dashed lines _ ) .
the models have been drawn by adopting a distance modulus @xmath38 ( from the homogeneous distance scale for 61 ggcs derived by * ? ? ? * ) and a reddening @xmath39 @xcite .
note that the distance modulus adopted here is consistent both with the estimate obtained from the wd cooling sequence ( @xmath40 ) by @xcite and with the most recent determination ( @xmath41 ) derived from ms fitting by @xcite .
inspection of fig .
1 reveals that luminosity and color of well agree with its being a he - wd of mass in the range @xmath42 in particular , from the @xmath43 models of masses @xmath44 and @xmath45 by @xcite , we derive the following estimates for the properties of @xmath0 : a temperature in the interval @xmath46 , a gravity @xmath47 , a luminosity @xmath48 , a radius @xmath49 cm and a cooling age in the range @xmath50 gyr .
we have identified the optical companion to the millisecond pulsar located at 3.3 half mass radii from the center of the globular cluster ngc6752 . @xmath51 and @xmath14 magnitudes ( and related colors ) of the optical source are compatible with its being a helium white dwarf hosted in the globular s halo . from the pulsar mass - function ( 0.0026887 ,
* ) , it results a minimum companion mass ( corresponding to a system seen edge - on , i.e. with orbital inclination @xmath52 ) @xmath53 for a pulsar mass of @xmath54 @xcite . given the upper limit ( @xmath55 ) of the range of masses of inferred from the cooling tracks of @xcite
, we can constrain the inclination @xmath56 to be larger than @xmath57 . adopting a larger value for the pulsar mass would result in a even larger lower limit for the orbital plane inclination .
dynamical friction can rapidly drive back to the cluster core a object of total mass of order @xmath58 moving on a highly eccentric orbit in the cluster potential .
in particular , ejection of from the core to the outskirts of ngc6752 ( implying a radial orbit ) can not be occurred more than @xmath59 gyr ago @xcite .
the cooling age of is longer than @xmath60 implying that any dynamical event responsible for the ejection probably acted on an already recycled millisecond pulsar .
hence , the optical identification of tends to exclude _
( i ) _ the scenarios in which the recycling process occurred after ( or was triggered by ) the dynamical encounter which propelled the pulsar in the cluster halo ( but see the discussion in sigurdsson , 2003 ) and _ ( ii ) _ any kinds of interactions imparting a significant eccentricity to the binary pulsar ( having excluded a recent mass transfer phase , the circularization of the system at the present level , @xmath61 would require a time much longer than @xmath62 ) .
in particular , among the proposed black hole hypotheses @xcite a binary black hole of mass in the range 10 - 200 seems the most plausible candidate . since the first discovery of an optical counterpart to a binary msp companion in ggc ( @xmath34 , the companion to psr j0024@xmath87203u in 47 tuc ; * ? ? ? * ) , the zoo of the optical msp counterparts in ggcs is rapidly increasing : a surprisingly bright object ( com j1740@xmath85340 , either a subgiant or a heated ms ) has been discovered @xcite in ngc6397 , and a very faint source ( w29 , the companion to psr j0024@xmath87204w ) has been found by @xcite in 47 tuc . moreover , an additional potential msp companion ( w34 in 47 tuc ) has been recently discussed by @xcite .
3 shows a comparison of the photometric properties of the available optical identifications of msp companions hosted in ggcs .
the ms stars of ngc6752 , the models of co - wd and two he - wd cooling sequences by @xcite are also plotted in fig . 3 as reference .
two among the five sources seem really peculiar : i.e. , the bright object in ngc6397 ( which is as luminous as the turn off stars and shows quite red colours ) and the faint w29 in 47 tuc , which is also too red to be a he - wd @xcite .
bassa et al .
( 2003 ) suggested a similiarity between the photometric properties of and w29 in 47 tuc .
as can be seen from figure 3 , is significantly brighter and bluer than w29 and it turns out to be more similar to @xmath34 .
indeed , @xmath34 and are found to lie nearly on the same mass he - wd cooling sequence and w34 in 47 tuc curiously shares the same photometric properties of . indeed ,
if confirmed as a msp companion , w34 would be the third he - wd companion orbiting a msp in ggcs roughly located on the same - mass cooling sequence .
if further supported by additional cases , this evidence could confirm that a low mass @xmath63 he - wd orbiting a msp is the favoured system generated by the recycling process of msps not only in the galactic field @xcite but also in ggcs @xcite .
is relatively bright and will allow detailed follow - up observations : both optical time series photometry and spectroscopy of this object are planned at the eso vlt , in order to determine the parameters of the he - wd companion , the mass ratio of the two components of the binary and eventually the mass of the neutron star .
colpi , m. , mapelli , m. , & possenti , a. 2003a , `` coevolution of black hols and galaxies '' , carnegie observatories astrophysics series , vol .
1 : ed . l. c. ho ( pasadena : carnegie observatories ) , astro ph/0302545
|
we have used deep high - resolution multiband images taken at the eso _ very large telescope _ to identify the optical binary companion to the millisecond pulsar ( @xmath0 ) located in the halo of the galactic globular cluster ngc6752 .
the object turns out to be a blue star whose position in the color magnitude diagram is consistent with the cooling sequence of a low mass ( @xmath1 ) , low metallicity helium white dwarf ( he - wd ) at the cluster distance .
this is the second he - wd which has been found to orbit a millisecond pulsar in ggcs .
curiously both objects have been found to lie on the same mass he - wd cooling sequence
. the anomalous position of with respect to the globular cluster center ( @xmath2 ) suggested that this system has recently ( @xmath3 gyr ) been ejected from the cluster core as the result of a strong dynamical interaction .
the data presented here allows to constrain the cooling age of the companion within a fairly narrow range ( @xmath4 gyr ) , therefore suggesting that such dynamical encounter must have acted on an already recycled millisecond pulsar .
|
the casimir effect @xcite implies that there is a force acting between closely spaced electrically neutral bodies following from the zero - point oscillations of the electromagnetic field .
the casimir force can be viewed as an extension of the van der waals force to large separations where the retardation effects come into play . within a decade of casimir s
work , lifshitz and collaborators @xcite introduced the role of optical properties of the material into the van der waals and casimir force . in the last few years
, the advances following from both fundamental physics and nanotechnology have motivated careful experimental and theoretical investigations of the casimir effect .
the first modern experiments were made with metal test bodies in a sphere - plate configuration , and their results are summarized in ref . @xcite . in subsequent experiments the lateral casimir force between corrugated surfaces @xcite and the pressure in the original casimir configuration @xcite have been demonstrated .
later experiments @xcite have brought the most precise determination of the casimir pressure between two metal plates .
the rapid theoretical progress has raised fundamental questions on our understanding of the casimir force between real metals at nonzero temperature .
specifically , the role of conductivity processes and the related optical properties of metals at quasi - static frequencies has become the subject of discussions @xcite .
one of the most important applications of the casimir effect is the design , fabrication and function of micro- and nanoelectromechanical systems such as micromirrors , microresonators , nanotweezers and nanoscale actuators @xcite .
the separations between the adjacent surfaces in such devices are rapidly falling below a micrometer , i.e. , to a region where the casimir force becomes comparable with typical electrostatic forces .
it is important that investigations of the casimir force be done in semiconductors as they are the material of choice for the fabrications of optomechanical , micro- and nanoelectromechanical systems .
while the role of conductivity and optical properties of materials can be checked in metals , semiconductors offer better control of the related parameters ( charge carrier density , defect density , size etc . ) and will provide an exhaustive check of the various models .
reference @xcite pioneered the measurement of the casimir force acting between a semiconductor surface , single crystal si wafer , and a gold coated sphere .
the experimental data obtained for a wafer with the concentration of charge carriers @xmath1 were compared with the lifshitz theory at zero temperature and good agreement was found at a 95% confidence level . at the same time , the theory describing a `` dielectric '' si plate with a concentration @xmath2 was excluded by experiment at 70% confidence .
this allows one to conclude that the casimir force is sensitive to the conductivity properties of semiconductors .
this conclusion has found direct experimental confirmation in ref .
@xcite where the casimir forces between a gold coated sphere and two different si wafers with the concentrations of charge carriers @xmath3 and @xmath4 have been measured .
the difference of the mesured forces for the two conductivities was found to be in good agreement with the corresponding difference of the theoretical results computed at zero temperature ( note that the sensitivity of force mesurements in refs .
@xcite was not sufficient to detect the thermal corrections predicted in refs .
@xcite ) . in the most precise experiment on the casimir force between a metal and a semiconductor @xcite , the density of charge carriers in a si membrane of 4@xmath5 m thickness
is changed from @xmath6 to @xmath7 through the absorption of photons from a laser pulse .
this is a differential experiment where only the difference of the casimir forces in the presence and in the absence of a laser pulse was measured .
this decreases the experimental error to a fraction of a 1pn and allows one to check the role of conductivity processes in semiconductors at the laboratory temperature @xmath8k .
the experimental data for the difference casimir force as a function of separation were compared with the lifshitz theory and the outcome was somewhat puzzling .
the data were found to be in excellent agreement with the theoretical difference force computed at @xmath8k under the assumption that in the absence of laser light si possesses a finite static dielectric permittivity .
by contrast , if theory takes into account the dc conductivity of si in the absence of laser light , it is excluded by the data at a 95% confidence level .
this is somewhat analogous to the above - mentioned problems for two real metals where the inclusion of the actual conductivity processes at low frequencies also leads to disagreement with experiment @xcite .
the fundamental questions on the role of scattering processes and conductivity at low frequencies in the casimir force have to be clarified for further progress in the field and this calls for new precise experiments . in this paper
we propose two experiments on the casimir force between a metal sphere and semiconductor plate which can shed light on the applicability of the lifshitz theory at nonzero temperature to real materials . in the first of these experiments , the patterned si plate with two sections of different dopant densities
is oscillated in the horizontal direction below the au coated sphere . as a result ,
the sphere is subject to the difference casimir force which can be measured using the static and dynamic techniques .
this experimental scheme promises a record sensitivity to force differences at the level of 1fn . in the second experiment
, we propose to demonstrate the modulation of the casimir force by optically switching the insulator - metal transition in vo@xmath0 films @xcite . the phase transition between the insulator and metal leads to a change in the charge carrier density of order @xmath9 , which is sufficient to bring about a large change in the casimir force .
for both experiments the related theory is elaborated and the magnitudes of casimir forces are computed with application to the experimental configurations .
the effects from using different theoretical approaches to the description of conductivity processes are carefully analyzed and shown to be observable in the proposed experiments .
the present status of developing the apparatus at uc riverside , its calibration and sensitivity is presented .
the proposed experiments offer a precision test of the role of conductivity , optical properties and scattering in the lifshitz theory of the van der waals and casimir force at nonzero temperature .
they also open up possibilities of radically new nanomechanical devices using the casimir force in imaging applications .
the paper is organized as follows . in sec .
ii , we describe the proposed experiment on the difference casimir force with the patterned semiconductor plate . the brief description of the experimental apparatus and preliminary results are also provided .
section iii contains the calculation of the difference casimir force and equivalent pressure in the patterned geometry using the lifshitz theory at nonzero temperature and different models of conductivity processes at low frequencies . in sec .
iv we propose the experiment on the modulation of the casimir force through a metal - insulator transition . the experimental scheme and
some preliminary tests are discussed .
section v presents theoretical computations of casimir forces in a insulator - metal transition on the basis of the lifshitz theory at nonzero temperature and using different models for the conductivity processes .
section vi contains our conclusions and a discussion .
the aim of this experiment is to gain a fundamental understanding of the role of carrier density in the casimir force using a nanofabricated patterned semiconductor plate .
the proposed design for the experiment is shown schematically in fig . 1
. the gold coated polystyrene sphere of about 100@xmath5 m radius is attached to a cantilever of an atomic force microscope ( afm ) specially adapted for making sensitive force measurements . instead of the simple single crystal si substrate used in the previous experiments @xcite , here a patterned si plate
is employed .
this plate is composed of single crystal si specifically fabricated to have adjacent sections of two different charge carrier densities @xmath10 and @xmath11 . in this range of doping densities ,
the plasma frequency @xmath12 will change by a factor of 100 .
additional changes in the @xmath12 can be brought about by using both @xmath13(b ) and @xmath14(p ) type dopants as electrons and holes differ in their effective mass by 30% . the preparation of the si sample with the two sections having different conductivities is shown schematically in fig .
, one half of the bare si wafer of 0.3 to 0.5 mm thickness @xcite having the lower conductivity shown in ( a ) is masked with a photoresist as shown in ( b ) .
next in ( c ) the exposed half of the si wafer is doped with p ions using ion implantation leading to a higher density of electrons in the exposed half in ( d ) .
rapid thermal annealing and chemical mechanical polishing of the patterned si plate will be done as the last step .
this is to ensure that there are no surface features resulting from the fabrication .
similar nanofabrication procedures of semiconductors were used in our previous work @xcite .
sharp transition boundaries between the two sections of si plate of width less than 200 nm are possible .
the limitation comes from interdiffusion and the resolution of the ion implantation procedure to be used .
it might be necessary to further limit interdiffusion by the creation of a narrow 100 nm barrier between the two doped regions .
identically prepared but unpatterned samples will be used to measure the properties which are needed for theoretical computations .
the carrier concentration will be measured using hall probes .
this will yield an independent measurement of the plasma frequencies .
a four probe technique will be used to measure the conductivity @xmath15 . from the conductivity and the charge carrier concentration ,
the scattering time @xmath16 will be found , where @xmath17 is the permittivity of free space .
the first important improvement of this experiment , as compared with refs .
@xcite , is the direct measurement of difference forces when the patterned plate is oscillated below the sphere .
this measurement is performed as follows .
the patterned si plate will be mounted on the piezo below a au coated sphere as is shown in fig . 1 .
the si plate is positioned such that the boundary is below the vertical diameter of the sphere .
the distance between the sphere and si plate @xmath18 will be kept fixed and the si plate will be oscillated in the horizontal direction using the piezo such that the sphere crosses the boundary in the perpendicular direction during each oscillation ( a similar approach was exploited in ref .
@xcite for constraining new forces from the oscillations of the au coated sphere over two dissimilar metals , au and ge ) .
the casimir force on the sphere changes as the sphere crosses the boundary .
this change corresponds to the differential force @xmath19 equal to the difference of the casimir forces due to two different charge carriers densities @xmath20 and @xmath14 , respectively .
this causes a difference in the deflection of the cantilever . in order to reduce the random noise by averaging
, the periodic horizontal movement of the plate will be of an angular frequency @xmath21hz .
the amplitude of the plate oscillations is limited by the piezo characteristics , but will be of order 100@xmath22 m , much larger than typical transition region of 200 nm
. the experiment will be repeated for different sphere - plate separations in the region from 100 to 300 nm .
the measurement of absolute separations will be performed by the application of voltages to the test bodies as described in ref .
@xcite .
the second major improvement in this experiment in comparison with all previous measurements of the casimir force is the increased sensitivity .
this will be achieved through the use of the interferometer based low temperature afm capable of operating over wide temperature range spanning from 360 to 4k , and the use of two measurement techniques , a static one and a dynamic one .
a picture of the newly constructed experimental apparatus of a low - temperature afm is shown in fig . 3 . here
the cantilever deflection is measured interferometrically and therefore has much higher sensitivity than photodiodes used in the previous work @xcite .
the detection of a difference force @xmath23 will be done by two alternative techniques .
the first technique , a static one , reduces to the direct measurement of @xmath23 as described above .
we are presently performing the initial tests and calibration trials at 77k .
an oil free vacuum with a pressure of around @xmath24torr is used .
the instrument is magnetically damped to yield low mechanical coupling to the environment .
the temperature can be varied with a precision of 0.2k .
we have fabricated special conductive cantilevers with a spring constant @xmath25n / m .
the magnitude of @xmath26 is found by applying electrostatic voltages to the plate as discussed in ref .
@xcite . to accomplish this ,
si cantilevers were thermal diffusion doped to achieve the necessary conductivity .
note that conductive cantilevers are necessary to reduce electrostatic effects .
the cantilever - sphere arrangement has been checked to be stable at 77k .
the experimental setup using the static measurement technique allows not only a demonstration but a detailed investigation of the influence of carrier density , conductivity and scattering in semiconductors on the casimir force . according to the results of refs .
@xcite and calculations below in sec .
iii , the magnitude of the difference force to be measured is about several pn .
different theoretical models of conductivity processes at low frequencies lead to predictions differing by approximately 1pn within a wide separation region ( see fig . 5 in sec .
the described setup provides excellent opportunity for precise measurement of forces of order and below 1pn .
we have measured a resonance frequency of the cantilever of @xmath27hz , a quality factor @xmath28 , and an equivalent noise bandwidth @xmath29hz .
the resultant force sensitivity of our cantilever at @xmath30k with the gold coated sphere attached was determined following ref .
@xcite to be @xmath31 where @xmath32 is the boltzmann constant . even bearing in mind that the systematic error may be up to an order of magnitude larger ,
the sensitivity ( [ eq2 ] ) presents considerable possibilities for the precise investigation of the difference casimir force @xmath33 .
the second technique for the detection of the difference casimir force is a dynamic one @xcite . here
this technique is applied not for a direct measurement of @xmath33 but rather for the experimental determination of the equivalent difference casimir pressure between the two parallel plates ( one made of au and the other one , a patterned si plate ) .
the cantilever - sphere system oscillates in the verical direction due its thermal noise with a resonant frequency @xmath34 in the absence of the casimir force , where @xmath35 is the mass of the system
. the thermal noise spectrum of the sphere - cantilever system is measured and fit to a lorentzian to identify the peak resonant frequency , @xmath36 .
the shift of @xmath36 in the presence of the casimir force when for example the sphere is positioned above a section of the patterned si plate with the density of charge carriers @xmath20 is equal to @xcite @xmath37 next the plate is oscillated in the horizontal direction with a frequency @xmath38 . as a result ,
the frequency shift @xmath39 between the resonant frequencies above the two different sections of the patterned si plate is measured .
using the proximity force approximation @xcite , we determine the difference casimir pressure @xmath40 between the two parallel plates ( the au one and the patterned si ) .
note that the systematic error from the use of the proximity force approximation was recently confirmed to be less than @xmath41 @xcite .
equations ( [ eq4 ] ) and ( [ eq5 ] ) express the difference casimir pressure through the measured shift of the resonance frequency above the two halfs of the patterned plate .
as is shown in the next section , the measurements of the difference casimir pressure using the dynamic technique provides us with one more test of the predictions of the different models of conductivity processes at low frequencies .
the experiment on the difference casimir force from a patterned si plate , as described in this section , allows variation of charge carrier density by the preparation of different semiconductor samples . thus , the proposed measurements should provide a comprehensive understanding on the role of conductivity and optical processes in the casimir force for nonmetallic materials and discriminate between competing theoretical approaches
the difference casimir force and the equivalent casimir pressure from the oscillation of the patterned si plate below an au coated sphere at @xmath8k in thermal equilibrium are given by the lifshitz theory .
in the static technique the data to be compared with theory is the difference of casimir forces acting between the sphere and two sections of the patterned plate .
this difference is obtained from the casimir energy between two parallel plates , as given by the lifshitz theory , using the proximity force approximation @xcite @xmath42\,\left[1- r_{te;\tilde{n}}(\xi_l , k_{\bot } ) r_{te}(\xi_l , k_{\bot})e^{-2q_lz}\right]}{\left[1- r_{tm;{n}}(\xi_l , k_{\bot } ) r_{tm}(\xi_l , k_{\bot})e^{-2q_lz}\right]\,\left[1- r_{te;{n}}(\xi_l , k_{\bot } ) r_{te}(\xi_l , k_{\bot})e^{-2q_lz}\right]}\ , .
\nonumber\end{aligned}\ ] ] here @xmath43 with @xmath44 are the matsubara frequencies , @xmath45 , and @xmath46 is the projection of the wave vector on the boundary planes .
the reflection coefficients on the au plane for the two independent polarizations of the electromagnetic field ( transverse magnetic and transverse electric modes ) are @xmath47 where @xmath48 and @xmath49 is the dielectric permittivity of au along the imaginary frequency axis . in a similar way ,
the reflection coefficients on the two sections of a patterned si plate with charge carrier densities @xmath20 and @xmath14 are given , respectively , by @xmath50 where @xmath51 and @xmath52 are the dielectric permittivities of si with charge carrier densities @xmath20 and @xmath14 along the imaginary frequency axis .
in the dynamic technique the data to be compared with theory is the equivalent difference casimir pressure between two parallel plates , one made of au and the other one a patterned si plate . using the same notations as above ,
the difference casimir pressure is given by @xmath53^{-1}+\left [ r_{te;\tilde{n}}^{-1}(\xi_l , k_{\bot } ) r_{te}^{-1}(\xi_l , k_{\bot})e^{2q_lz}-1\right]^{-1}\right .
\nonumber \\ & & \phantom{aaaa}\left .
-\left[r_{tm;{n}}^{-1}(\xi_l , k_{\bot } ) r_{tm}^{-1}(\xi_l , k_{\bot})e^{2q_lz}-1\right]^{-1}- \left[r_{te;{n}}^{-1}(\xi_l , k_{\bot } ) r_{te}^{-1}(\xi_l , k_{\bot})e^{2q_lz}-1\right]^{-1}\right\}\ , .
\nonumber\end{aligned}\ ] ] note that in eqs .
( [ eq6 ] ) and ( [ eq9 ] ) we have replaced the 100 nm au coating and the 0.30.5 mm si plate for an au and si semispaces , respectively . using the lifshitz formula for layered structures @xcite it is easy to calculate the force and pressure errors due to this replacement . for example , for an au layer at a typical separation of 100 nm this error is about 0.01% . for si a finite thickness of the plate @xmath54 markedly affects the casimir force when the separation distance @xmath18 exceeds the thickness , i.e. , @xmath55 @xcite . in our case , however , even at the largest separation considered ( @xmath56 nm ) the ratio of the separtion to the plate thickness @xmath57 .
this is similar to the case of the experiment @xcite where the finite thickness of si membrane also does not influence the magnitude of the casimir force because at separations @xmath58 nm where statistically meaningful results were obtained @xmath59 .
we have performed computations of the difference casimir force ( [ eq6 ] ) and difference casimir pressure ( [ eq9 ] ) for samples with typical values of charge carrier concentrations @xmath20 and @xmath14 as used in experiments @xcite .
both sections of the si plate were chosen to have electron conductivity and doped with p. for the section of the plate with higher concentration of charge carriers the values @xmath60 ( such a sample was fabricated in ref .
@xcite ) and @xmath61 were used in the computations .
the respective dielectric permittivity along the imaginary frequency axis can be represented in the form @xcite @xmath62 here @xmath63 is the permittivity of high - resistivity ( dielectric ) si along the imaginary frequency axis computed in ref .
@xcite by means of the dispersion relation using the tabulated optical data for the complex index of refraction @xcite .
the values of the plasma frequencies and relaxation parameters are the following @xcite : @xmath64rad / s , @xmath65rad / s , @xmath66rad / s , @xmath67rad / s . in fig .
4 the dielectric permittivities of the samples with high concentrations of charge carriers @xmath68 and @xmath69 are shown as solid lines 1 and 2 , respectively . in the same figure , the dashed line @xmath70 shows the permittivity of high - resistivity , dielectric , si @xcite and the dotted line the permittivity of au computed in ref .
@xcite using the tabulated optical data of ref .
@xcite .
below we will use two models for the permittivity of the section of the si plate with lower concentration of charge carriers @xmath14 .
calculations show that for any @xmath71 ( this interval includes the experimental value of @xmath72 in ref .
@xcite ) , the obtained values of @xmath73 and , thus , of @xmath23 do not depend on @xmath14 .
because of this we use in the computations @xmath74 , the plasma frequency @xmath75rad / s and the relaxation parameter @xmath76rad / s @xcite ( note that for @xmath77 the value of the relaxation parameter does not effect the magnitude of the casimir force ) .
then the dielectric permittivity of this section of the si plate along the imaginary frequency axis is given by @xmath78 and is shown as the dashed line @xmath79 in fig .
this is one model of si with a lower concentration of charge carriers referred to below as model ( @xmath79 ) . as is seen in fig . 4 ,
the dashed line @xmath79 , and thereby all respective lines for samples with the concentration of charge carriers smaller than @xmath80 , deviate from the permittivity of dielectric si ( line @xmath70 ) only at frequencies below the first matsubara frequency @xmath81 . because of this , it is common ( see , e.g. , @xcite ) to neglect the small conductivity of high - resistivity materials at low frequencies and describe them in the frequency region below the first matsubara frequency by the static dielectric permittivity . in our case
this leads to @xmath82 which is the other model for si with a lower concentration of charge carriers referred to below as model ( @xmath70 ) . from eq .
( [ eq12 ] ) and fig .
4 at all frequencies @xmath83 it follows : @xmath84 . to be exact , at any @xmath85 the density of free charge carriers @xmath14 in semiconductors ( and even in dielectrics ) and thus the conductivity are nonzero ( @xmath86 ) .
thus , the model ( [ eq11 ] ) should be considered as more exact than the model ( [ eq12 ] ) . at the same time , if we note that for @xmath77 the conductivity is small , it should be expected that both models should lead to practically identical results .
this is , however , not so . in fig .
5 we present the computational results for the difference casimir force using eq .
( [ eq6 ] ) .
the solid line @xmath87 demonstrates the values of the difference casimir force versus separation for the patterned si plate with a higher concentration of charge carriers @xmath68 computed under the assumption that the lower concentration section of the plate is described by eq .
( [ eq12 ] ) , i.e. , the conductivity processes at low frequencies are neglected .
the dashed line @xmath88 shows the difference casimir force as a function of separation computed with the same @xmath68 but taking into account the conductivity processes at low frequencies in accordance with eq .
( [ eq11 ] ) . as is seen from the comparison of lines @xmath87 and @xmath88 , the difference casimir forces computed using eqs .
( [ eq12 ] ) and ( [ eq11 ] ) differ by 1.2pn at a separation @xmath89 nm and this difference slowly decreases to approximately 0.14pn at a separation @xmath56 nm . the lines @xmath90 and @xmath91 present similar results for the case when the higher charge carrier density is equal to @xmath69 . as is seen from fig .
5 , decreasing the higher concentration by an order of magnitude decreases the predicted magnitude of the difference casimir force by more than two times , but leaves the same gap between the predictions of two different models of the permittivity at low frequencies .
importantly , our predictions do not depend on the discussions mentioned in the introduction on the optical properties of metals at quasi - static frequencies @xcite .
the resolution of this controversy affects only the value of the au reflection coefficient @xmath92 at zero frequency .
the latter , however , does not contribute to the difference casimir force ( [ eq6 ] ) and pressure ( [ eq9 ] ) because for dielectrics and semiconductors @xmath93 regardless of what model ( [ eq11 ] ) or ( [ eq12 ] ) is used for the description of the dielectric permittivity at low frequencies .
the obtained difference between the lines @xmath94 and @xmath95 in fig .
5 is completely explained by the different contributions of semiconductor reflection coefficient @xmath96 when one uses eq .
( [ eq11 ] ) or eq .
( [ eq12 ] ) to describe the dielectric permittivity at low frequencies . regarding the semiconductor section with a higher charge carrier density @xmath20 , from eqs .
( [ eq8 ] ) , ( [ eq10 ] ) it is always valid that @xmath97 however , for the section of the plate with a lower charge carrier density @xmath14 it follows from eq .
( [ eq8 ] ) that @xmath98 when eq .
( [ eq11 ] ) or eq .
( [ eq12 ] ) are used , respectively .
thus , the difference between the lines @xmath87 and @xmath88 ( and the same difference between the lines @xmath90 and @xmath91 ) can be found analytically .
taking only the zero - frequency contribution in eq .
( [ eq6 ] ) and subtracting the difference casimir force calculated using eq .
( [ eq11 ] ) [ model ( @xmath79 ) ] from the difference casimir force calculated using eq .
( [ eq12 ] ) [ model ( @xmath70 ) ] one obtains @xmath99\right\}. \label{eq15}\ ] ] here @xmath100 is the riemann zeta function , and li@xmath101 is the polylogarithm function . the results using the analytic eq .
( [ eq15 ] ) coincide with the differences between the lines @xmath94 and @xmath95 in fig . 5 computed numerically . in the experiment
@xcite the difference casimir force between au coated sphere and si plate illuminated with laser pulses was first measured . in the presence of light
the charge carrier density was about @xmath7 and in the absence of light of about @xmath6 .
the experimental data were shown to be in agreement with model ( @xmath70 ) which uses the finite static dielectric permittivity of si . the model ( @xmath79 ) which includes the dc conductivity of si
was excluded at 95% confidence within the separation region from 100 to 200 nm . as was discussed above , in the framework of the lifshitz theory
this result is rather unexpected .
bearing in mind that illumination with laser pulses leads to several additional sources of errors discussed in ref .
@xcite , it is of vital interest to verify the obtained conclusions in a more precise experiment with patterned si plates . the comparison of the experimental sensitivities presented in sec .
ii with the magnitudes of the difference casimir forces computed here using different theoretical models demonstrate that the proposed experiment with patterned semiconductor plate will bring decisive results on the discussed problems in the lifshitz theory at nonzero temperature .
the calculations of the difference casimir pressure determined in the dynamic mode of the proposed experiment leads to results analogous to those for the difference force . the calculation results using eq .
( [ eq9 ] ) with the same values of parameters as above and two models of lower conductivity si are presented in fig .
the difference casimir pressures between an au plate and a patterned si plate with the higher densities of charge carriers @xmath102 ( one section of the plate ) and lower @xmath14 ( another section of the plate ) are shown with solid lines @xmath87 and @xmath90 , respectively , computed under the assumption that si with the lower @xmath14 possesses a finite permittivity ( [ eq12 ] ) at zero frequency .
the dashed lines @xmath88 and @xmath91 are obtained under the assumption that si with the lower @xmath14 is described by the permittivity ( [ eq11 ] ) which goes to infinity when the frequency goes to zero .
as is seen in fig .
6 , the difference casimir pressure with a patterned plate with charge carrier densities @xmath68 and @xmath14 equals 250mpa at a separation @xmath89 nm [ model ( @xmath70 ) of low conductivity section of the plate ] and the difference in predictions for the two models equals 38.6mpa
. the proposed experiment of the difference casimir pressure can reliably discriminate between the solid and dashed lines in fig .
6 thus providing one more test for the lifshitz theory at nonzero temperature .
notice that in a similar way to the force , the differences between the lines @xmath94 and @xmath95 in fig .
6 are expressed analytically by taking the zero - frequency contributions in eq .
( [ eq9 ] ) : @xmath103\right\}. \label{eq16}\ ] ] calculations using eq .
( [ eq16 ] ) lead to the same differences between the lines @xmath94 and @xmath95 as were computed numerically in fig . 6 .
the exciting possibility for the modulation of the casimir force due to a change of charge carrier density is offered by semiconductor materials that undergo the insulator - metal transition with the increase of temperature . such a transition leads to a change of the carrier density of order @xmath9 .
although in literature it is common to speak about insulator - metal transition , this can be considered as a transformation between two semiconductor phases with lower and higher charge carrier densities @xmath14 and @xmath20 , respectively . as was shown above ,
this is sufficient to bring about a large change in the casimir force . from a fundamental point of view , the modulation of the casimir force due to the phase transition will offer one more precision test of the role of conductivity and optical properties in the lifshitz theory of the casimir force .
this experiment suggests some advantages as compared to the difference force measurement with a patterned plate considered in secs .
ii and iii .
first , because of the large change in the magnitude and bandwidth of the optical properties in a phase transition , the modulation of the casimir force will be larger .
second , an insulator - metal transition does not require the special fabrication of patterned plates with one section having a high carrier density , which might not be compatible with robust device design . keeping in mind that the increase of temperature necessary for the phase transition can be induced by laser light , this opens up the possibility of radically new nanomechanical devices using the casimir force in image detection .
the phase transition can be also brought about through electrical heating of the material . in this experiment
we propose to measure the change of the casimir force acting between an au coated sphere and a vanadium dioxide ( vo@xmath0 ) film deposited on sapphire substrate which undergoes the insulator - metal transition with the increase of temperature .
it has been known that vo@xmath0 crystals and thin films undergo an abrupt transition from semiconducting monoclinic phase at room temperature to a metallic tetragonal phase at 68@xmath104c @xcite .
the phase transition causes the resistivity of the sample to decrease by a factor of @xmath9 from 10@xmath105 cm to @xmath106 cm ( i.e. , the same change as for two semiconductor half plates in sec .
ii with lower and higher charge carrier densities ) .
in addition , the optical transmission for a wide region of wavelengths extending from @xmath107 m to greater than @xmath108 m , decreases by more than a factor of 10100 .
the schematic of the experimental setup is shown in fig .
7 . in this figure ,
light from a chopped 980 nm laser will be used to heat the vo@xmath0 film @xcite .
about 10100mw power of the 980 nm laser is required to bring about all optical switching of vo@xmath0 films . the same procedure as outlined in sec .
ii ( the static technique ) will be used in the measurement of the modulation of the casimir force including the interferometric detection of cantilever deflection .
the schematic of the setup is similar to the one used in ref .
@xcite in the demonstration of optically modulated dispersion forces .
an important point is that in ref .
@xcite the absorption of light from a 514 nm ar laser led to an increase of charge carrier density .
by contrast , here the wavelength of a laser is selected in such a way that light only leads to heating of a vo@xmath0 film but does not change the number of free charge carriers @xcite . as a first step towards studying the role of the insulator - metal transition in the casimir force , we have recently fabricated thin films of vo@xmath0 on sapphire plates .
the preliminary results are shown in fig .
it is observed that we have obtained more than a factor of 10 change in the resistivity of the film .
these films were prepared by thermal evaporation of vo@xmath0 powder . while films of appropriate thickness approaching 100 nm and roughness of about 2 nm ( shown in fig .
9 ) can be obtained by this procedure , it is not optimal as it leads to the non - stoichimetric formation of the mixed valence states of the vanadium oxide ( vo@xmath109 ) . in the future
rf magnetron sputtering will be used to make the films @xcite .
vo@xmath0 films using this technique have been shown to have the @xmath9 change in resistivity and a corresponding large change in optical reflectivity and spectrum .
the aim of the proposed experiment on the influence of insulator - metal transition on the casimir force is two fold : applications for actuation of nanodevices through a modulation of the casimir force , and to perform fundamental tests on the theory of dispersion forces . to accomplish this ,
two types of measurements are planned . in the first we plan to demonstrate the modulation of the casimir force through an optical switching of the insulator - metal transition .
this modulation will lead to novel microdevices as optical and electrical switches , optical modulators , optical filters and ir detectors that can be actuated optically through the absorption of ir radiation .
importantly , such devices can be integrated with si technology which is used in the fabriaction of microelectromechanical systems @xcite . in the second type of measurements , the variable temperature atomic force microscope described in sec .
ii will be used to perform precision measurements of the casimir force between a gold coated sphere and vo@xmath0 film . here , the casimir force will be measured at different temperatures from room temperature through 80@xmath104c .
this temperature range spans the dielectric ( semiconducting ) and metal regions of vo@xmath0 .
careful comparison of the experimental data and the theory ( see the next section ) will be done to understand the role of conductivity and losses in both phases of vo@xmath0 .
the casimir force acting between a metal coated sphere and the vo@xmath0 film on a sapphire plate both before and after the phase transition ( i.e. , in the insulating and metal phases or , more exactly , in the semiconductor phases with lower and higher charge carrier densities ) are expressed by the lifshitz formulas in accordance with eqs .
( [ eq1 ] ) and ( [ eq6 ] ) . as above ,
we label the higher concentration of charge carriers @xmath20 and the lower concentration @xmath14 . to compute the casimir force before and after the phase transition one needs the optical properties of vo@xmath0 on a sapphire plate in a wide frequency region . in ref .
@xcite the dielectric permittivity of vo@xmath0 is measured and fitted to the oscillator model for both bulk vo@xmath0 and for a system of 100 nm thick vo@xmath0 film deposited on bulk sapphire plate within the frequency region from 0.25ev to 5ev .
this modelling was performed both before and after the phase transition .
typical thickness of sapphire substrate is of about 0.3 mm , i.e. , the same as the thickness of patterned si plate in secs .
ii and iii . because of this , when calculating the casimir force between gold coated sphere and vo@xmath0 film on sapphire substrate , we can use the lifshitz formala for bulk test bodies ( see sec .
iii for details ) .
the application region of the models presented in ref .
@xcite should be extended in order to perform computations of the casimir force within the separation region from 100 to 300 nm where contribution from optical data up to about 10ev have to be taken into account . for this purpose , we have supplemented equations of ref .
@xcite with additional terms taking into account the frequency - dependent electronic transitions at high frequencies @xcite . as a result
, the effective dielectric permittivity of the vo@xmath0 film on a sapphire substrate before the phase transition ( at @xmath8k ) is given by @xmath110 here the values of the oscillator frequencies @xmath111 , dimensionless relaxation parameters @xmath112 and of the oscillator strengths @xmath113 taken from fig .
5 in ref .
@xcite are presented in table i. the constants related to the contribution of high - frequency electronic transitions [ the last term on the right - hand side of eq .
( [ eq17 ] ) ] are @xmath114 @xcite and @xmath115ev . if we put @xmath116 in the last term on the right - hand side of eq .
( [ eq17 ] ) , this equation is the same as the result in ref .
@xcite . after the phase transition we have a phase with increased charge carrier density @xmath20 .
similar to eq .
( [ eq10 ] ) the effective dielectric permittivity of the vo@xmath0 film on a sapphire substrate can be described by the dielectric permittivity @xmath117 parameters @xmath118 , @xmath119 and @xmath120 can be found in fig .
6 of ref .
@xcite and are listed in table ii .
the other parameters are @xmath121 @xmath122ev , @xmath123ev @xcite . setting @xmath116 in the last term on the right - hand side of eq .
( [ eq18 ] ) ( this term takes high - frequency electronic transitions into account ) , returns ( [ eq18 ] ) to the original form suggested in ref .
note that the recently suggested model for the dielectric permittivity of vo@xmath0 films @xcite is applicable not only before and after a phase transition but also at intermediate temperatures .
this model is , however , restricted to a more narrow frequency region from 0.73 to 3.1ev and uses the simplified description of two oscillators before the phase transition and only one oscillator with nonzero frequency after it . in fig .
10 , the effective dielectric permittivity of vo@xmath0 film of 100 nm thickness on sapphire substrate before and after the phase transition , as given in eqs .
( [ eq17 ] ) and ( [ eq18 ] ) is shown by the solid lines 1 and 2 , respectively . in the same figure ,
the dielectric permittivity of au versus frequency is shown as dots .
the vertical line indicates the position of the first matsubara frequency at @xmath124k ( i.e. , in the region of the phase transition ) . in fig .
11 we present the computational results for the casimir force between the au coated sphere and vo@xmath0 film on sapphire substrate versus separation obtained by the substitution of the dielectric permittivity ( [ eq17 ] ) ( vo@xmath0 before the phase transition in solid line 1 ) and ( [ eq18 ] ) ( vo@xmath0 after the phase transition in solid line 2 ) into the lifshitz formula . as is seen in fig . 11 , after the phase transition the magnitudes of the casimir force increase due to an increase in the charge carrier density .
for a comparison with the proposed experiment on the difference casimir force from a patterned si plate , in fig . 12 ( solid line ) we plot the difference of the casimir forces after and before the phase transition , i.e. , the difference of lines 2 and 1 in fig . 11 .
it is seen that the difference casimir force from a phase transition changes from 13pn at @xmath89 nm to 1.2pn at @xmath56 nm , i.e. , the magnitudes of the difference from the phase transition are greater than that from the patterned si plate .
the difference casimir force in the insulator - metal phase transition provides us with one more test on the proper modelling of the dielectric permittivity in the lifshitz theory of dispersion forces .
similar to sec .
iii , we arrive at different results for the difference casimir force after and before the phase transition if the conductivity of a dielectric vo@xmath0 at zero frequency is taken into account in our computations .
the shift in the values of the difference casimir force is completely determined by the change of the zero - frequency term in the lifshitz formula . by analogy with eq .
( [ eq15 ] ) it follows : @xmath125\right\ } , \label{eq19}\ ] ] where @xmath79 represents the case when the dc conductivity of an insulating vo@xmath0 is taken into account , and @xmath70 represents the case when insulating vo@xmath0 is described by the permittivity ( [ eq17 ] ) . from eq .
( [ eq17 ] ) and table i one obtains @xmath126 in fig . 12 the difference casimir force between an au coated sphere and vo@xmath0 film on sapphire substrate after and before the phase transition computed including the dc conductivity of insulating vo@xmath0 versus separation is plotted with the dashed line .
the difference between the solid and dashed lines is determined by eq .
( [ eq19 ] ) .
this difference changes from 1.6pn at @xmath89 nm to 0.2pn at @xmath56 nm .
thus , in the phase transition experiment the predicted discrepances between the two theoretical approaches to the description of conductivity properties at low frequencies are larger than in the experiment with the patterned semiconductor plate .
this will help to experimentally discriminate between the two approaches and deeply probe the role of the material properties in the lifshitz theory at nonzero temperature .
in the above we have proposed two experiments on the measurement of the difference casimir force acting between a metal coated sphere and a semiconductor with different charge carrier densities .
one of these experiments is based on the formation of a special patterned si plate , two sections of which have charge carrier densities differing by several orders of magnitude .
the measurement scheme in this experiment is differential , i.e. , adapted for the direct measurement of the difference in the casimir forces between the sphere and each section of the patterned plate .
this allows one to obtain high precision within a wide measurement range . using the dynamic measurement technique
, this experiment also permits the measurement of the difference casimir pressure between two parallel plates one of which is coated with gold and the other is patterned and consists of two sections with different charge carrier densities . another proposed experiment directed to the same objective is novel and uses
the insulator - metal phase transition brought about by an increase of temperature in the measurements of the casimir force .
this transition also leads to the change of charge carrier density by several orders of magnitude while not requiring the formation of special patterned samples .
the expected difference in the casimir forces after and before the phase transition is even larger than in the experiment with the patterned si plate .
both proposed experiments are motivated by the uncertainties in the application of the theory of dispersion forces at nonzero temperature . as was shown above ,
different models of the conductivity of semiconductors at low frequencies used in the literature predict variations of the difference casimir force at the level of 1pn .
an even greater concern is that the model taking into account the dc conductivity of dielectrics violates the nernst heat theorem @xcite .
we have reported an apparatus developed at uc riverside that has the sensitivity of force measurements on the level of 1fn and is well adapted for the systematic investigation of the proposed effects in a wide range of separations .
this apparatus includes an interferometer based atomic force microscope operated in high vacuum over a temperature range from 360k to 4k .
the proposed experiments are feasible using the developed techniques and will aid in the resolution of theoretical problems on the application of the lifshitz theory at nonzero temperature to real materials .
another motivation of the proposed experiments is in the application to nanotechnology .
the separations between the adjacent surfaces in micro- and nanoelectromechanical devices are rapidly falling to a region below a micrometer where the casimir force becomes dominant . keeping in mind that semiconducting materials are used for micromachines , the detailed investigation of the dependence of the casimir force on the properties of semiconductors is important .
the proposed experiments and related theory clearly demonstrate that it is possible to control the casimir force with semiconductor surfaces by changing the charge carrier density with doping or excitation .
this opens new opportunities discussed above for using the casimir force in both the operation and function of novel nanomechanical devices .
in addition to the previously performed experiments on the casimir force ( see review @xcite and refs .
@xcite ) currently a number of new experiments have been proposed in the literature .
thus , ref .
@xcite proposes to measure the casimir torque between two parallel birefringent plates with in - plane optical anisotropy separated by either a vacuum or ethanol . in ref .
@xcite it is suggested to measure the vacuum torque between corrugated mirrors .
references @xcite propose the measurements of the casimir force between metallic surfaces at large separations of a few micrometers as a spherical lens and a plate , a cylinder and a plate or two parallel plates .
these experiments are aimed at resolving the theoretical problems arising in the lifshitz theory when it is applied to real metals . in refs .
@xcite a proposal to measure the influence of the casimir energy on the value of the critical magnetic field in the transition from a superconductor to a normal state has been made .
references @xcite proposed the measurement of the dynamic casimir effect resulting in the creation of photons .
the experiments proposed here on the difference casimir force through the use of patterned semiconductor samples or using the insulator - metal phase transition indicate important new promising directions for future investigations in the casimir effect .
g.l.k.and v.m.m . are grateful to the department of physics and astronomy of the university of california ( riverside ) for its kind hospitality .
the work on the difference casimir force with patterned semiconductor samples and insulator - metal phase transition was supported by the doe grant no .
de - fg02 - 04er46131 .
the development of the interferometer based low temperature afm was supported by the nsf grant no .
g . is grateful for the financial support of ucmexus and conacyt .
99 h. b. g. casimir , proc .
akad . wet . *
51 * , 793 ( 1948 ) .
e. m. lifshitz , zh .
teor . fiz .
* 29 * , 94 ( 1956 ) [ sov .
jetp * 2 * , 73 ( 1956 ) ] .
i. e. dzyaloshinskii , e. m. lifshitz , and l. p. pitaevskii , usp .
nauk * 73 * , 381 ( 1961 ) [ sov .
( usa ) * 4 * , 153 ( 1961 ) ] .
m. bordag , u. mohideen , and v. m. mostepanenko , phys . rep . *
353 * , 1 ( 2001 ) .
f. chen , u. mohideen , g. l. klimchitskaya , and v. m. mostepanenko , phys .
rev . lett . * 88 * , 101801 ( 2002 ) ; phys .
a * 66 * , 032113 ( 2002 ) .
g. bressi , g. carugno , r. onofrio , and g. ruoso , phys . rev
. lett . * 88 * , 041804 ( 2002 ) .
r. s. decca , e. fischbach , g. l. klimchitskaya , d. e. krause , d. lpez , and v. m. mostepanenko , phys .
d * 68 * , 116003 ( 2003 ) .
r. s. decca , d. lpez , e. fischbach , g. l. klimchitskaya , d. e. krause , and v. m. mostepanenko , ann .
( n.y . ) * 318 * , 37 ( 2005 ) .
r. s. decca , d. lpez , e. fischbach , g. l. klimchitskaya , d. e. krause , and v. m. mostepanenko , phys .
d * 75 * , 077101 ( 2007 ) .
m. bostrm and b. e. sernelius , phys .
* 84 * , 4757 ( 2000 ) . c. genet , a. lambrecht , and s. reynaud , phys .
a * 62 * , 012110 ( 2000 ) .
m. bordag , b. geyer , g. l. klimchitskaya , and v. m. mostepanenko , phys .
lett . * 85 * , 503 ( 2000 )
. b. geyer , g. l. klimchitskaya , and v. m. mostepanenko , phys .
a * 67 * , 062102 ( 2003 ) .
f. chen , g. l. klimchitskaya , u. mohideen , and v. m. mostepanenko , phys .
* 90 * , 160404 ( 2003 ) .
i. brevik , j. b. aarseth , j. s. hye , and k. a. milton , phys .
e * 71 * , 056101 ( 2005 ) . v. b. bezerra , r. s. decca , e. fischbach , b. geyer , g. l. klimchitskaya , d. e. krause , d. lpez , v. m. mostepanenko , and c. romero , phys .
e * 73 * , 028101 ( 2006 ) .
j. s. hye , i. brevik , j. b. aarseth , and k. a. milton , j. phys
. a : mat .
* 39 * , 6031 ( 2006 ) . y. srivastava , a. widom , and m. h. friedman , phys .
* 55 * , 2246 ( 1985 ) .
e. buks and m. l. roukes , phys .
b * 63 * , 033402 ( 2001 ) .
g. palasantzas and j. th .
m. dehosson , phys .
b * 72 * , 1213409 ( 2005 ) .
h. b. chan , v. a. aksyuk , r. n. kleiman , d. j. bishop , and f. capasso , science , * 291 * , 1941 ( 2001 ) ; phys .
87 * , 211801 ( 2001 ) .
r. e. slusher and c. weisbuch , solid state comm . * 92 * , 149 ( 1994 ) .
f. chen , u. mohideen , g. l. klimchitskaya , and v. m. mostepanenko , phys .
a * 72 * , 020101(r ) ( 2005 ) ; * 74 * , 022103 ( 2006 ) .
f. chen , g. l. klimchitskaya , v. m. mostepanenko , and u. mohideen , phys .
lett . * 97 * , 170402 ( 2006 ) .
f. chen , g. l. klimchitskaya , v. m. mostepanenko , and u. mohideen , optics express * 15 * , 4823 ( 2007 ) .
a. zylbersztejn and n. f. mott , phys .
b * 11 * , 4383 ( 1975 ) .
u. mohideen , w. s. hobson , s. j. pearton , r. e. slusher , and f. ren , appl .
. lett . * 64 * , 1911 ( 1994 ) .
u. mohideen , r. e. slusher , f. jahnke and s. koch , phys .
* 73 * , 1785 ( 1994 ) .
r. s. decca , d. lpez , h. b. chan , e. fischbach , d. e. krause , and c. r. jamell , phys .
* 94 * , 240401 ( 2005 ) .
h. j. mamin and d. rugar , appl .
* 79 * , 3358 ( 2001 ) .
j. blocki , j. randrup , w. j. swiatecki , and c. f. tsang , ann .
( n.y . ) * 105 * , 427 ( 1977 ) .
t. emig , r. l. jaffe , m. kardar , and a. scardicchio , phys .
rev . lett . * 96 * , 080403 ( 2006 ) .
a. bulgac , p. magierski , and a. wirzba , phys .
d * 73 * , 025007 ( 2006 ) .
m. bordag , phys .
d * 73 * , 125018 ( 2006 ) .
h. gies and k. klingmller , phys .
* 96 * , 220401 ( 2006 ) ; phys .
d * 74 * , 045002 ( 2006 ) .
d. e. krause , r. s. decca , d. lpez , and e. fischbach , phys .
lett . * 98 * , 050403 ( 2007 ) .
a. lambrecht , i. pirozhenko , l. duraffourg , and p. andreucci , eur .
. lett . * 77 * , 44006 ( 2007 ) .
j. d. jackson , _ classical electrodynamics _
( john wiley & sons , new york , 1999 ) .
a. o. caride , g. l. klimchitskaya , v. m. mostepanenko , and s. i. zanette phys . rev .
a * 71 * 042901 ( 2005 ) . ,
e. d. palik ( academic , new york , 1985 ) . t. vogel , g. dobel , e. holzhauer , h. salzmann , and a. theurer , appl . opt . * 31 * , 329 ( 1992 ) .
n. w. ashcroft and n. d. mermin , _ solid state physics _
( saunders colledge , philadelphia , 1976 ) . m. soltani , m. chaker , e. haddad , and r. v. kruzelecky , j. vac .
a * 22 * , 859 ( 2004 ) .
j. y. suh , r. lopez , l. c. feldman , and r. f. haglund , j. appl . phys . * 96 * , 1209 ( 2004 ) .
f. y. gan and p. laou , j. vac .
a * 22 * , 879 ( 2004 ) .
d. h. youn , h. t. kim , b. y. chae , y. j. hwang , j. w. lee , s. l. maeng , and k. y. kang , j. vac .
a * 22 * , 719 ( 2004 ) .
h. w. verleur , a. s. barker , and c.n .
berglund , phys . rev . * 172 * , 788 ( 1968 ) .
b. m. gorelov , k. p. kronin , v. v. koval , and v. m. ogenko , techn .
* 27 * , 157 ( 2001 ) .
v. a. parsegian , _ van der waals forces : a handbook for biologists , chemists , engineers , and physicists _ ( cambridge university press , cambridge , 2005 ) .
l. bergstrm , adv .
colloid interface sci . *
70 * , 125 ( 1997 ) .
h. kakiuchida , p. jin , s. nakao , and m. tazawa , jap .
phys . * 46 * , l113 ( 2007 ) .
b. geyer , g. l. klimchitskaya , and v. m. mostepanenko , phys .
d * 72 * , 085009 ( 2005 ) ; int .
a * 21 * , 5007 ( 2006 ) .
g. l. klimchitskaya , b. geyer , and v. m. mostepanenko , j. phys .
a : mat . gen . *
39 * , 6495 ( 2005 ) .
b. geyer , g. l. klimchitskaya , and v. m. mostepanenko , arxiv:0704.3818 ; ann .
( n.y . ) , 2007 , to appear
. j. n. munday , d. iannuzzi , yu .
barash , and f. capasso , phys .
a * 71 * , 042102 ( 2005 ) .
r. b. rodrigues , p.
a. maia neto , a. lambrecht , and s. reynaud , europhys . lett .
* 76 * , 822 ( 2006 ) . s. k. lamoreaux and w. t. buttler , phys .
e * 71 * , 036109 ( 2005 ) .
m. brown - hayes , d. a. r. dalvit , f. d. mazzitelli , w. j. kim , and r. onofrio , phys .
a * 72 * , 052102 ( 2005 ) .
antonini , g. bressi , g. carugno , g. galeazzi , g. messineo , and g. ruoso , new j. phys . *
8 * , 239 ( 2006 ) .
g. bimonte , e. calloni , g. esposito , l. milano , and l. rosa , phys .
94 * , 180402 ( 2005 ) .
g. bimonte , e. calloni , g. esposito , and l. rosa , nucl .
b * 726 * , 441 ( 2005 ) .
c. braggio , g. bressi , g. carugno , c. del noce , g. galeazzi , a. lombardi , a. palmieri , g. ruoso , and d. zanello , europhys .
lett . * 70 * , 754 ( 2005 ) .
kim , j. h. brownell , and r. onofrio , phys .
* 96 * , 200402 ( 2006 ) . + *
fig . 1 . *
( color online ) schematic diagram of the experimental setup for the measurement of the difference casimir force .
the patterned si plate with two sections of different dopant densities is mounted on a piezo below the au coated sphere attached to a cantilever of an atomic force microscope .
the piezo oscillates in the horizontal direction above different regions of the plate causing the flexing of the cantilever in response to the casimir force . *
fig . 2 .
* ( color online ) steps in the fabrication of the patterned si plate with patterned doping ( see text for more details ) . * fig .
( color online ) image of the interferometer based variable temperature atomic force microscope with the force sensitivity up to 1fn fabricated at uc , riverside .
the critical components are labeled .
4 . * the dielectric permittivity of si along the imaginary frequency axis for samples with high concentration of charge carriers @xmath68 and @xmath69 is shown by the solid lines 1 and 2 , respectively .
for the sample with a low concentration of charge carriers @xmath14 the permittivity versus frequency is shown by dashed lines @xmath70 and @xmath79 based on whether the static permittivity is finite or infinitely large .
the premittivity of au is indicated by the dotted line . *
the difference casimir forces versus separation in the case when the higher concentration of charge carriers is equal to @xmath68 and the sample with a lower concentration , @xmath14 , is described by a finite or infinitely large static permittivity are shown by the solid line 1@xmath70 and dashed line 1@xmath79 , respectively .
the analogous difference forces when the higher concentration of charge carriers is equal to @xmath69 are shown by the solid line 2@xmath70 and dashed line 2@xmath79 . * fig . 6 . * the difference casimir pressures versus separation in the case when the higher concentration of charge carriers is equal to @xmath68 and the sample with a lower concentration , @xmath14 , is described by a finite or infinitely large static permittivity are shown by the solid line 1@xmath70 and dashed line 1@xmath79 , respectively .
the analogous difference pressures when the higher concentration of charge carriers is equal to @xmath69 are shown by the solid line 2@xmath70 and dashed line 2@xmath79 . * fig . 7 .
* schematic of the experimental setup for the observation of modulation of the casimir force in an insulator - metal phase transition .
light from a chopped 980 nm laser heats a vo@xmath0 film leading to a phase transition to a state with higher concentration of charge carriers ( sapphire substrate is not shown ) .
cooling in between pulses causes the transition to a state with lower concentration of carriers .
the cantilever of an atomic force microscope flexes in response to the difference casimir force . *
fig . 8 . *
( color online ) preliminary results on the resistance of vo@xmath0 film grown at uc riverside as a function of temperature are shown as black squares ( heating ) and dots ( cooling ) . *
fig . 9 . *
( color online ) morphology of the same vo@xmath0 film , as in fig . 8 , grown by thermal evaporation . the heights of roughness peaks are of about 2 nm . *
fig . 10 .
* the effective dielectric permittivity of vo@xmath0 film on sapphire substrate along the imaginary frequency axis before and after the phase transition are shown by the solid lines 1 and 2 , respectively .
the permittivity of au is indicated by the dotted line . *
fig . 11 . *
the casimir force between an au coated sphere and vo@xmath0 film on sapphire substrate versus separation before and after the phase transition are shown by the solid lines 1 and 2 , respectively . *
fig . 12 . *
the difference of the casimir forces after and before the phase transition versus separation computed using a finite static dielectric permittivity ( solid line ) and taking into account the dc conductivity of vo@xmath0 in a dielectric state ( dashed line ) .
.values of the oscillator resonant frequencies @xmath111 , dimensionless relaxation parameters @xmath112 and oscillator strengths @xmath113 of vo@xmath0 film on sapphire substrate before the phase transition . [ cols="<,<,<,<",options="header " , ]
|
we propose two novel experiments on the measurement of the casimir force acting between a gold coated sphere and semiconductor plates with markedly different charge carrier densities . in the first of these experiments
a patterned si plate is used which consists of two sections of different dopant densities and oscillates in the horizontal direction below a sphere . the measurement scheme in this experiment is differential , i.e. , allows the direct high - precision measurement of the difference of the casimir forces between the sphere and sections of the patterned plate or the difference of the equivalent pressures between au and patterned parallel plates with static and dynamic techniques , respectively .
the second experiment proposes to measure the casimir force between the same sphere and a vo@xmath0 film which undergoes the insulator - metal phase transition with the increase of temperature .
we report the present status of the interferometer based variable temperature apparatus developed to perform both experiments and present the first results on the calibration and sensitivity .
the magnitudes of the casimir forces and pressures in the experimental configurations are calculated using different theoretical approaches to the description of optical and conductivity properties of semiconductors at low frequencies proposed in the literature .
it is shown that the suggested experiments will aid in the resolution of theoretical problems arising in the application of the lifshitz theory at nonzero temperature to real materials .
they will also open new opportunities in nanotechnology .
|
since the discovery of high - temperature superconductivity in the cuprates@xcite , many strongly correlated models have been intensively studied .
one of the simplest of these is the @xmath0-@xmath1 model@xcite : @xmath2 where @xmath3 is the electron operator defined in the no - double - occupancy subspace .
this model can be derived from the strong - coupling limit of the hubbard model . despite its simplicity and extensive study , the nature of the ground states of eq .
is still controversial .
a strong correlation view of the @xmath0-@xmath1 model was advanced by anderson , who conjectured the relevance of a resonating valence bond ( rvb ) state@xcite as a low energy state for eq . when doped . when undoped , the rvb state is a spin singlet , with no symmetry breaking , and describes a `` quantum spin liquid '' . at low temperature the mobile carriers in the doped rvb state
behave as bosons and condense , forming a state indistinguishable in terms of symmetry from a singlet bcs superconductor .
a further development was the introduction of a projected mean - field wavefunction the projection removing all components of the wavefunction with doubly occupied sites which could be used variationally@xcite presently , this variational method remains one of the few numerical tools to study @xmath0-@xmath1 like models which work directly at @xmath4 and can treat significant system sizes .
however , due to the special form of the variational wavefunction , one may be concerned about bias : very general states in the low energy subspace can not be investigated .
recently new numerical tools have been developed to investigate much more general low energy states in @xmath0-@xmath1 like models beyond the projective method .
one novel construction builds tensor product states ( tpss)@xcite , which can be conveniently studied and have been applied to many spin systems@xcite .
this new class of variational states do not assume any specifical ordering pattern a priori and can describe very general states as long as their entanglement entropy satisfy perimeter law .
recently , this method has been generalized to fermionic systems@xcite . among many different generalizations ,
the grassmann tensor product states ( gtpss)@xcite were shown to be closely related to projective states .
they are able to describe a class of projective wavefunctions faithfully , including in particular the short - range rvb states .
the tps and gtps construction is particularly well suited to trivalent lattices , the simplest of which is the honeycomb lattice in two dimensions . like the square lattice appropriate for the cuprates , the honeycomb lattice is bipartite and naturally supports an antiferromagnetic ( af ) state at half - filling in the strong coupling ( heisenberg ) limit .
similarities to cuprate physics may be expected . moreover ,
several numerical studies have identified a possible quantum spin liquid state on this lattice at half - filling when additional quantum fluctuations are included in the hubbard@xcite and heisenberg@xcite models .
thus the doped @xmath0-@xmath1 model on the honeycomb lattice seems a promising venue to explore rvb ideas . here
, we investigate the ground state of this system using the recently developed gtps approach .
the gtps results are benchmarked by comparison with exact diagonalization ( ed ) and density matrix renormalization group ( dmrg ) calculations .
our results are systematically consistent with these non - variational , exact methods on small clusters .
the principle result of the gtps calculations is that the ground state at non - zero doping is a time reversal symmetry breaking @xmath5 wave superconductor .
some physical rationale for this result are given at the end of this letter . and
@xmath6 , which contain @xmath7 are defined on the sublattices @xmath8 and @xmath9 for each unit cell .
the grassmann metric @xmath10 containing @xmath11 is defined on the links that connect the grassmann tensors @xmath12 and @xmath6 .
the blue lines represent the fermion parity even indices while the red lines represent the fermion parity odd indices of the virtual states .
notice an arrow from @xmath8 to @xmath9 represents the ordering convention @xmath13 that we use for the grassmann metric.,title="fig:",scaledwidth=35.0% ]
* the variational ansatz : * we use the standard form of gtps as our variational wavefunction . we further assume a translationally invariant ansatz , and thus is specified by just two different grassmann tensors @xmath14 on sublattice @xmath15 of each unit cell : @xmath16 with @xmath17 we notice the symbol @xmath18 means tensor contraction of the inner indices @xmath19 . here
@xmath20 are the grassmann numbers and dual grassmann numbers , respectively , defined on the link @xmath21 , which satisfy the grassmann algebra : @xmath22 as shown in fig.[fig : grassmanntps ] , @xmath23 are the virtual indices carrying a fermion parity @xmath24 . in this paper , we choose @xmath25 to be even and assume there are _ equal _ numbers of fermion parity even / odd indices , which might be not necessary in general . those indices with odd parity
are always associated with a grassmann number on the corresponding link and the metric @xmath26 is the grassmann generalization of the canonical delta function .
the complex coefficients @xmath27 and @xmath28 are the variational parameters . notice that @xmath29 is the physical index of the @xmath0-@xmath1 model on site @xmath30 , which can take three different values , @xmath31 and @xmath32 , representing the hole , spin - up electron and spin - down electron states .
we choose the hole representation in our calculations , and thus the hole state has an odd parity @xmath33 while the electron states have even parity @xmath34 .
on each site , the non - zero components of the grassmann tensors should satisfy the parity conservation constraint : @xmath35 since the wavefunction eq.([gtps ] ) does not have a definite fermion number , we use the grand canonical ensemble , adding a chemical potential term to eq.([tj ] ) to control the average hole concentration .
we use the imaginary time evolution method@xcite to update the gtps from a random state .
then we use the weighted grassmann - tensor - entanglement renormalization group ( wgterg ) method@xcite to calculate physical quantities.(see methods section for details . )
the total system size ranges up to @xmath36 sites and all calculations are performed with periodic boundary conditions .
the largest virtual dimension of the gtps considered was @xmath37 . to ensure convergence of the wterg method
, we kept @xmath38 ( defined in refs .
@xcite ) up to @xmath39 for @xmath40 , which gave relative errors for physical quantities of order @xmath41 . for @xmath42
, we kept @xmath38 up to @xmath43 .
* ground state energy and staggered magnetization : * at half - filling , the @xmath0-@xmath1 model reduces to the heisenberg model . in this case
, we find the converged ground state energies per site are @xmath44 for @xmath45 and @xmath46 for @xmath42(the term @xmath47 is subtracted here ) , which is consistent with previous tps study@xcite ( with virtual dimension @xmath48 and @xmath49 , since all the components of gtps with odd fermion parity virtual indices vanish in this case ) and a recent quantum monte carlo ( qmc ) result @xmath50 . despite good agreement with the ground state energy ,
the staggered magnetization @xmath51 obtained from our calculations is larger than the qmc result @xmath52 .
we find @xmath53 for @xmath45 and @xmath54 for @xmath42 , also consistent with the previous tps study@xcite .
nevertheless , we emphasize that the variational approach indeed obtains the correct phase .
actually , a recent study for square lattice heisenberg model shows @xmath51 can be consistent with the quantum monte carlo ( qmc ) result if @xmath25 is sufficient large@xcite .
much more interesting physics arises after we dope the system.(we consider @xmath55 ) . as seen in fig.[fig : energy ] , the ground state energy shows a marked increase in @xmath25 dependence with increasing hole doping @xmath56 . as a benchmark , we perform the ed and dmrg calculations for small periodic clusters with @xmath57 sites ( @xmath58 for ed and @xmath59 for dmrg ) .
these two methods are the only unbiased methods for frustrated systems that avoid the sign problem , but are restricted to relatively small systems .
to ensure the convergence of the dmrg , we keep up to @xmath60 states and make the truncation errors less than @xmath61 in our @xmath62 calculations . up to @xmath45 , we find a systematic convergence of the ground energy .
some data points around @xmath63 for @xmath42 have slightly higher energy than @xmath45 , because @xmath64 is still not large enough for convergence at @xmath42 . as shown in the insert of fig.[fig : energy ] , the staggered magnetization @xmath51 has an even larger @xmath25 dependence than energy at finite doping
. however , up to @xmath42 , the data appears to converge to a relatively well - defined curve indicating vanishing af order for @xmath65 .
* superconductivity : * next we turn to the interesting question of whether the doped antiferromagnetic mott insulator on the honeycomb lattice supports superconductivity , and if so , what its pairing symmetry is . to answer this
, we calculate the real space superconducting ( sc ) order parameters in the spin singlet channel @xmath66 , where @xmath30 and @xmath67 are nearest neighboring sites.because we use a chemical potential to control the hole concentration , the charge @xmath68 symmetry can be spontaneous broken in the variational approach , which allows @xmath69 to be directly measured rather than through its two - point correlation function . as shown in the main panel of fig.[fig : sc ] , up to @xmath70 we find a non - zero singlet sc order parameter for the whole region .
strikingly , we find that the sc state breaks time reversal symmetry . by measuring the sc order parameters for the three inequivalent nearest - neighbor bonds
, we found @xmath71 with @xmath72 .
this pairing symmetry is usually called @xmath5 wave . to exclude the possibility that this results from trapping in a unstable local minimum , we repeated the calculations with many different random tensors , and all cases converged to the same results .
moreover , we also checked that the sc order parameter vanishes at large @xmath73 to make sure that the existing of sc order is consequence of spontaneous symmetry breaking.(for @xmath74 , the critical value is around @xmath75 at @xmath76 ) the existence of sc order is observed in our numerical study up to @xmath77 .
however , a much larger inner dimension @xmath25 is required for the convergence of ground state energy at larger hole concentration , which is beyond the scope of this paper.(the gtps variational ansatz we use in this paper is designed to help us understand the nature of mott physics ; at large doping , the mott physics becomes less important and can be studied much better by other methods . ) * coexisting phases at low doping : * interestingly , we find the sc and af order coexisting in the regime @xmath78 .
a physical consequence of the microscopic coexistance is that triplet pairing is induced .
the insert of fig.[fig : sc ] shows the amplitude of the triplet order parameter as a function of doping .
since the triplet pairing order parameter has three independent components @xmath79(here @xmath80 , @xmath81 and @xmath82 is the phase of sc order parameter . )
, we can define the amplitude of triplet order parameter as @xmath83 .
the phase shift of @xmath84 on the three inequivalent bonds is also observed for all the triplet components .
we further check the internal spin direction of the triplet @xmath85 vector and find it is always anti - parallel to the neel vector(@xmath86 ) . at larger doping @xmath87 ,
the triplet order parameter has a very strong @xmath25 dependence , so at present we are unable to determine whether it ultimately vanishes or remains non - zero in the @xmath88 limit .
we leave this issue for future work . by fully using all symmetry quantum numbers and other techniques like high performance simulation on gpus ,
we can in principle handle @xmath25 up to @xmath89 .
the calculations in this paper by construction ignore any possible phases breaking translational symmetry .
natural candidates are spiral or striped anti - ferromagnetic phases , as have been heavily discussed for the square lattice .
a weak coupling perspective suggests this may be unlikely . in the weak - coupling limit of hubbard model on honeycomb lattice ,
the system is a semimetal with two dirac cones . with a non - zero hubbard @xmath90 , commensurate af fluctuations at zero momentum manifest in the inter - band susceptibility , becoming stronger and stronger with increasing @xmath90 . at sufficiently large u
commensurate af order develops ( recent numerics find a narrow region of intermediate spin liquid phase@xcite ) . at small but finite doping
, the dirac cones become pockets . in this case , the total intra - band spin susceptibility shows a constant behavior for small @xmath91(@xmath92)@xcite while the total inter - band spin susceptibility still peaks at zero momentum .
thus , we argue the commensurate af fluctuations at zero momentum still dominate for sufficient small hole concentration . .
, title="fig:",scaledwidth=45.0% ] our dmrg and ed calculations also support this argument . up to 54 sites
, we do not find any evidence of incommensurate spin - spin correlation .
fig.[fig : compare ] shows the comparison of dmrg calculations and gtps calculations for staggered magnetization @xmath51 on small systems .
we find the gtps results are comparable with extrapolations of @xmath51 for infinite size systems ( the gtps results for @xmath51 are somewhat larger close to half - filling due to insufficient tensor dimension @xmath25 ) .
this agreement also implies our variational wavefunction is very close to the true ground state .
a more general concern is whether the gtps tends to overestimate sc order at large doping , due to its non - conservation of charge(note that the projected wavefunction approach has a rather strong tendency to produce superconducting states ) .
the observed sc order parameter is `` small '' in terms of the natural upper limit @xmath93 .
nevertheless , our results are consistent with other approaches : ( a ) in the mean field theory for the honeycomb @xmath94 model , only the @xmath5 pairing channel gains energy@xcite , and ( b ) in the weak coupling limit of hubbard model , very recent renormalization group studies also find @xmath5 superconductivity around quarter filling@xcite . in conclusion , we report the theoretical discovery of a @xmath5 wave superconducting state in @xmath0-@xmath1 model on a honeycomb lattice , based on a recently developed variational method - the gtps approach . at low doping , af ordering coexists with the sc ordering . in the coexistance regime ,
a spin triplet pairing with the same phase shift is induced and its triplet @xmath95 vector is anti - parallel with the neel vector
. it would be interesting to search for this physics in experiment .
the recently discovered spin @xmath96 honeycomb lattice antiferromagnet inv@xmath97cu@xmath98o@xmath99@xcite would be an appealing candidate if it could be doped .
* the imaginary time evolution algorithm of gtps : * in this paper , we use the ( simplified ) imaginary time evolution method@xcite to update the gtps variational wave function . in the hole representation
, we can decompose the @xmath3 as @xmath100 . here
the holon @xmath101 is a fermion while the spinon @xmath102 is a boson .
the no - double - occupancy constraint reads : @xmath103 under this representation , we can rewrite the @xmath94 model as : @xmath104 where @xmath105 due to the no - double - occupancy constraint eq .
( [ constraint ] ) , the spin up / down states @xmath106 and the hole state @xmath107 form a complete basis for each site .
the closure condition reads : @xmath108 as already has been discussed in ref.@xcite , we need to use the fermion coherent state representation to perform the imaginary time evolution algorithm for gtps .
let us introduce the fermion coherent state of holon @xmath109.(@xmath110 is a grassmann variable here . ) in this new basis , the closure relation eq .
( [ closure ] ) becomes : @xmath111 the variational ground state can be determined through imaginary time evolution : @xmath112 for sufficient thin time slice @xmath113 , we can decompose @xmath114 as : @xmath115 here @xmath116 represent three different directions of the honeycomb lattice and each @xmath117 term contains a summation of non - overlapped two body hamiltonian @xmath118 . thus , for sufficient thin time slice , we can apply the evolution operator along @xmath119 direction separately . in the fermion coherent state representation , the two body evolution operator along @xmath120 direction @xmath121 can be expressed as a grassmann valued matrix acting on the two sites hilbert space : @xmath122 here @xmath123 is the fermion coherent states indices valued as @xmath124 , @xmath32 or @xmath125 while @xmath126 is the usual fock state indices valued as @xmath124 , @xmath32 or @xmath127 .
we notice the spin states remain the same in both representations . by applying the method developed in ref.@xcite
, we can successfully update the ( complex ) variational parameters @xmath27 and @xmath28 as : @xmath128 where @xmath90 and @xmath129 are determined by the singular value decomposition(svd ) of the following matrix @xmath130 : @xmath131p^f(a ) } ( -)^{p^f(m_j^\prime)\left[p^f(m_i^\prime)+p^f(b)+p^f(c)\right ] } \nonumber\\ & \times ( -)^{m_j\left[p^f(m_i)+p^f(b)+p^f(c)\right ] } e_{m_i^\prime m_j^\prime}^{m_i m_j } { t}^{m_{i}^\prime}_{a;abc}{t}^{m_{j}^\prime}_{b;a b^\prime c^\prime}\label{m}\end{aligned}\ ] ] we keep the largest @xmath25th singular values : @xmath132 similar as in the usual tps case@xcite , the environment weight vectors @xmath133 can be initialized as @xmath134 and then updated during the time evolution .
for example , @xmath135 is updated as @xmath136 in the above evolution scheme . *
the gterg / wgterg algorithm * : after determining the variational ground state of gtps by performing the imaginary time evolution , we can compute the physical quantities by using the gterg / wgterg method developed in ref@xcite .
this method can be regarded as the grassmann variable generalization of the usual terg / wterg method@xcite , where a coarse graining procedure is designed to calculate the physical quantities of tps efficiently . in fig.[fig : error ] , we plot the variational ground state energy(@xmath45 ) as a function of doping with different @xmath38(number of eigenvalues kept in the wgterg algorithm ) .
we find the relative error is of order @xmath41 . to mimic the environment effect .
as has been discussed in ref.@xcite , the initial value of @xmath137 can be determined by @xmath138 on the corresponding link and will be updated during the rg scheme.,title="fig:",scaledwidth=50.0% ] ) as a function of doping with different @xmath38(number of eigenvalues kept in the wgterg algorithm ) .
it is shown all the data points almost collapse on the same curve .
we find the relative error is of order @xmath41 .
, title="fig:",scaledwidth=50.0% ]
the authors would like to thank f. verstraete , j. i. cirac , m. p. a. fisher , f. c. zhang , p. a. lee , z. y. weng , l. fu , k. sun , c. wu , f. yang and y. zhou for valuable discussions .
z.c.g . is supported by nsf grant no .
phy05 - 51164 ; d.n.s is supported by dmr-0906816 ; l.b . is supported by nsf grant dmr-0804564 and a packard fellowship ; h.y .
was partly supported by doe grant de - ac02 - 05ch11231 ; x.g.w . is supported by nsf grant no .
dmr-1005541 and nsfc 11074140 .
|
we report the theoretical discovery of a novel time reversal symmetry breaking superconducting state in the @xmath0-@xmath1 model on the honeycomb lattice , based on a recently developed variational method - the grassmann tensor product state approach . as a benchmark ,
we use exact diagonalization ( ed ) and density matrix renormalization ( dmrg ) methods to check our results on small clusters .
remarkably , we find systematical consistency for the ground state energy as well as other physical quantities , such as the staggered magnetization . at low doping ,
the superconductivity coexists with anti - ferromagnetic ordering .
|
quantum entanglement is a fundamental resource for many quantum information processing tasks , e.g. super - dense coding , quantum cryptography and quantum error correction [ 1 - 4 ] .
entangled states can be used in constructing number of protocols , e.g. teleportation [ 5 ] , key distribution and quantum computation [ 6 ] . during recent past ,
entanglement sudden death ( esd ) has been investigated by different authors for bipartite and multipartite states [ 7 - 10 ] .
yu and eberly [ 11 , 12 ] have shown that entanglement loss occurs in a finite time under the action of pure vacuum noise in a bipartite qubit system .
a geometric interpretation of the phenomenon of esd has been given in ref .
furthermore , experimental evidences of esd have been reported for optical setups [ 14 , 15 ] and atomic ensembles [ 16 ] .
peres - horodecki [ 17 , 18 ] have studied entanglement of qubit - qubit and qubit - qutrit states and established separability criterion . according to this criterion , the partial transpose of a separable density matrix must have non - negative eigenvalues . for non - separable states ,
the sum of the absolute values of the negative eigenvalues of the partial transpose matrix gives the degree of entanglement of a density matrix also termed as negativity .
ann et al .
[ 19 ] have studied a qubit - qutrit system where they have shown the existence of esd under the influence of dephasing noise .
bipartite entangled states are divided into free - entangled states and bound - entangled states [ 20 , 21 ] .
free - entangled states can be distilled under local operations and classical communication ( locc ) whereas bound - entangled states can not be distilled to pure - state entanglement irrespective of the number of copies of the initial state available .
wang et al . proposed that free - entangled states may be converted into bound - entangled states under multi - local [ 22 ] and collective dephasing processes [ 23 ] .
they have shown that certain free - entangled states of qutrit - qutrit systems become non - distillable in a finite time under the influence of classical noise . for a qutrit - qutrit system
there are many bound entangled states and no single criterion can fully describe all of them [ 24 ] .
however , the realignment criterion can be used to detect certain bound entangled states [ 25 ] . since , it is not possible to completely isolate a quantum system from its environment .
therefore , one needs to investigate the behavior of entanglement in the presence of environmental effects .
a major problem of quantum communication is to faithfully transmit unknown quantum states through a noisy quantum channel .
when quantum information is sent through a channel , the carriers of the information interact with the channel and get entangled with its many degrees of freedom . in order to quantify entanglement ,
two measures are the negativity [ 26 ] , a measure of a state having negative partial transpose and the realignment criterion [ 25 ] .
the negativity is equal to the absolute value of the sum of negative eigenvalues of partial transpose of a state .
when the negativity becomes zero , we need to study the time evolution of realignment criterion to determine the existence of bound - entangled states .
however , it should be kept in mind that realignment criterion could not detect all entangled states . in this paper ,
the effect of flipping noise on the entanglement dynamics of a qutrit - qutrit system under the influence of global , local and multilocal decoherence is investigated by considering phase flip , trit flip and trit phase flip channels .
it is shown that the entanglement sudden death can be avoided in the presence of phase flip , trit flip and trit phase flip environments .
it is also shown that particular local operations can not avoid the non - distillability of the distillable states .
further more , no esd and dsd is seen for trit flip channel in the presence of global noise .
let us consider that the initial state is interacting with the noisy environment both collectively and individually .
local and multi - local couplings describe the situation when both the qutrits are independently influenced by their individual noisy environments . whereas , the global decoherence corresponds to the situation when it is influenced by both local and multilocal noises at the same time .
the state shared by the two parties is an entangled qutrit - qutrit state of the form@xmath0where @xmath1 , @xmath2 is a maximally entangled bipartite qutrit state and @xmath3 are separable states given as @xmath4 , @xmath5 the above density matrix is separable for @xmath6 , bound entangled for @xmath7 and free entangled for @xmath8 [ 27 ] .
the interaction between the system and its environment introduces the decoherence to the system , which is a process of the undesired correlation between the system and the environment .
the dynamics of the composite system in the presence of flipping noise can be described in kraus operators formalism [ 29 ] @xmath9 where the kraus operators @xmath10 satisfy the following completeness relation @xmath11 the single qutrit kraus operators for phase flip channel can be written as @xmath12 and the single qutrit kraus operators for the trit flip channel are given as @xmath13 whereas , the single qutrit kraus operators for the trit phase flip channel are given by @xmath14 where @xmath15 represents the quantum noise parameter usually termed as decoherence parameter . here
the bounds [ 0 , 1 ] of @xmath16 correspond to @xmath17 , @xmath18 respectively .
the evolution of the initial density matrix of the composite system when it is influenced by local and multi - local environments is given in kraus operator form as @xmath19and the evolution of the system when it is influenced by global environment is given in kraus operator representation as @xmath20where @xmath21 @xmath22 are the kraus operators of the multilocal couplings of the individual qutrits and @xmath23 are the kraus operators of the collective coupling of the qutrit system . using equations ( 2 - 5 ) along with the initial density matrix of as given in equation ( 1 ) and taking the partial transpose over the second qutrit , the eigenvalues of the final density matrix can be easily found .
let the decoherence parameters for local and global noise be @xmath24 , @xmath25 and @xmath16 respectively .
the eigenvalues of the partial transpose matrix , when both qutrits are influenced by multi - local noise of the phase flip channel , are given as @xmath26 \\
\lambda _ { 7,8,9 } & = & \frac{1}{42}[5+\left .
\sqrt{% \begin{array}{c } 4\text{$\alpha $ } ^{2}-20\text{$\alpha $ } + 16\text{$p_{1}$}^{2}(\text{$p_{2}$}% -1)^{2 } \\ -32\text{$p_{1}$}(\text{$p_{2}$}-1)^{2}+16\text{$p_{2}$}^{2}-32\text{$p_{2}$}% + 41% \end{array}% } \right . ]
\end{aligned}\]]and the realignment criterion in this case reads@xmath27the expressions for local noise can be obtained by setting @xmath28 .
the eigenvalues of the partial transpose matrix , when both the qutrits are influenced by the global noise of phase flip channel , are given by@xmath29and the realignment criterion in this case reads@xmath30the eigenvalues of the partial transpose matrix when both the qutrits are influenced by the trit flip and trit phase flip channels are not provided with their analytical relations as these are too lengthy in size .
therefore , they are just interpreted in the graphs .
the entanglement for all mixed states @xmath31 can be quantified by the negativity@xmath32where @xmath33 represents the negative eigenvalues of the partial transpose of the density matrix @xmath31 .
in this work , the effect of decoherence on a qutrit - qutrit system is investigated . in figure 1 , the negativity and realignment criterion are plotted as a function of decoherence parameter @xmath16 for local noise introduced through phase flip , trit flip and trit phase flip channels .
it is seen that the partial transpose criterion fails to detect the behavior of entanglement in density matrices of larger @xmath34 for higher values of decoherence .
the quantity @xmath35 is positive for @xmath36 which shows that these density matrices are bound entangled .
however for @xmath37 , the realignment criterion also fails to detect the possible entanglement for these states . in figure 2 , negativity and realignment criterion
are plotted as a function of decoherence parameter @xmath16 for multi - local noise introduced through phase flip , trit flip and trit phase flip channels .
it can be seen that no esd occurs in any density matrix as the negativity is positive for all the density matrices . however , for @xmath38 , some states that correspond to large values of @xmath34 become positive partial transpose states ( ppt ) . in figure 3 , negativity and realignment criterion
are plotted as a function of decoherence parameter @xmath16 for global noise introduced through phase flip , trit flip and trit phase flip channels .
it is seen that the free entangled distillable states convert into bound entangled or separable states and therefore become completely non - distillable in the presence of global noise . in figure 4 ,
three - dimensional graphs for negativity are plotted as a function of decoherence parameter @xmath16 and parameter @xmath34 for multi - local and global noises introduced through phase flip , trit flip and trit phase flip channels .
it is seen that esd and dsd can be completely avoided for the case of trit flip channel . in order to see the effect of local unitary operation on the dynamics of qutrit - qutrit state ,
let the unitary operator @xmath39 = @xmath40 , with @xmath41 . by applying this operator locally to the state ( equation 1 )
, it converts the maximally entangled state @xmath42 into another maximally entangled state given by @xmath43 and the two separable states @xmath44 are transformed to @xmath45 , @xmath46 the final density matrix takes the form @xmath47the evolution of @xmath48 in the presence of flipping noise and its partial transpose can be found in a similar fashion as calculated for equation ( 5 ) .
it is seen that the eigenvalues for each case remains unchanged .
therefore , the local operation does not change the behavior of entanglement in the presence of flipping noise .
entanglement dynamics of a qutrit - qutrit system under the influence of global , local and multilocal flipping noise is investigated . in order to quantify the entanglement of the system ,
the negativity and realignment criterion are used .
it is shown that the entanglement sudden death and distibility sudden death can be avoided in the presence of phase flip , trit flip and trit phase flip environments .
it is shown that certain free entangled distillable qutrit states become bound entangled or separable i.e. convert into non - distillable states under different flipping noises .
it is also seen that local operations do not have any effect on the entanglement dynamics of the system .
further more , no esd and dsd is seen for the case of trit flip channel .
figures captions**figure 1**. ( color online ) .
the negativity and realignment criterion are plotted as a function of decoherence parameter @xmath16 for local noise introduced through phase flip , trit flip and trit phase flip channels.*figure 2*. ( color online ) .
the negativity and realignment criterion are plotted as a function of decoherence parameter @xmath16 for multi - local noise introduced through phase flip , trit flip and trit phase flip channels.*figure 3*. ( color online ) .
the negativity and realignment criterion are plotted as a function of decoherence parameter @xmath16 for global noise introduced through phase flip , trit flip and trit phase flip channels.*figure 4*. ( color online ) .
the negativity is plotted as a function of decoherence parameter @xmath16 and parameter @xmath34 for multi - local and global noises introduced through phase flip , trit flip and trit phase flip channels .
|
entanglement dynamics of a qutrit - qutrit system under the influence of global , local and multilocal decoherence introduced by phase flip , trit flip and trit phase flip channels is investigated .
the negativity and realignment criterion are used to quantify the entanglement of the system .
it is shown that the entanglement sudden death and distibility sudden death can be avoided in the presence of phase flip , trit flip and trit - phase flip environments .
it is shown that certain free entangled distillable qutrit states become bound entangled or separable i.e. convert into non - distillable states under different flipping noises .
it is also seen that local operations do not have any effect on the entanglement dynamics of the system .
further more , no esd and dsd is seen for the case of trit flip channel .
keywords : quantum channels ; qutrit entanglement ; global noise .
|
the davies - unruh effect ( dhe ) @xcite , essentially predict that in an accelerated frame of reference ; a vacuum state may seen as a thermal bath of photons with a black boddy spectrum at a temperature t , the main point of the due is that this temperature is proportional to the acceleration of the frame .
the connection between thermodynamic and gravity began in the 70s with bekenstein @xcite and hawking @xcite , researching the nature of black holes . in 1995
jacobson @xciteshown a thermodynamic description of gravity obtaining the einstein s equations .
according to padmanabhan @xcite , the association between gravity and entropy leads in a natural way to describes gravity as an emergent phenomenon , and a formalism of gravity as a entropic force is derived by verlinde @xcite in 2010 .
the dependence of information on surface area , rather than volume ( holographic principle ) @xcite , it is one of the key of black hole thermodynamic theory , as well as in egt .
on the other hand , the tully - fisher relation @xcite is an empirical result , very well established for spiral galaxies .
this relation is hard to be obtained from newtonian gravity , at least if only the visible mass of the galaxy is considered .
the output for this impasse was postulating the presence of a galactic halo of dark matter . nowadays
the empirical roots of the missing mass problem goes from the flat rotation curves of galaxies , cluster of galaxies , gravitational lensing , large scale structure , and it is needed to describe the spectrum of the cosmic microwave background radiation ( cmb ) .
the nature of dark matter is unknown .
but the most widely accepted hypothesis is that dark matter is composed of weakly interacting massive particles ( wimps ) that interact only through gravity and the weak force @xcite .
however , so far , there is no direct evidence of wimps or other dark particles such as the axions , and only upper limits were reported @xcite .
in addition , so far , there is no evidence of a new physics beyond standard model , in the large hadron collider ( lhc ) data @xcite .
there is no evidence for susy ( super - symmetric particles ) @xcite , where the `` neutralino '' is a kind of natural candidate for a dark matter particle . in the 80s , milgrom @xcite ,
proposed a modification in the newtonian law of gravity as solution to the missing mass problem , without dark matter . rather than a scientific theory mond
is considered only as an empirical model by most of the scientific community , because predicts a violation of the equivalence principle .
however , mond has been very well successful to describe the galaxies dynamics @xcite .
indeed , mond predicted the tully - fisher relation .
the connection between entropic gravity and mond is not new , there are some literature on this topic such as reported in @xcite . in this letter
we gives emphasis to the formalism of debye model @xcite of the specific heat of solid at low temperatures , incorporated to the entropic gravity . in section
ii , the basis of egt within the debye formalism is presented and we defined the debye temperature , @xmath0 , in egt .
section iii , is devoted to a analysis of the inertia at low temperatures ( @xmath1 ) . in the intermediate region ,
i.e. , temperatures close to @xmath0 , egt seems to indicate that is the galaxy clusters region , this topic is discussed in section iv , and the section v is devoted for our conclusions .
in 1912 , debye @xcite developed a theory to explain the heat capacity of solid as low temperatures .
he assumed that the vibration of the atoms of the lattice of a solid , follows a continuous range of frequencies , such as an elastic structure , that cuts off at a maximum frequency , @xmath2 . in this theory
each solid has a specific temperature , called as debye temperature , @xmath3 .
the debye model correctly predicts the low temperature dependence of the heat capacity of solid and coincides with the approaching the dulong - petit law at high temperatures . in egt
an holographic screen is the closed area where is stored the information of the surrounding matter enclosed by the screen .
the information is codified in n bits and is considered as the freedom degrees of the system .
the holographic screens coincides with the newtonian equipotential surfaces .
a central argument of the holographic principle is consider that each bit of information on the screen carries an energy @xmath4 and the number of bits n on the screen surface is proportional to the area of the screen , and expressed as @xmath5 , where g is newton gravitational constant . taking into account the equipartition of energy principle , the specific potential energy on the screen can be written as @xmath6 following the analogy with debye model , we have substituted @xmath7 , for @xmath8 .
the main difference with the debye theory is in that the third debye function , @xmath9 was replaced by the first debye function , @xmath10 , because the information bits located on the screen have only a vibrational state along of the gradient of newton potential and it is assumes that the vibrations follows a continuous range of frequencies .
@xmath11 is defined as @xmath12 the shape of debye function reflects the bose - einstein statistic formula , used in its derivation .
if m represents the all mass enclosed by the screen surface , the specific potential energy can be written as @xmath13 , and considering that the entropy variation , @xmath14 of the screen , happens when a particle of mass m is at a distance @xmath15 ( close to the compton wave length ) .
the bekestain entropy variation can be expressed as @xmath16 these relations , allows to obtain the entropic force defined as @xmath17 .
the more simple case is for a spherical screen of radius r , ( @xmath18 ) , and the acceleration of the mass m is @xmath19 considering that @xmath20 ( see below ) , the debye function can be parametrized by a power function of type @xmath21 , as shows in fig . 1 and eq.4 becomes @xmath22 the two asymptotically cases are : ( a ) @xmath23 and means high temperature regime @xmath24 and the eq.5 coincides with the newtonian gravity theory , and ( b ) @xmath25 means low temperature regime @xmath26 and
eq.5 reproduce the tully fisher relation @xmath27 observed in the galaxies dynamic and plotted as @xmath28 @xcite .
the slope , 4 , fall precisely with that observed in galaxies , whereas the normalization require @xmath29 @xcite .
the acceleration @xmath30 is the milgrom acceleration parameter @xcite and in this limit egt coincides with the mond theory . in this limit
the debye first function ( eq.2 ) can be related as @xmath31 with @xmath32 and lead to @xmath33 .
finally the intermediate region , @xmath34 , egt seems to indicate that is the galaxy cluster region , see section iv .
the starting point for development of the general theory of relativity was the equivalence principle , it is also valid in the newtonian gravity .
there are strong evidences indicating that the equivalence principle holds in all experiments at earth@xcite .
we starting the analysis , taken into account the bekenstein entropy variation expressed in the eq .
3 , and the entropic force concept @xmath35 to obtain @xmath36 m_gc/\hbar.\ ] ] according to the debye framework , the quantity , @xmath7 was substituted by @xmath37 . in eq.6 , f in the left side represent the force of inertia , that can be expressed as @xmath38 , where @xmath39 is the inertia mass , while the mass , m , in the right side of the equation , represent the mass of the particle , at a distance equal to compton wavelength of the holographic screen , when the entropic force emerges , then it is linked with the gravitational mass , @xmath40 .
here we have two possibility : \(a ) if the equivalence principle holds for all temperature regions , we have @xmath41 and eq .
6 becomes @xmath42 this equation we called as generalized due , at high temperatures , @xmath43 , @xmath44 , and coincides with the conventional due .
2 shows a comparison between the generalized and conventional due . the discrepancy for @xmath26 can emulate dark matter . considering that @xmath45 and @xmath46 for dark matter and calling as @xmath47 and @xmath48 the accelerations according to the modified and conventional due
, we have @xmath49 and multiplying this last expression by a , we have @xmath50 for high temperatures @xmath24 , @xmath51 and @xmath52 .
\(b ) if the equivalence principle is violated @xmath53 and keeping the due without modification eq.6 becomes @xmath54 taking in account eq .
5 @xmath55 , the ratio @xmath56 is plotted in fig .
3 as function of acceleration . following the fig.3 , we can see that the ratio @xmath57 for low accelerations increases as the acceleration decreases , the extreme case ( strong violation ) coincides with the mond theory prediction . while , for high accelerations ( temperatures ) , @xmath58 , and it includes all solar system , the equivalence principle remains valid .
the fact that @xmath53 for motions in low temperatures regimen , can emulate dark matter .
if we called as @xmath59 the dark matter mass , it can be obtained as @xmath60 and taking into account eq.8 , we obtain @xmath61 this expression coincides with eq . 8 , and means that we have a duality . however , in both cases , the central mechanism is the due .
this suggested that the modification of due is more fundamental in egt and the emulated dark matter is seen as an apparent violation of the equivalence principle .
it is well known also that the mond theory has its `` achilles heel '' . the galaxy cluster , seems to indicate that still is necessary a residual mass . in most cases
the mond critics largely use this to reject mond .
indeed , the residual mass required by mond was supply with an exotic neutrino , the `` sterile neutrino '' , considered as promising candidates to hot dark matter @xcite .
so far , there is no direct evidence of these neutrinos @xcite . on the other hand , in the central part of clusters the observed acceleration is usually slightly larger than @xmath30 @xcite .
this clearly shows the limitations of mond in cluster analysis .
however , there is not this limitation in egt and means that the temperature of the holographic screen that surrounding clusters enclosed by the screen has a temperature slightly larger than the debye temperature .
this means that the clusters analysis requires @xmath34 .
the relationship between various galaxy cluster mass estimators and x - ray gas temperature agree to within 40% @xcite .
virial theorem mass estimates based on cluster galaxy velocity dispersions seem to be accurately related to the x - ray temperature as @xmath62 with @xmath63 @xcite .
this results are consistent with that predicted by simulations @xcite
. however , when is combined several independent observation , i.e. , a wide range of temperatures of galaxy clusters , seems to indicate a steeper @xmath64 index @xcite .
even so , in all cases , the greatest discrepancy is in normalization , with differences around 40% to 50% .
an analysis of galaxy cluster on the basis of entropic gravity is presented in @xcite . however , in this section we present an alternative straightforward analysis on galaxy clusters , on the basis of egt within the debye formalism .
cdm @xcite .
the solid line indicates the prediction of mond @xcite , ( @xmath25 ) , and the red dot line is the prediction of egt within the debye formalism to @xmath65 .
, width=415,height=340 ] under the assumption of spherical symmetry and following the eq .
5 , the asymptotic @xmath66 , allow us calculate the mass of cluster as @xmath67 the acceleration in the gravitational potential of the cluster is related by @xmath68 , with @xmath69 .
the density distribution @xmath70 is well described by the
@xmath71 , whose asymptotic expression is @xmath72 , where @xmath73 has a typical value of @xmath74 @xcite . under these conditions the eq .
11 becomes @xmath75 for @xmath25 we have the mond prediction to the mass - xray temperature relation of clusters , expressed as @xmath76 .
this relation is represented by the black solid line in fig .
4 @xcite , where the mass function of an x - ray flux of several samples of galaxy clusters are plotted .
we can see that data is closer with mond s predicted slope than that previsioned by standard @xmath77cdm and expressed as @xmath78 @xcite and represented by the dashed black line in fig .
4 , and it is better than mond only at high x - ray temperatures .
but the previsioned mond s normalization is around two times greater than observed .
the red dots line in fig .
4 , represent the prediction of eq . 12 , for @xmath65 and expressed as @xmath79 with @xmath80 .
the normalization for this case differs from mond normalization by a factor @xmath81 , for @xmath65 and @xmath82 .
egt within the debye formalism can emulate dark matter of two equivalent ways , the first case requires modification of the due in order to maintain the equivalence principle and the second one requires modification of equivalence principle in order to maintain the due . in both cases
the due is the fundamental mechanism .
this means that the modification of the due emulate dark matter which in turn emulates a violation of the equivalence principle . this scheme is promising , because recently results @xcite , on the sparc database of galaxies @xcite , seems indicate a challenge to the dark matter hypothesis .
the egt within the debye formalism is also promising in the analysis of galaxy clusters .
however , there is also a second dark entity to consider , the dark energy , responsible of the accelerating expansion of universe .
the dark energy in egt is discussed in @xcite .
even so , there is more a complication , the increase in a factor of 10 of the number of supernovae ia , in relation to the first analyses , seems to indicate that the accelerated expansion signal is only marginal @xcite .
p. c. w. davies , j. phys .
a , 8 , 4 ( 1775 ) .
w. g. unruh , phys .
d. 14 , 870 ( 1976 ) .
j. d. bekenstein , phys .
d , 7 , 2333 ( 1973 ) . s. w. hawking , nature , 248 , 30 ( 1974 ) .
t. jacobson , phys .
75 , 1260 ( 1995 ) .
t. padmanabhan , mod .
a , 30 , 1540007 ( 2015 ) .
e. verlinde , jhep 1104:029 ( 2011 ) .
g. t hooft , pre - print arxiv : gr - qc/9310026v2 r. b. tully and j. r. fisher , astron .
, 54 , 661 ( 1977 ) .
e. behnke , et al .
106 , 021303 ( 2011 ) .
d. s. akerib et al .
( lux collaboration ) , phys .
116 , 161301 ( 2016 ) .
a. tan et al .
( panda ii collaboration ) , phys .
117 , 121303 ( 2016 ) .
e. gibney , nature 536 , 133 ( 2016 ) .
g. jungman , m. kamionkowski and k. griest , phys .
267 , 195 ( 1996 ) .
m. milgrom , astrophys .
j. 270 , 365 ( 1983 ) .
m. milgrom , astrophys .
j. 270 , 384 ( 1983 ) .
s. mcgaugh , phys .
106 , 121303 ( 2011 ) .
b. famaey and s. s. macgaugh , living rev .
relativity , 15 , 10 ( 2012 ) .
r. sanders , astron .
astrophys . rev .
1990 , 2 , 1 ( 1990 ) .
p. kroupa , et al .
523 , 32 ( 2010 ) . c. man ho , d. minic , and y. jack ng , phys .
let b , 9 , 8 ( 2010 ) .
e. verlinde , pre - print arxiv:1611.02269v2 [ hep - th ] p. debye , annalen der physik ( in german ) , leipzig .
39 , 789 ( 1912 ) t. w. murphy , rep .
76 , 076901 ( 2013 t. chaichian , m. oksanen and a. tureanu , phys .
lett b , 712 , 272 ( 2012 ) .
d. gerbal , f. burret , m. lachieze - rey and g. lima - neto , astron .
262 , 395 ( 1992 ) .
g. w. angus , b. famaey , and d. a. buote , mon . not .
r. astron .
387 , 1470 ( 2008 ) .
g. w. angus , and a. diaferio , mon . not .
r. astron .
417 , 941 ( 2011 ) .
m. g. aartsen et al .
( icecube collaboration ) , phys .
117 , 071801 ( 2016 ) .
p. adamson et al .
( daya bay collaboration , minos collaboration ) , phys . rev .
117 , 151801 ( 2016 ) . e. f. bell , d. h. mcintosh , n. katz , m. d. weinberg , astrophys .
j. , 585 , l117 ( 2003 ) .
j. d. horner , r. f. mushotzky and c. a. scharf , astrophys .
j. 520 , 78 ( 1999 ) .
a. e. evrard , c. a. metzler and j. f. navarro , astrophys .
j. 469 , 494 ( 1996 ) .
t. h. reiprich , and h. bohringer , astrophys .
j. 567 , 716 ( 2002 ) .
r. h. sanders , mon . not .
r. astron .
342 , 901 ( 2003 ) .
s. s. mcgaugh , f. lelli , and j. m. schombert , phys .
117 , 201101 ( 2016 ) .
f. lelli , s. s. mcgaugh and j. m. schombert , pre - print arxiv:1606.09251v1 d. a. easson , p. h. frampton , g. f. smoot , phys .
b , 696 , 273 ( 2011 ) .
j. t. nielsen , a. guffanti and s. sarkar sci .
6 , 35596 ( 2016 ) .
|
the debye model of the specific heat of solid at low temperatures is incorporate in the entropic gravity theory ( egt ) . rather of a smooth surface ,
the holographic screen is considered as an oscillating elastic membrane , with a continuous range of frequencies , that cuts off at a maximum ( debye ) temperature , @xmath0 .
we show that at low temperatures @xmath1 , the conservation of the equivalence principle in egt requires a modification of the davies - unruh effect . while the maintenance of davies - unruh effect requires a violation of the equivalence principle .
these two possibilities are equivalents , because both can emulate the same quantity of dark matter .
however , in both cases , the central mechanism is the davies - unruh effect , this seems to indicate that the modification of the davies - unruh effect emulates dark matter which in turn can be see as a violation of the equivalence principle .
this scenario is promising to explain why mond theory works at very low temperatures ( accelerations ) regime , i. e. , the galaxies sector .
we also show that in the intermediate region , for temperatures slightly lower or slightly higher than debye temperature , egt predicts the mass - temperature relation of hot x - ray galaxy clusters .
|
the covariant density functional theory ( cdft ) @xcite is one of standard tools of nuclear theory which offers considerable potential for further development .
built on lorentz covariance and the dirac equation , it provides a natural incorporation of spin degrees of freedom @xcite and an accurate description of spin - orbit splittings @xcite ( see also fig . 2 in ref .
@xcite ) , which has an essential influence on the underlying shell structure . note that the spin - orbit interaction is a relativistic effect , which arises naturally in the cdft theory .
lorentz covariance of the cdft equations leads to the fact that time - odd mean fields of this theory are determined as spatial components of lorentz vectors and therefore coupled with the same constants as time - like components @xcite which are fitted to ground state properties of finite nuclei .
in addition , pseudo - spin symmetry finds a natural explanation in the relativistic framework @xcite .
cdft in its different incarnations both on the mean field and beyond mean field levels provides succesful description of many properties of ground state and excited configurations in nuclei @xcite .
however , the majority of applications of cdft have been focused on collective properties of nuclei .
there are only few features on nuclear systems , strongly dependent on single - particle degrees of freedom , which have been addressed in the cdft studies on the mean field level and compared with experiment .
these are the single - particle properties in spherical and deformed nuclei , spin - orbit splittings in spherical nuclei , magnetic moments in the @xmath1 neighbours of the @xmath2o nucleus ( @xmath3 ) @xcite , and alignment properties of single - particle orbitals in rotating nuclei @xcite .
the calculated single - particle spectra of spherical nuclei are frequently compared with experiment ( see , for example , refs .
however , only in refs .
@xcite deformation polarization effects and time - odd mean fields are included in the calculations which makes them more realistic .
the spin - orbit splittings , as extracted from the single - particle energies obtained in spherical cdft calculations , reproduce the experimental spin - orbit splittings fairly well , although there are deviations up to 20% of absolute value of the splitting ( see figs .
11 - 12 and table 4 in ref .
@xcite and fig . 2 and sect .
ivb in ref .
@xcite ) which only weakly depend on the rmf parametrization .
note that the comparison of ref .
@xcite includes the uncertainty of the spin - orbit splittings due to deformation polarization effects and time - odd mean fields as they are found in ref.@xcite .
@xcite clearly shows that spin - orbit splittings are better described in cdft as compared with skyrme energy density functional theory ( sedft ) despite the fact that no single - particle information ( contrary to sedft ) has been used in the fit of the rmf lagrangian . these comparisons
, however , do not reveal the accuracy of the description of the single - particle states because the particle - vibration coupling ( pvc ) , which can affect considerably the energies of single - particle states in odd - mass nuclei @xcite , has been neglected .
the modification of the quasiparticle states by particle - vibration coupling is weaker in deformed nuclei @xcite since the surface vibrations are more fragmented ( less collective ) than in spherical nuclei @xcite . as a consequence , the corrections to the energies of quasiparticle states in odd nuclei due to particle - vibration coupling
are expected to be less state - dependent in deformed nuclei .
hence the comparison between experimental and calculated energies of single - particle states is expected to be less ambiguous in deformed nuclei as compared with spherical ones @xcite , at least at low excitation energies , where vibrational admixtures in the wave functions are small .
the analysis of the energies of one - quasiparticle deformed states in actinide region , performed within the relativistic hartree - bogoliubov ( rhb ) approach in refs.@xcite , reveals that while the majority of the states are described with an accuracy better than 0.5 mev , there are a number of states which deviate from experiment by as much as 1 mev .
the question arises of which features of the dft are most decisive for the single - particle structure .
the most crucial ingredients in this respect are * _ the effective nucleon mass and its radial dependence which determines the level density near the fermi surface_. it is well known that the one-(quasi- ) particle spectra calculated on the mean field level are less dense than in experiment both at spherical @xcite and deformed @xcite shapes .
the average level density of the single - particle states on the mean field level is related to the effective mass ( lorentz mass in the notation of ref .
@xcite for the case of cdft theory ) of the nucleons at the fermi surface @xmath4 .
the cdft is characterized by a low effective mass @xmath5 @xcite . as a consequence ,
this low effective mass of cdft has a bigger impact on the single - particle spectra than the effective masses of non - relativistic dft which are typically larger .
note that at spherical shape the inclusion of particle - vibrational coupling brings the calculated level density closer to experimental one characterized by an effective mass @xmath6 @xcite .
* _ the spin - orbit potential which determines the energetic distance of the spin - orbit partners_. it is well known that cdft describes rather well spin - orbit splittings @xcite .
however , these results have been obtained on the mean field level , and it is necessary to test how they will be affected by the inclusion of particle - vibrational coupling . in cdft theory
the dirac effective mass @xmath7 is closely related to the effective spin - orbit single - nucleon potential , and empirical energy spacings between spin - orbit partner states in finite nuclei determine a relatively narrow interval of allowed values : @xmath8 @xcite .
this requirement restricts considerably the possible parametrizations of the rmf lagrangian on the mean field level , and thus it would be interesting to see whether the inclusion of pvc will loosen this requirement and thus allow wider range of effective masses in the cdft .
* _ the density dependence of the effective potential and effective mass which has an influence on the relative position of the states with different orbital angular momentum @xmath9 .
_ the investigation of ref .
@xcite , covering spherical odd nuclei bordering doubly magic nuclei clearly showed that the relative placement of the states with different orbital angular momenta @xmath10 is not so well reproduced on the mean field level in cdft .
the problems occur predominantly in connection with the states of high @xmath10 which hints that the surface profile of the mean - field and kinetic terms are involved .
microscopic consideration in many - body theory indicates that the effective mass has a pronounced surface profile which is insufficiently parametrized in the present mean - field models @xcite .
thus , it is important to see whether the inclusion of pvc will improve the description of the relative positions of the states with different orbital angular momentum @xmath9 . in ref .
@xcite , a relativistic particle - vibration coupling model has been developed and the calculations based on the nl3 @xcite parametrization of the cdft have been performed for the doubly magic @xmath11pb nucleus . in this work we present a more systematic investigation within this model including the calculations covering the nuclei from light @xmath0ni and medium - mass @xmath12sn , @xmath13sn to heavy @xmath11pb and superheavy @xmath14120 doubly magic nuclei using improved ( as compared with nl3 ) nl3 * @xcite parametrization of the cdft .
this investigation covers neutron - deficient near proton drip - line nuclei ( @xmath0ni and @xmath12sn ) , neutron - rich @xmath13sn nucleus , and the @xmath11pb nucleus located at the beta - stability line .
in addition , the polarization effects in odd - mass nuclei due to deformation and time - odd mean fields are treated by means of cranked relativistic mean field ( crmf ) calculations .
the article is organized as follows .
the method of comparison of experimental and theoretical single - particle levels is discussed in sec .
[ exp - vs - th ] .
section [ sec - theory ] presents relativistic particle - vibration model and cranked relativistic mean field theory and their details related to the study of single - particle states .
the comparison of model calculations with experiment is presented in sec .
[ res - disc ] .
the accuracy of the description of single - particle spectra and spectroscopic factors , the impact of particle - vibration coupling on shell gaps , spin - orbit splittings and pseudo - spin doublets as well as the role of polarization effects in odd - mass nuclei due to deformation and time - odd mean fields are discussed in detail .
finally , sec .
[ sec - final ] reports the main conclusions of our work .
the experimental information about single - particle level structure of even - even doubly magic nucleus with the proton number @xmath15 and neutron number @xmath16 originates from the odd - mass @xmath17 and @xmath18 nuclear neighbors . in odd - mass nuclei , the polarization effects due to deformation and time - odd ( to ) mean fields induced by odd particle play an important role ( see sec . [ pol - eff ] ) . in addition , the particle - vibration coupling modifies the energies and the structure of the levels ( secs . [ sp - acc ] and [ spec - fac ] ) ; neither of them remains purely single - particle in nature . in order to describe all these effects we use a hybrid model ( labelled as `` hybrid '' below ) , in which the polarization effects due to deformation and to mean fields are treated on the mean field level within cranked relativistic mean field ( crmf ) approach ( see sect . [ mean - field ] ) , while the corrections due to the pvc are treated within relativistic particle - vibration coupling model ( see sect .
[ beyond - mean - field ] ) .
the need for a such hybrid model is dictated by the fact that the pvc model is not variational in nature and thus it does not allow to calculate polarization effects due to odd particle . since the polarization effects and the corrections due to the pvc have a different origin , they are treated as additive quantities in the hybrid model which is a reasonable approximation .
the details of the hybrid model are discussed in sec .
[ sec - theory ] .
the experimental and calculated energies of the particle [ @xmath19 and hole [ @xmath20 states closest to the fermi level are determined from the difference of the binding energies of the core [ @xmath21 and the corresponding adjacent odd nuclei [ @xmath22 and @xmath23 as @xcite @xmath24 and @xmath25 these quantities correspond to one - particle removal energies . we will call them as single - particle energies for simplicity of discussion .
doubly magic spherical nuclei @xmath0ni , @xmath26sn , and @xmath11pb are used as the cores in our analysis .
their ground states are not affected by pvc .
the experimental energies of particles and holes defined according to these equations include the polarization effects due to deformation and to mean fields induced by odd particle or hole as well as the energy corrections due to particle - vibration coupling in odd - mass nuclei .
thus , their comparison with the results of `` hybrid '' calculations is straightforward . on the other hand , there are some unresolved questions when these energies are compared with pure mean field calculations as it is done frequently in the literature .
the problem is related to the fact that experimental single - particle levels and their energies and structure are affected by the particle - vibration coupling which is neglected on the mean field level .
each mean field state @xmath27 with the energy @xmath28 is split into many levels due to particle - vibration coupling , so the single - particle strength is fragmented over many levels . in the diagonal approximation for the nucleonic self - energy , these levels have the same quantum numbers as the original mean field state @xmath27 , but different energies @xmath29 and spectroscopic factors @xmath30 . unlike the hartree or hartree - fock approximations without pairing where the single - particle states are
either fully occupied or empty , in the pvc model their occupation probabilities are the real numbers between zero and one , so that sum rule @xmath31 is satisfied . for the states in the vicinity of the fermi surface one usually obtains one dominant level with @xmath32 and many other levels with small @xmath30 . for the states far from the fermi surface one observes very strong splitting over many levels with much smaller and comparable spectroscopic factors , see the detailed analysis in ref .
@xcite for @xmath11pb .
this discussion clearly indicates that some procedure has to be employed in order to extract pure ( or `` bare '' ) single - particle energies from experimental data which have to be compared with mean field single - particle energies @xmath28 .
two such procedures have been discussed in the literature . in the first procedure , the energy of the dominant fragment is corrected by the energy shift due to particle - vibration coupling leading to a `` bare '' single - particle energy .
such energies of the doubly - magic nuclei are extracted in some publications , see , for example , ref .
however , this procedure relies on the choice of particle - vibration coupling model . however , different pvc models lead to different energy shifts ( see table 1 in ref .
@xcite ) . in addition
, such procedure frequently neglects polarization effects due to deformation and to mean fields which are again model dependent .
for example , there is no time - odd mean fields in the phenomenological models based on the nilsson or woods - saxon potentials . in the second procedure ,
the center of gravity of the fragmented levels with a given quantum number @xmath33 ( @xmath34 is the total angular momentum and @xmath35 is a parity ) and energies @xmath29 is obtained via weighted - average @xcite procedure @xmath36 / \left [ \sum_\nu s_k^{\nu } \right]\end{aligned}\ ] ] this energy is then associated with a `` bare '' single - particle energy .
this expression is fulfilled exactly in the pvc model discussed in sec .
[ beyond - mean - field ] and it is satisfied with high accuracy in this model numerical implementation .
the normalization in this procedure ( which is omitted in some publications ) is used to minimize the uncertainties due to large experimental errors on spectroscopic factors , since these errors lead to the fact that the sum rule @xmath37 is poorly fulfilled in the majority of experiments .
the application of this procedure requires that all @xmath38-levels , to which the single - particle state @xmath27 is fragmented , are identified in the experiment and that spectroscopic factors are measured accurately .
these conditions are definetely not satisfied in the @xmath0ni and @xmath26sn nuclei .
the situation is somewhat better in @xmath11pb , for which the @xmath39 values have been extracted in ref .
@xcite for a number of levels .
however , even in this case the accuracy of the definition of @xmath39 is not known because of the number of the reasons listed below .
the absolute values of spectroscopic factors are characterized by large ambiguities and depend strongly on the reaction employed in experiment and the reaction model used in the analysis @xcite .
the spectroscopic factors in odd mass nuclei have been obtained in different reactions : for example , in the ( d,@xmath40he ) reaction for @xmath41tl @xcite , ( @xmath42 , t ) for @xmath43bi @xcite , ( @xmath40he , @xmath42 ) @xcite and @xmath44 @xcite for @xmath41pb . even in the same nucleus
different reactions give different spectroscopic factors ( see , for example , table i in ref .
@xcite and table [ table - spec-1 ] in the current manuscript ) .
in addition , the spins and parities of many high - lying fragments are uncertain ( see , for example , table 1 in ref .
@xcite ) and/or only the part of the sum rule strength has been observed in experiment ( see , for example , ref .
@xcite ) . as a consequence ,
it is not clear whether the complete set of the @xmath38-levels , to which the single - particle state @xmath27 is fragmented , and their spectroscopic factors have been identified in experiment . in the light of this discussion
it is clear that `` bare '' single - particle energies can not be defined with controllable precision in the nuclei under study . as a result ,
we compare observed experimental levels directly with mean field calculations .
such an approach is frequently used in the literature ( see , for example , refs .
it has clear limitations , but at least it allows to see ( i ) whether the inclusion of particle - vibration coupling improves the description of specific physical observables and ( ii ) whether the conclusions reached earlier on the mean field level are valid or not .
mean field results related to polarization effects due to deformation and to mean fields in odd - mass nuclei have been obtained using cranked relativistic mean field ( crmf ) theory @xcite .
although this theory has initially been developed for the description of rotating nuclei , it is also able to describe the nuclear systems with broken time - reversal symmetry in intrinsic frame at no rotation in which to mean fields play an important role ( see ref .
@xcite ) .
the application of the crmf theory has the following advantages .
first , the crmf computer code is formulated in the signature basis . as a result
, the breaking of kramer s degeneracy of the single - particle states in odd - mass nuclei is taken into account in a fully self - consistent way .
this is important for an accurate description of to mean fields ( ref .
these fields have considerable impact on the moments of inertia in rotating nuclei @xcite and the fact that the properties of such nuclei are well described in the crmf code ( see refs .
@xcite adds confidence to proper description of to mean fields in non - rotating systems @xcite .
second , the crmf computer code is formulated in three - dimensional cartesian coordinates which allows to describe not only axially deformed but also triaxial nuclear shapes .
this is important for a proper description of polarization effects due to deformation .
it is well known fact that the description of pairing correlations in doubly magic nuclei and their close neighborhood is notoriously difficult and model dependent ( ref .
the presence of large shell gaps leads to a pairing collapse in the models with the treatment of pairing on the bcs or hfb [ hartree - fock - bogoliubov ] levels .
this is seen , for example , in non - relativistic hfb calculations with the gogny d1s force of @xmath26sn in ref .
the calculations within the relativistic hartree - bogoliubov ( rhb ) framework of ref .
@xcite with the same force for pairing also show pairing collapse in the ground states of @xmath0ni , @xmath26sn and @xmath11pb .
the pairing correlations are restored when the particle number projection is implemented .
however , there is a substantial difference between different approaches ( such as lipkin - nogami ( ln ) method of approximate particle number projection or projection after variation ( pav ) approach ) and variation after projection ( vap ) approach ; the latter represents the best approach for the treatment of pairing correlations @xcite .
the rhb+ln calculations with the gogny d1s force for pairing show very weak pairing in the nuclei of interest .
the additional binding due to pairing ( relatively to the results with no pairing ) is only 0.333 mev , 0.284 mev , 0.493 mev and 0.557 mev in @xmath0ni , @xmath26sn and @xmath11pb , respectively .
the blocking of single - particle orbital in odd - mass nucleus leads to the weakening of pairing correlations and to frequent unphysical pairing collapse even in the rhb+ln calculations .
the physical quantities , such as given by eqs .
( [ eq - part ] ) and ( [ eq - hole ] ) , are relative in nature .
thus , we believe that in the case of weak pairing they are sufficiently well described in unpaired calculations . as a consequence ,
we neglect pairing correlations in the crmf calculations .
note that they are also neglected in the pvc calculations .
similar to ref .
@xcite , binding energies of odd-@xmath45 nuclei were calculated by blocking the @xmath46 state ( amongst the @xmath47 states ) with the largest projection @xmath48 onto the symmetry axis . here ,
the @xmath49 and @xmath34 are the single - nucleon radial and total angular momentum quantum numbers .
the spherical subshell label of this state is used to label the configuration of odd mass nucleus since the deformations ( induced by odd particle or hole ) of immediate odd-@xmath45 neighbors of doubly magic nuclei are rather small .
the mean field approach , being a very useful and convenient theoretical tool to describe finite nuclei , is , however , only a static and local approximation to the many - body problem . the non - locality ( or , equivalently , momentum dependence ) is implicitly taken into account by an effective mass which is considerably smaller than the bare nucleon mass .
this enables one to reproduce reasonably well the general properties of the single - particle motion in the nuclear potential well .
however , the mean field approximations based on the widely used skyrme , gogny as well as covariant density functionals generate too stretched single - particle spectra around the fermi surface and often fail to describe the sequence of the single - particle levels .
proper inclusion of the time dependence in the nucleonic potential requires to go beyond the mean field description which is static by construction . the underlying physics which is responsible for the time dependence is related to many - body correlations , i.e. to the part of the hamiltonian which can not be expressed in terms of the one - body density . in the medium mass and heavy nuclei the leading order contribution of such correlations to
the nucleonic self - energy comes from coupling of the nucleonic motion to the surface and volume vibrational modes ( phonons ) .
the most important collective vibrational modes are generated by the nuclear oscillations of the coherent nature .
taking into account the coupling of these modes to the single - particle motion , the shell model acquires a dynamic content .
these surface and volume oscillations , especially their low - lying modes , modify considerably the picture of the single - particle motion in nuclei .
the main assumption of the quasiparticle - phonon coupling model is that two types of elementary degrees of freedom
one - quasiparticle ( 1-qp ) and collective ones are coupled in such a way that configurations of 1-qp@xmath50phonon type with low - lying phonons strongly compete with simple 1-qp configurations located close in energy or , in other words , that quasiparticles can emit and absorb phonons with rather high probabilities @xcite .
taking the quasiparticle - phonon coupling into account as an additional time dependent part of the nucleonic self - energy leads to the splitting of each mean - field single - particle state into many energy levels .
the dominant levels , i.e. the levels with the largest spectroscopic factors , shift , as a rule , towards the fermi level improving considerably the agreement with data , see the review of various applications in ref . @xcite and references therein .
this result is obtained , however , only for the states in about 10 mev vicinity of the fermi energy . for the states lying far from the fermi surface one
observes a very strong fragmentation making an extraction of the dominant levels impossible .
similar general picture has been obtained within the particle - vibration coupling extension of the covariant density functional theory @xcite . unlike the non - relativistic versions of the model , in the extended cdft the particle - phonon coupling self - energy allows intermediate nucleonic propagation through the dirac sea states with negative energies , in addition to the particle and hole states in the fermi sea . the contribution of the dirac - sea terms to the self - energy at the fermi level are , however , found negligibly small due to their large energy denominators . in the present investigation
we retain basically calculational scheme of ref .
this approach implies linearized version of particle - vibration coupling model : the dyson equation for single - particle green function contains phonon - coupling self - energy with mean - field intermediate nucleonic propagators and phonon energies and vertices computed within relativistic random phase approximation ( rrpa ) @xcite . for medium - mass and heavy nuclei
the rrpa gives a very reasonable description of the phonon spectra .
although a more precise description of the phonons can be achieved within an approach like relativistic time blocking approximation ( rtba ) @xcite , particle - vibration coupling model of ref .
@xcite by using the rrpa phonons takes into account the major contribution of the vibrational motion to the single - particle spectra . the application of the model to the lightest doubly magic nuclei @xmath2o , @xmath51ca , and @xmath52ca requires , however , a special consideration because their mean - field single - particle level densities are too small to provide a sizable configuration mixing around the fermi surface . for @xmath2o and @xmath51ca ,
it is not possible , in principle , to reproduce within the rrpa the low - lying phonons with positive parity ( first positive parity particle - hole excitations are at too high energies ) . in such cases
the phonon spectra for the nucleonic self - energy should be calculated allowing an extension for coupling to more complex configurations ; that implies a generalized particle - vibration coupling model .
this will be done in a separate work .
in @xmath52ca we have obtained a rather reasonable phonon spectrum , but the strong fragmentation of the hole states does not allow extraction of dominant single - particle levels .
the phonons of unnatural parities are known to play only a marginal role in the nucleonic self - energy , therefore , they are not included into the phonon space .
moreover , the inclusion of pairing vibrations into the phonon space has not been studied yet within the relativistic framework . in principle
, they may bring some additional corrections to the results , although our approximation is justified by the results of refs .
@xcite , where the fragmentation of the collective modes has been reproduced very well without inclusion of the pairing vibrations .
the crmf equations are solved in the basis of an anisotropic three - dimensional harmonic oscillator in cartesian coordinates characterized by the deformation parameters @xmath53 and @xmath54 as well as the oscillator frequency @xmath55 mev .
the truncation of the basis is performed in such a way that all states belonging to the shells up to fermionic @xmath56=20 and bosonic @xmath57=20 are taken into account in the calculations .
numerical analysis indicates that this truncation scheme provides sufficient numerical accuracy for the physical quantities of interest .
note that the same truncation of basis is used in the rrpa and pvc calculations , but they are performed at spherical shape .
the nl3 * @xcite parametrization of the rmf lagrangian is used in the crmf , rrpa and pvc calculations .
this recently fitted parametrization has been successfully applied to the description of binding energies @xcite , ground state properties of deformed nuclei @xcite , fission barriers @xcite , rotating nuclei @xcite , giant resonances @xcite , and breathing mode @xcite .
phonons of the multipolarities 2@xmath58 , 3@xmath59 , 4@xmath58 , 5@xmath59 , 6@xmath58 with energies below 15 mev are included in the model space of the pvc calculations .
we have neglected only the phonon modes with very small transition probabilities , less than 5% of the maximal ones for each j@xmath60 .
the phonon energies and their coupling vertices have been computed within the self - consistent rrpa .
a rather good convergence of the phonon coupling self - energy with extension of the phonon space has been obtained numerically : with very few exceptions , addition of phonon modes with energies above 15 mev does not affect the results .
partial contributions from the phonons with low angular momenta and low energies have been also investigated .
an example of such a study is shown in fig .
[ n1g92 ] for the neutron 1g@xmath61 state in @xmath13sn .
this typical deeply bound hole state is strongly fragmented when the pvc is taken into account , see the panel ( d ) with the results obtained with the full phonon set below 15 mev .
panels ( a ) and ( b ) show the results obtained when only the phonons below 5 and 10 mev , respectively , are taken into account . one can see that in the former case the picture is very different from the one displayed in the panel ( d ) . in the latter case
the results are already close to saturation , but there are still two states between @xmath62 and @xmath63 mev with the spectroscopic factors larger than 0.1 .
panel ( c ) shows how the angular momentum cut - off weakens the fragmentation effect .
similar conclusion can be drawn for other states either far from or around the fermi surface : cut - off on the angular momenta and on the energies of phonons has to be made with care .
an example of the comparison of the results of calculations and experiment is shown in fig .
[ ni56-fig ] .
column `` sph '' shows single - particle spectra obtained in spherical rmf calculations of even - even @xmath0ni nucleus .
column `` def '' displays one - nucleon separation energies [ defined according to eqs .
( [ eq - part ] ) and ( [ eq - hole ] ) ] obtained in triaxial crmf calculations in which only deformation polarization effects are taken into account .
it is seen that these energies are close to the single - particle energies obtained in spherical calculations .
this shows that deformation polarization effects induced by extra particle or hole are rather modest . in general
, they lead to a slight compression of the calculated spectra as compared to those obtained in spherical calculations .
note that the magnitude and impact of deformation polarization effects is similar in our calculations and in the calculations of ref .
@xcite .
-[pb208-fig ] , the absolute values of the difference of the energies obtained in spherical ( the @xmath64 energies ; column `` sph '' in figs . [ ni56-fig]-[pb208-fig ] ) and deformed ( the @xmath65 energies of the calculations with to mean fields included ; column `` def+to '' in figs . [ ni56-fig]-[pb208-fig ] )
calculations are shown .
the average polarization effects per state @xmath66 ( @xmath67 ) are also shown.,width=302 ] the inclusion of time - odd mean fields induces additional binding in odd - mass nuclei ( see ref .
@xcite for more detail ) which is rather modest being around 100 - 200 kev in the nuclei around @xmath0ni .
this again introduces slight compression of the calculated spectra ( fig .
[ ni56-fig ] ) .
one should note that additional binding due to to mean fields in odd - mass nuclei ( and a subsequent compression of the spectra ) is substantially larger in the calculations of ref .
for example , it is around 0.5 mev in the nuclei around @xmath11pb and reaches @xmath68 mev in @xmath0ni .
note that for the latter nucleus additional binding due to to mean fields is only @xmath69 mev in our crmf calculations ( see fig .
[ ni56-fig ] ) .
the reason for this difference is not clear .
however , additional bindings due to to mean fields in the nuclei under study obtained in our crmf calculations are within the typical ranges obtained in the systematic study of the impact of to mean fields on the binding energies of odd mass nuclei ( ref.@xcite ) .
in addition , the crmf code employed here has been extensively and successfully used in the description of rotating systems in which to fields have very large impact on the moments of inertia ( see refs .
@xcite and references quoted therein ) .
these two facts strongly suggest that the impact of to mean fields is correctly described in the current work .
the dominant levels , i. e. the levels with the largest spectroscopic factors , as obtained in spherical pvc calculations , are shown in column `` sph+pvc '' .
one can see two effects of particle - vibration coupling : ( i ) the general compression of the spectra leading to a better agreement with experiment and ( ii ) in some cases the change of the level sequences . with few exceptions , the spectroscopic factors of the dominant levels in the vicinity of the fermi level vary between 0.5 and 0.9 ( see tables [ table - spec-1 ] and [ table - spec-2 ] below ) ,
thus these states retain basically their single - particle nature .
the impact of particle - vibration coupling on specific state can be quantified by the energy difference between the energies of this state in the columns `` sph '' and `` sph+pvc '' of fig .
[ pb208-fig ] .
these energy differences treated as the corrections due to pvc are then added to one - neutron separation energies obtained in triaxial crmf calculations ; this leads to column `` hybrid '' .
the results in this column are compared with experimental data presented in column `` exp '' .
the single - particle spectra of doubly magic @xmath12sn , @xmath13sn and @xmath11pb nuclei , obtained in the same way as fig .
[ ni56-fig ] , are displayed in figs .
[ sn100-fig ] , [ sn132-fig ] and [ pb208-fig ] .
they show similar effects due to pvc and the polarizations induced by deformation and to mean fields as in the case of @xmath0ni .
their details will be discussed below .
let consider major conclusions emerging from these calculations .
they are related to the role of polarizations effects due to deformation and to mean fields and the impact of pvc on the accuracy of the description of single - particle spectra , shell gaps , pseudospin doublets and spin - orbit splittings .
combined polarization effects due to deformation and to mean fields are shown in fig .
[ polarization ] .
two features are clearly seen .
first , combined polarization effects due to deformation and time - odd mean fields decrease with an increase of mass number .
indeed , with an increase of the size of nucleus the relative role of each single - particle orbital becomes smaller which leads to the fact that the calculated quadrupole deformations of odd - mass nuclei decrease with an increase of mass .
for example ,
the average proton and neutron deformations of calculated single - particle states in @xmath70cu and @xmath70ni are 0.045 and 0.060 , respectively . on the contrary , these deformations are significantly smaller in @xmath43bi and @xmath43sn ; they are only 0.0055 and 0.006 , respectively .
thus , odd nuclei neighboring to @xmath11pb remain basically spherical .
the decrease of calculated deformations of odd mass nuclei neighboring to doubly magic nuclei with increasing mass clearly indicates that deformation polarization effects also decrease with mass number .
in addition , additional binding due to to mean fields decreases with mass ( see ref .
@xcite for details ) .
the combination of these two effects is responsible for the features seen in fig .
[ polarization ] .
[ polarization ] also reveals that combined polarization effects due to deformation and to mean fields are smaller for protons than for neutrons by a factor of approximately two .
this is especially true for the @xmath71 @xmath0ni and @xmath12sn nuclei which have the same proton and neutron single - particle states .
the part of this difference comes from the fact that additional binding due to to mean fields is smaller for odd - proton nuclei as compared with odd - neutron ones ( see ref .
@xcite ) . the analysis of ref .
@xcite clearly indicates that the contributions of the coulomb force to the proton single - particle energies in the presence of to mean fields is responsible for this feature .
deformation polarization effects show the same features , namely , they are smaller for odd - proton nuclei as compared with odd - neutron ones . for example , average deformation polarizations per single - particle state are 0.157 mev and 0.326 mev for proton and neutron single - particle states of @xmath0ni and 0.060 mev and 0.164 mev for proton and neutron single - particle states of @xmath12sn . similar to the case of polarizations due to to mean fields , the origin of this difference has to be traced back to the coulomb force since compared sets of proton and neutron single - particle states are the same in a given @xmath71 nucleus . the majority of the comparisons of the energies of the single - particle states obtained in different particle vibration coupling models with experiment neglect deformation and to mean field polarization effects induced by the odd particle ( see , for example , refs .
the current analysis indicates that within the framework of cdft this neglect is more or less justified only for heavy nuclei , and it is more justified for proton subsystem than for neutron one .
the analysis of figs .
[ ni56-fig ] , [ sn132-fig ] , and [ pb208-fig ] reveals that the inclusion of particle vibration coupling substantially improves the accuracy of the description of single - particle spectra . this statement
is quantified in fig .
[ dev - fig ] and table [ table - dev ] .
[ dev - fig ] displays the distribution of the deviations between calculated and experimental energies of the single - particle states in proton and neutron subsystems of @xmath0ni , @xmath13sn and @xmath43pb .
it clearly shows that _ in average _ the inclusion of particle - vibration coupling substantially improves the description of single - particle spectra , and that neutron single - particle states are better described than proton ones .
however , even in the `` hybrid '' calculations there are few ( mostly proton ) states which deviate from experiment by more than 1 mev .
note that the inclusion of particle - vibration coupling can make the agreement between theory and experiment worse for some states .
this is seen , for example , in the case of proton @xmath72 and @xmath73 subshells in @xmath11pb ( fig .
[ pb208-fig ] ) .
.average deviations per state @xmath74 between calculated and experimental energies of the single - particle states for a proton ( neutron ) subsystem of a given nucleus .
the results obtained in the `` def+to '' and `` hybrid '' calculational schemes are shown .
[ cols="^,^,^",options="header " , ] [ table - dev ] sn , the experimental shell gaps are extracted from extrapolated single - particle energies of refs .
@xcite ( a ) and @xcite ( b ) [ see also table [ table - sn100 ] ] . note that the size of the @xmath75 shell gap in @xmath11pb is almost the same in both calculations.,width=302 ] average deviations per state @xmath74 between calculated and experimental energies of the single - particle states are shown in table [ table - dev ] .
they are defined as @xmath76 where @xmath77 is the number of the states with known experimental single - particle energies , and @xmath78 ( @xmath79 ) are calculated ( experimental ) energies of the single - particle states .
one can see that the inclusion of pvc substantially improves the description of the single - particle states in @xmath13sn and @xmath11pb . on the contrary
, pvc introduces no ( small ) improvement in the description of the proton ( neutron ) single - particle states of @xmath0ni .
it is interesting to mention that the best agreement with experiment is obtained in the `` hybrid '' calculations of neutron - rich @xmath13sn .
while the accuracy of the description of proton and neutron states is comparable in @xmath0ni , neutron states are appreciable better described than proton ones in the `` hybrid '' calculations of @xmath13sn and @xmath11pb . the analysis of figs .
[ sn132-fig ] and [ pb208-fig ] strongly suggests that the increases of the depth of proton potential in these two nuclei by few hundreds kev ( and as a result the lowering of the energies of proton single - particle states by similar energy ) will improve the agreement with experiment .
this clearly indicates that the inclusion of the single - particle energies into the fit of the cdft parametrizations can provide an extra information on the depth of the proton and neutron potentials and their evolution with particle numbers .
@xmath80(mev ) @xcite + + @xmath81 & -8.6(5 ) & -7.8(8 ) + @xmath82 & -9.2(5 ) & -8.8(8 ) + @xmath72 & -9.3(5 ) & -9.3(9 ) + @xmath83 & -10.93(20 ) & -10.4(10 ) + @xmath84 & -11.13(20 ) & -11.1(10 ) + @xmath85 & -17.93(30 ) & + @xmath86 & -18.38(20 ) & + + @xmath87 & + 3.90(15 ) & 2.6(3 ) + @xmath84 & + 3.00(80 ) & 2.8(3 ) + @xmath85 & -2.92(20 ) & -2.9(3 ) + @xmath86 & -3.53(20 ) & -3.5(3 ) + @xmath88 & -6.38 & + @xmath89 & -8.71 & + [ table - sn100 ] the detailed comparison between theory and experiment has not been performed for @xmath12sn since very little experimental information is available on neighboring to @xmath12sn nuclei with one nucleon less ( more ) ( see , for example , refs .
unfortunately , the extrapolations of the single - particle energies from other odd nuclei towards @xmath12sn do not bring unique values .
this is illustrated in table [ table - sn100 ] where the results of the extrapolations of refs .
@xcite are shown .
note that the estimates of the energies of the single - particle states of ref .
@xcite coincide within the estimation errors ( given in table [ table - sn100 ] in parentheses ) with the values reported in ref .
@xcite , except for the energy of the @xmath87 proton state . however , estimation errors are significant which makes detailed comparison between theory and experiment meaningless . on the other hand , one can clearly see in fig .
[ sn100-fig ] that pvc improves the agreement with experimental estimates .
= 1.40em [ c]ccccc|ccccnucleus & state & @xmath90 & @xmath91 & @xmath91 & nucleus & state & @xmath90 & @xmath91 + & & & & & & & + @xmath43pb & 2g@xmath61 & 0.85 & [email protected] @xcite & 0.94 @xcite & @xmath93sn & 2f@xmath94 & 0.89 & [email protected] + & 1i@xmath95 & 0.89 & [email protected] @xcite & 1.05 @xcite & & 3p@xmath96 & 0.91 & [email protected] + & 1j@xmath97 & 0.66 & [email protected] @xcite & 0.57 @xcite & & 1h@xmath61 & 0.88 & + & 3d@xmath98 & 0.89 & [email protected] @xcite & & & 3p@xmath99 & 0.91 & [email protected] + & 4s@xmath99 & 0.92 & [email protected] @xcite & & & 2f@xmath98 & 0.89 & [email protected] + & 2g@xmath94 & 0.87 & [email protected] @xcite & & & & & + & 3d@xmath96 & 0.89 & [email protected] @xcite & & & & & + & & & & & & & & + @xmath43bi & 1h@xmath61 & 0.88 & 1.17 @xcite & 0.80 @xcite & @xmath93sb & 1g@xmath94 & 0.86 & + & 2f@xmath94 & 0.78 & 0.78 @xcite & 0.76 @xcite & & 2d@xmath98 & 0.82 & + & 1i@xmath100 & 0.63 & 0.56 @xcite & 0.74 @xcite & & 2d@xmath96 & 0.63 & + & 2f@xmath98 & 0.61 & 0.88 @xcite & 0.57 @xcite & & 1h@xmath95 & 0.79 & + & 3p@xmath96 & 0.62 & 0.67 @xcite & 0.44 @xcite & & 3s@xmath99 & 0.70 & + & 3p@xmath99 & 0.37 & 0.49 @xcite & 0.20 @xcite & & & & + & & & & & & & & + @xmath41pb & 3p@xmath99 & 0.90 & & 1.08 @xcite & @xmath101sn & 2d@xmath96 & 0.88 & + & 2f@xmath98 & 0.87 & 1.13 @xcite & 1.05 @xcite & & 1h@xmath95 & 0.86 & + & 3p@xmath96 & 0.86 & 1.00 @xcite & 0.95 @xcite & & 3s@xmath99 & 0.87 & + & 1i@xmath100 & 0.82 & 1.04 @xcite & 0.61 @xcite & & 2d@xmath98 & 0.70 & + & 2f@xmath94 & 0.64 & 0.89 @xcite & 0.64 @xcite & & 1g@xmath94 & 0.72 & + & 1h@xmath61 & 0.38 & & & & & & + & & & & & & & & + @xmath41tl & 3s@xmath99 & 0.84 & 0.95 @xcite & 0.85 @xcite & @xmath101 in & 1g@xmath61 & 0.85 & + & 2d@xmath96 & 0.86 & 1.15 @xcite & 0.90 @xcite & & 2p@xmath96 & 0.70 & + & 1h@xmath95 & 0.80 & 0.89 @xcite & 0.88 @xcite & & 2p@xmath99 & 0.85 & + & 2d@xmath98 & 0.68 & 0.62 @xcite & 0.63 @xcite & & 1f@xmath98 & 0.37 & + & 1g@xmath94 & 0.22 & 0.40 @xcite & 0.27 @xcite & & & & + [ pbsn ] [ table - spec-1 ] the size of the shell gap ( between last occupied and first unoccupied states in even - even doubly magic nucleus ) is defined as @xmath102 where @xmath103 and @xmath104 stand for the energies of the single - particle states above and below the shell gap .
this definition of the shell gap incorporates the information from three different nuclei because the single - particle energies are defined according to eqs .
( [ eq - part ] ) and ( [ eq - hole ] ) .
the advantage of this definition is that it allows to incorporate the polarization ( due to deformation and to mean fields ) and pvc effects on the size of shell gap .
[ gaps - fig ] compares experimental shell gaps with the shell gaps obtained in the `` def+to '' and `` hybrid '' calculations .
the extrapolations of refs .
@xcite for the single - particle energies in @xmath12sn lead to quite different values of the proton @xmath105 shell gap .
it is reasonable to expect the same situation for the neutron @xmath106 shell gap , the size of which , however , can not be determined in the extrapolations of ref .
@xcite since no estimates for the single - particle states below it are provided ( see table [ table - sn100 ] ) .
the shell gaps are largest in the spherical mean field calculations .
the gaps become smaller with the inclusion of each additional type of correlations . the inclusion of polarization effects due to deformation and to mean fields
decreases slightly their sizes ( see figs .
[ ni56-fig ] , [ sn100-fig ] , [ sn132-fig ] and [ pb208-fig ] ) .
thus , we compare with experiment in fig .
[ gaps - fig ] only the pvc ( `` hybrid '' ) and best mean field ( `` def+to '' ) calculations .
this figure shows that particle - vibration coupling decreases substantially the size of the shell gaps .
the effect is most pronounced in @xmath0ni where pvc decreases the proton @xmath107 and neutron @xmath108 shell gaps by almost 2 mev . on the contrary
, the effect of particle - vibration coupling is least pronounced in @xmath11pb in which it decreases the size of the neutron @xmath109 gap by only few hundreds kev and has almost no impact on the proton @xmath75 shell gap . comparing experimental and calculated shell gaps in @xmath0ni , @xmath13sn and @xmath11pb
, one can conclude that both calculational schemes provide similar accuracy of the description of experimental data .
however , experimental shell gaps are typically overestimated in the `` def+to '' calculations .
the exceptions are the @xmath105 gap in @xmath13sn and @xmath75 gap in @xmath11pb . on the contrary , with exception of the @xmath109 gap in @xmath11pb , the size of experimental gaps
is underestimated in the calculations with particle - vibration coupling .
= 1.40em [ c]cccc|cccccnucleus & state & @xmath90 & & nucleus & state & @xmath90 & @xmath91 @xcite & @xmath91 @xcite + & & & & & & & & + @xmath110sn & 2d@xmath98 & 0.85 & & @xmath70ni & 2p@xmath96 & 0.83 & [email protected] & [email protected] + & 1g@xmath94 & 0.85 & & & 1f@xmath98 & 0.79 & [email protected] & + & 2d@xmath96 & 0.78 & & & 2p@xmath99 & 0.76 & [email protected] & + & 3s@xmath99 & 0.81 & & & 1g@xmath61 & 0.79 & & + & 1h@xmath95 & 0.80 & & & & & & + & & & & & & & & + @xmath110sb & 2d@xmath98 & 0.87 & & @xmath70cu & 2p@xmath96 & 0.85 & & + & 1g@xmath94 & 0.86 & & & 1f@xmath98 & 0.80 & & + & 2d@xmath96 & 0.83 & & & 2p@xmath99 & 0.80 & & + & 3s@xmath99 & 0.87 & & & 1g@xmath61 & 0.80 & & + & 1h@xmath95 & 0.81 & & & & & & + & & & & & & & & + @xmath111sn & 1g@xmath61 & 0.84 & & @xmath112ni & 1f@xmath94 & 0.78 & & + & 2p@xmath99 & 0.85 & & & 2s@xmath99 & 0.71 & & + & 2p@xmath96 & 0.71 & & & 1d@xmath96 & 0.62 & & + & 1f@xmath98 & 0.62 & & & 1d@xmath98 & 0.20 & & + & 1f@xmath94 & 0.11 & & & & & & + & & & & & & & & + @xmath111 in & 1g@xmath61 & 0.85 & & @xmath112co & 1f@xmath94 & 0.78 & & + & 2p@xmath99 & 0.86 & & & 2s@xmath99 & 0.73 & & + & 2p@xmath96 & 0.73 & & & 1d@xmath96 & 0.64 & & + & 1f@xmath98 & 0.66 & & & 1d@xmath98 & 0.20 & & + & 1f@xmath94 & 0.14 & & & & & + [ table - spec-2 ] of the spin - orbit doublets in @xmath11pb .
insert shows the colors used for different values of orbital angular momentum @xmath9 .
theoretical results are shown by symbols , while the experimental values by solid lines .
solid ( open ) symbols are used for spin - orbit doublets which are build from the particle ( hole ) states with respect of the @xmath11pb core .
the stars and crosses are used for the spin - orbit doublets which involve both particle and hole states.,width=302 ] the calculated spectroscopic factors for odd nuclei with doubly - magic cores are compared with experiment in tables [ pbsn ] and [ snni ] . in odd nuclei neighboring to @xmath11pb , the experimental spectroscopic factors are reasonably well reproduced in the pvc calculations . both in experiment and in calculations
the proton states are found to be somewhat more fragmented than the neutron ones . when comparing theory with experiment one should keep in mind that the experimental spectroscopic factors depend considerably on the parameters used in the model analysis and on employed reaction , which is clearly seen in the staggering of experimental values .
note that the results obtained for @xmath11pb with the nl3 * parametrization are very close to those obtained previously in ref .
@xcite with the nl3 force . for other nuclei ,
the experimental data are available only for @xmath93sn ( table [ pbsn ] ) and @xmath70ni ( table [ snni ] ) .
the level of agreement between theory and experiment in @xmath93sn is comparable with the one seen in odd nuclei neighbouring to @xmath11pb .
the situation in @xmath70ni is more controversial .
for example , the sum rule ( [ sum_rule ] ) is strongly violated in the case of the @xmath89 state in experimental data of ref .
in addition , experimental spectroscopic factors for the @xmath88 state in @xmath70ni extracted by means of the @xmath113 transfer @xcite and one - neutron knockout @xcite reactions differ by a factor of almost two . currently there is no satisfactory explanation for this difference ( ref .
the results of the pvc calculations are lower than the experimental values obtained in the @xmath113 transfer reaction but higher than the ones obtained in the one - neutron knockout reaction .
on the other hand , the spectroscopic factors obtained in the pvc calculations are very close to the ones obtained in the spherical shell model calculations in full @xmath114 model space which employ gxpf1a interaction ( see table ii in ref .
note that the latter calculations provide best agreement with the data in the ni isotopes @xcite .
one should also mention that non - relativistic particle - vibration calculations of ref .
@xcite give the spectroscopic factors for the @xmath88 , @xmath89 , and @xmath86 states which are very close to 0.6 .
covariant density functional theory naturally describes the spin - orbit interaction in nuclei @xcite , which is a relativistic effect that has to be added phenomenologically in non - relativistic models .
the comparison of experimental spin - orbit splittings with calculations has been performed only on the mean field level so far ( see , for example , refs .
however , the current work clearly indicates that particle - vibration coupling has an impact on the energies of the single - particle states .
thus , it is reasonable to expect that it will also modify the spin - orbit splittings .
but for the spin - orbit doublets of @xmath13sn.,width=302 ] in order to understand the impact of polarization effects ( due to deformation and to mean fields ) and particle - vibration coupling on spin - orbit splittings , the results of the `` sph '' , `` sph+pvc '' , `` def '' , `` def+to '' , and `` hybrid '' calculations are compared with available experimental data on the spin - orbit doublets in figs .
[ so - pb208 ] , [ so - sn132 ] and [ so - ni56 ] . the impact of deformation and to mean field polarization effects on spin - orbit splittings is rather modest in @xmath11pb ( fig .
[ so - pb208 ] ) since these effects are smallest in heavy systems ( see sec . [ pol - eff ] ) .
in addition , both of them lead to a compression of the single - particle spectra . as a consequence ,
they act in the same direction for the members of the spin - orbit doublet which is built either from hole ( below the shell gap ) or particle ( above the shell gap ) states . moreover , these polarization effects are more or less similar in magnitude for the members of such spin - orbit doublets . as a consequence , they cancel to a large extent when spin - orbit splitting is calculated as the difference of the energies of the members of the spin - orbit doublet .
such behavior is typical for the spin - orbit doublets based on the orbitals with low orbital angular momentum _
l _ , both members of which are located either below or above the shell gap . for this type of the spin - orbit doublets , the impact of polarization effects on spin - orbit splittings is also small in @xmath13sn ( fig .
[ so - sn132 ] ) . however , it becomes more pronounced ( especially , in neutron subsystem ) in @xmath0ni ( fig .
[ so - ni56 ] ) .
the situation is different for the spin - orbit doublets built on the orbitals with high orbital angular momentum _
l_. for such doublets , one member is located above the shell gap , while another below the shell gap .
the compression of the single - particle spectra due to deformation and to mean field polarization effects leads to the decrease of spin - orbit splitting .
this decrease is especially pronounced for light nuclei and for neutron subsystem in a given nucleus ( compare figs .
[ so - ni56 ] , [ so - sn132 ] , and [ so - pb208 ] ) because of particle number and subsystem ( proton or neutron ) dependencies of polarization effects discussed in sec .
[ pol - eff ] .
the impact of particle - vibration coupling on spin - orbit splittings is state - dependent .
it is small or modest in the proton @xmath115 @xmath116 and neutron @xmath117 @xmath118 , @xmath119 @xmath120 , and @xmath121 @xmath122 doublets of @xmath11pb ( fig .
[ so - pb208 ] ) , proton @xmath119 @xmath120 and neutron @xmath117 @xmath118 , @xmath115 @xmath116 doublets of @xmath13sn ( fig .
[ so - sn132 ] ) and neutron and proton @xmath117 @xmath123 doublets in @xmath0ni ( fig .
[ so - ni56 ] ) . in these doublets ,
the corrections to the energies of the members of the spin - orbit doublet due to pvc are more or less the same . however , this is not always the case for the doublets based on the orbitals with low orbital angular momentum _
l_. for example , the corrections due to pvc differ significantly for the proton @xmath124 and @xmath125 states of @xmath11pb ( fig .
[ pb208-fig ] ) . as a consequence
, particle - vibration coupling has significant effect on the spin - orbit splitting of the proton @xmath119 @xmath126 doublet .
in addition , spin - orbit splittings of the proton @xmath117 @xmath118 and neutron @xmath115 @xmath116 doublets in @xmath11pb ( fig .
[ so - pb208 ] ) as well as of neutron @xmath119 @xmath120 doublet in @xmath13sn ( fig .
[ so - sn132 ] ) are strongly affected by particle - vibration coupling .
the effect of particle - vibration coupling is especially pronounced for the spin - orbit doublets with high orbital angular momentum @xmath9 , the one member of which is located below the shell gap and another above the shell gap .
these are the proton @xmath127 @xmath128 and neutron @xmath129 @xmath130 doublets in @xmath11pb ( fig .
[ so - pb208 ] ) , the proton @xmath121 @xmath131 and neutron @xmath127 @xmath128 doublets in @xmath13sn ( fig .
[ so - sn132 ] ) and the proton and neutron @xmath115 @xmath132 doublets in @xmath0ni ( fig .
[ so - ni56 ] ) . for these doublets ,
the spin - orbit splittings are largest in spherical calculations . the inclusion of polarization effects due to deformation and to mean fields decreases these splittings .
the pvc coupling further reduces the energy splittings by approximately 1 mev in medium and heavy mass nuclei ( figs .
[ so - pb208 ] and [ so - sn132 ] ) and by approximately 1.5 mev in @xmath0ni ( fig .
[ so - ni56 ] ) .
[ so - dev ] presents percentual deviations @xmath133 of the calculated spin - orbit splittings from experimental ones defined as @xmath134 where @xmath135 is the spin - orbit splitting energy .
negative ( positive ) values of @xmath133 indicate that experimental spin - orbit splitting is underestimated ( overestimated ) in the calculations .
one can see that in average the mean - field results are more or less evenly scattered around zero percentual deviation . on the contrary , with few exceptions particle - vibration coupling calculations underestimate experimental spin - orbit splittings .
spin - orbit doublets built on the orbitals with high orbital momentum @xmath9 , involving the members across the shell gap , are rather well ( within the 20% deviation ) described in both calculations .
these doublets are characterized by large spin - orbit splittings @xmath135 in the range of 5 - 7 mev , so the 20% deviation means that the absolute deviation from experiment is typically less than 1 mev .
the largest percentual deviations are observed for the spin - orbit doublets built on the orbitals with low orbital angular momentum @xmath9 .
these doublets are characterized by low spin - orbit splittings so relatively low absolute deviations from experimental values of the order of few hundreds kev result in appreciable percentual deviations .
the analysis of the results of calculations allows to conclude that the inclusion of particle - vibration coupling decreases the accuracy of the description of spin - orbit splittings .
this is clearly visible when absolute and/or percentual deviations per doublet are compared in the mean field and pvc calculations .
the absolute deviations per doublet are 0.34 mev [ 0.50 mev ] , 0.23 mev [ 0.56 mev ] and 0.26 mev [ 0.45 mev ] in the mean field ( `` def+to '' ) [ particle - vibration coupling ( `` hybrid '' ) ] calculations in @xmath0ni , @xmath13sn and @xmath11pb , respectively .
the percentual deviations per doublet are 11.8 % [ 10.3% ] , 14% [ 21.5% ] and 19.3% [ 36.4% ] in the mean field ( `` def+to '' ) [ particle - vibration coupling ( `` hybrid '' ) ] calculations in @xmath0ni , @xmath13sn and @xmath11pb , respectively .
these results are not surprising and should not be viewed negatively .
first of all , as discussed in secs .
[ exp - vs - th ] and [ spec - fac ] , the experimental levels are not pure single - particle levels , and as such they are closer in nature to the calculated levels of the pvc model .
second , relativistic description @xcite predicts a definite connection between the dirac effective mass @xmath136 of the nucleon in the kinetic energy and the strength of the spin - orbit force because the same effective mass appears in both terms .
in addition , the spin - orbit term is sensitive to the spatial variations of the effective mass .
particle - vibration coupling affects the effective mass of the nucleon and as a result it has an impact on calculated spin - orbit splittings .
third , these results point to a new direction of improving the covariant energy density functionals .
this is because empirical energy spacings between spin - orbit partner states in finite nuclei determine a relatively narrow interval of allowed values for the dirac effective mass @xmath137 on the mean field level @xcite .
the observed impact of particle - vibration coupling on spin - orbit splittings suggests that this interval of the dirac effective mass @xmath136 may in reality be broader .
but for the spin - orbit doublets of @xmath0ni.,width=302 ] it was shown in ref .
@xcite that quasidegenerate pseudospin doublets in nuclei , discovered more than 40 years ago @xcite , arise from the near equality in magnitude of attractive scalar @xmath138 and repulsive vector @xmath139 relativistic mean fields in which the nucleons move .
pseudospin doublets have nonrelativistic quantum numbers @xmath140 and @xmath141 where @xmath49 , @xmath9 and @xmath34 are the single - nucleon radial , orbital and total angular momentum quantum numbers , respectively in the first position of the label . ] .
the pseudospin doublet can be characterized by pseudospin doublet splitting energy which is defined as the difference of the energies of pseudospin doublet members @xmath142 with this definition of @xmath143 , the low-@xmath9 ( high-@xmath9 ) member of the pseudospin doublet is higher in energy than its high-@xmath9 ( low-@xmath9 ) counterpart if @xmath144 ( @xmath145 ) .
note that the calculations fail to reproduce the relative order ( in energy ) of the members of pseudospin doublet if the signs of @xmath143 are different in the calculations and experiment .
[ pseudo ] compares available experimental data on pseudospin doublet splitting energies in @xmath0ni , @xmath13sn and @xmath11pb with pvc ( `` hybrid '' ) and best mean field ( `` def+to '' ) calculations .
note that the @xmath143 values obtained in the `` sph '' and `` def '' calculations deviate from the ones obtained in the `` def+to '' calculations by less than 200 kev ( in the majority of the cases the difference is less than 100 kev ) .
this is because both members of pseudospin doublet are located either above or below the shell gaps .
the polarization effects due to deformation and to mean fields act in the same direction for these states . as a result
, these effects cancel each other to a large degree when the difference of the energies of the members of pseudospin doublet is taken .
[ pseudo ] clearly shows that particle - vibration coupling substantially improves the description of splitting energies in pseudospin doublets ; the average deviations from experiment for @xmath143 are 0.93 mev and 0.6 mev in the mean field and pvc calculations , respectively .
there are still some doublets in which the energy splitting @xmath143 is poorly reproduced in model calculations . these are @xmath146 and @xmath147 doublets in @xmath11pb ( see figs .
[ pb208-fig ] and [ pseudo ] ) , the @xmath143 values of which deviate from experiment by 2.18 mev ( 1.66 mev ) and 2.16 mev ( 1.2 mev ) in the mean field ( pvc ) calculations , respectively .
we did not find clear explanation for such large differences .
proton and neutron pseudospin doublets with the same single - particle structure have similar splitting energies in experiment ( see pseudospin doublets indicated by the brackets with arrows in fig . [ pseudo ] ) .
this feature is rather well reproduced both in the mean field and pvc calculations for the proton and neutron @xmath148 pseudospin doublets of @xmath0ni . on the contrary , the mean field ( `` def+to '' )
calculations completely fail to reproduce this feature for the proton and neutron @xmath149 pseudospin doublets in @xmath13sn and @xmath150 pseudospin doublets in @xmath11pb ( fig . [ pseudo ] ) ; the calculated splitting energies for neutron doublets are by more than 1 mev larger than the ones for proton doublets . however , in the pvc calculations the splitting energies for these pairs of proton and neutron pseudospin doublets are similar in agreement with experiment .
this is a consequence of the fact that particle - vibration coupling decreases ( as compared with mean field calculations ) the splitting energies in neutron pseudospin doublets by more than 1 mev leaving at the same time the splitting energies in proton pseudospin doublets unchanged .
the superheavy @xmath14120 nucleus has been predicted to have proton @xmath152 and neutron @xmath153 spherical shell closures within the rmf theory @xcite .
thus , it represents the center of the `` island of stability '' of shell - stabilized superheavy nuclei .
the analysis of the deformed one - quasiparticle states in the @xmath154 mass regions in ref .
@xcite supports the presence of the @xmath152 shell gap in superheavy nuclei .
in addition , it indicates @xmath153 as a likely candidate for magic neutron number in superheavy nuclei .
this nucleus represents a challenge for future experimental synthesis since it is located at the limits of accessibility by available cold fusion reactions .
therefore , as accurate as possible estimations of its characteristics are needed from the theoretical side .
first results for its single - particle spectra obtained within the relativistic particle - vibration coupling model are presented in fig .
[ fig-120 - 172 ] .
they are compared with the results obtained on the mean field level .
note that we restrict ourselves in the case of the @xmath14120 nucleus to spherical calculations .
this is because the polarization effects due to deformation and to mean fields decrease with mass ( see sect . [ pol - eff ] ) , and thus their impact on the single - particle spectra is expected to be rather small in the @xmath14120 nucleus .
the density of the single - particle states is substantially larger than in lighter nuclei at the mean field level . as a consequence ,
this nucleus exhibits a very rich spectrum of low - lying collective phonons already within the rrpa .
for example , the lowest vibrational @xmath155 mode appears at 1.41 mev with b(e2)@xmath156 = 7.1 @xmath157 e@xmath158 @xmath159 , accompanied by several rather collective modes at 3.18 mev , 5.25 mev , 6.74 mev etc .
similar picture is obtained for other multipolarities .
although contributions to the nucleonic self - energy from the high - lying modes decrease quickly with energy , we have included phonons with all j@xmath60 values mentioned in sec .
[ cut - off ] below 15 mev , altogether about 100 phonons . fig .
[ fig-120 - 172 ] shows that the particle - vibration coupling gives rise to the general compression of the spectra . as a consequence ,
the size of the @xmath152 shell gap decreases from 3.35 mev ( in mean field calculations ) down to 2.67 mev ( in the pvc calculations ) , and the size of the neutron shell gap at @xmath153 from 2.42 mev down to 1.83 mev .
although these gaps are smaller by 20 - 25% than the ones obtained in the mean field calculations , they are still significant . as a result , the @xmath14120 nucleus still remains doubly magic nucleus representing the center of the `` island of stability '' of superheavy nuclei even when the correlations beyond mean field are taken into account .
the relativistic particle - vibration model has systematically been applied in combination with cranked relativistic mean field approach for the study of the impact of surface vibrations on single - particle motion .
the polarization effects in odd mass nuclei due to deformation and time - odd mean fields have been treated in the crmf framework .
the main results can be summarized as follows : * particle - vibration coupling has to be taken into account when model calculations are compared with experiment since this coupling is responsible for observed fragmentation of experimental levels .
the inclusion of particle - vibration coupling substantially improves the description of the energies of dominant single - particle states in @xmath13sn and @xmath11pb .
however , the accuracy of the description of single - particle spectra in @xmath0ni is similar in the mean field and pvc calculations .
note that dominant neutron single - particle states are in average better described than proton ones in the pvc calculations . * the polarization effects in odd - mass nuclei due to deformation and time - odd mean fields induced by odd particle are important . they have to be taken into account when
experimental and calculated single - particle energies are compared .
however , they are usually neglected when the results of particle - vibration coupling model calculations are compared with experiment .
the current analysis indicates that within the framework of cdft this neglect is more or less justified only for heavy nuclei , and it is more justified for proton subsystem than for neutron one . *
particle - vibration coupling leads to a shrinkage of the shell gaps .
the size of the shell gaps is typically underestimated in the pvc calculations as compared with experiment and overestimated in the mean field calculations . *
the inclusion of particle - vibration coupling decreases the accuracy of the description of spin - orbit splittings .
the analysis suggests that the mean field constraint on the allowed range of the dirac effective mass @xmath137 is too restrictive and the range of the @xmath136 values can be broader in the models which take into account the correlations beyond the mean field . *
particle - vibration coupling substantially improves the description of splitting energies in pseudospin doublets as compared with mean field calculations .
observed similarity of the splitting energies of proton and neutron pseudospin doublets with the same single - particle structure in medium and heavy mass nuclei can only be reproduced when the particle - vibration coupling is taken into account .
* the spherical shell closures in superheavy nuclei are still found at proton @xmath152 and neutron @xmath153 numbers even when particle - vibration coupling is taken into account
. however , the size of these gaps becomes smaller ( as compared with mean field values ) in the presence of particle - vibration coupling due to general compression of the spectra caused by the increase of the effective mass of the nucleon at the fermi level .
* the nl3 * parametrization employed in the current work has been adjusted at the mean - field level .
the remaining discrepancies between the results of particle vibration coupling model and experiment clearly suggest that this parametrization is not completely adequate for the description of the energies of the single - particle states .
we believe that this statement is not limited to nl3 * but it is also valid for all existing cdft parametrizations which have been adjusted at the mean - field level .
this calls for the parametrizations specifically tailored to describe single - particle degrees of freedom in the models taking into account the correlations beyond the mean field .
this work was supported by the loewe program of the state of hesse ( helmholtz international center for fair ) and by the u.s .
department of energy under the grant de - fg02 - 07er41459 .
h. langevin - joliot , j. van de wiele , j. guillot , e. gerlic , l. h. rosier , a. willis , m. morlet , g. duhamel - chretien , e. tomasi - gustafsson , n. blasi , s. micheletti , and s. y. van der werf , phys
. rev . * 47 * , 1571 ( 1993 ) .
k. l. jones , a. s. adekola , d. w. bardayan , j. c. blackmon , k. y. chae , k. a. chipps , j. a. cizewski , l. erikson , c. harlin , r. hatarik , r. kapler , r. l. kozub , j. f. liang , r. livesay , z. ma , b. h. moazen , c. d. nesaraja , f. m. nunes , s. d. pain , n. p. patterson , d. shapira , j.f .
shriner jr , m. s. smith , t. p. swan and j. s. thomas , nature * 465 * , 454 ( 2010 ) .
d. seweryniak , m. p. carpenter , s. gros , a. a. hecht , n. hoteling , r. v. f. janssens , t. l. khoo , t. lauritsen , c. j. lister , g. lotay , d. peterson , a. p. robinson , w. b. walters , x. wang , p. j. woods , and s. zhu , phys .
* 99 * , 022504 ( 2007 ) .
i. g. darby , r. k. grzywacz , j. c. batchelder , c. r. bingham , l. cartegni , c. j. gross , m. hjorth - jensen , d. t. joss , s. n. liddick , w. nazarewicz , s. padgett , r. d. page , t. papenbrock , m. m. rajabali , j. rotureau , and k. p. rykaczewski , phys . rev .
lett . * 105 * , 162502 ( 2010 ) .
|
the impact of particle - vibration coupling and polarization effects due to deformation and time - odd mean fields on single - particle spectra is studied systematically in doubly magic nuclei from low mass @xmath0ni up to superheavy ones .
particle - vibration coupling is treated fully self - consistently within the framework of relativistic particle - vibration coupling model .
polarization effects due to deformation and time - odd mean field induced by odd particle are computed within covariant density functional theory .
it has been found that among these contributions the coupling to vibrations makes a major impact on the single - particle structure .
the impact of particle - vibration coupling and polarization effects on calculated single - particle spectra , the size of the shell gaps , the spin - orbit splittings and the energy splittings in pseudospin doublets is discussed in detail ; these physical observables are compared with experiment .
particle - vibration coupling has to be taken into account when model calculations are compared with experiment since this coupling is responsible for observed fragmentation of experimental levels ; experimental spectroscopic factors are reasonably well described in model calculations .
|
there have been a number of publications analyzing citation patterns in the astronomical literature , see my previous paper for more discussion ( stanek 2008 ) . in that paper
i addressed the question of whether longer astronomical papers are cited more ( they are ) . following a number of suggestions i received after posting that seminal study , here i address the question of whether astronomical papers with more authors are also cited more .
that this would be the case is somewhat expected , for a number of reasons , but the expected size of the effect varied quite widely among my informally polled colleagues .
i therefore decided to investigate the citation vs. number of authors correlation for astronomical papers , and i found the results interesting enough to warrant this posting . in section 2
i describe the data , namely citation and number of authors statistics for about 30,000 refereed astronomical papers published between 2000 and 2004 in apj , a&a , aj and mnras . in section 3
i discuss the results , and in section 4 i conclude with some brief remarks .
i used the ads abstract serviceservice.html ] to obtain the citations , page length and number of authors for all papers published between 2000 and 2004 ( i.e. , five complete years ) in the four largest astronomical journals , a&a , aj , apj and mnras .
i start in 2000 so that page lengths will be reported for all papers and end with 2004 to allow significant time for papers to mature ( or not ) and obtain citations .
the journals were chosen so there would be plenty of papers for statistics , with the page length being about the same for each journal .
there were 12,854 apj ( including apjl and apjs ) papers , 9,275 a&a papers , 5,328 mnras papers and 2,564 aj papers , for a total of 30,021 papers .
originally , that number was somewhat larger , but i have removed all one - page papers from the list , as these were mostly errata or editorials , usually cited 0 times .
i procured these data using ads in mid - november 2009 .
as this is the same sample of papers i have used in my previous study ( stanek 2008 ) , but previously with citations accumulated until june 2008 , i will also discuss briefly the growth of citations for the whole sample between the two studies . in figures 1 and 2
i show the basic properties of our sample , such as the distribution of number of authors ( fig.@xmath11 ) and the distribution of citations ( fig.@xmath12 )
. while the distributions presented in fig.@xmath11 are more or less the same , except for aj having a larger fraction of papers with more authors , the distributions presented in fig.@xmath12 are not identical in shape .
the median number of authors per paper for the entire sample of @xmath2 papers is 3 , with a mode of 2 authors , and only @xmath3 papers with 50 or more authors . in the distribution of citations , @xmath4 papers in the sample
have 3 or fewer citations , while @xmath5 papers have 100 or more citations , and only @xmath6 papers have 1,000 or more citations .
th median number of citations for the entire sample grew from 15 in june 2008 ( stanek 2008 ) to 19 in november 2009 .
i have not tried in any way to correct for or to remove self - citations .
having the number of authors and the number of citations , i can produce a citation - authors diagram ( hereafter : cad ) , which i present in fig.@xmath13 . for display purposes ,
i have added @xmath7 to the number of citations for each paper ( `` an honorary citation '' ) , to avoid the @xmath8 problem .
also , for display purposes , i add a random variable uniformly distributed between 0 and 0.9 to both @xmath9 and the number of citations , otherwise all papers with the same number of authors and citations would produce only one dot in the cad . for bins of journals articles with a given number of authors ,
the median number of citations is shown by large open circles . as can be seen from our cad
, these medians clearly increase with the number of authors , and range from 12 for articles with 1 author to about 50 for articles with 50 and more authors .
one obvious thing to notice is that , as expected , for any given number of authors on a paper there is a very wide range in the number of citations , as seen by the 10% and 90% ranges ( filled blue dots ) .
somewhat surprisingly , barring any ads errors , there seem to exist papers with very many authors and very few citations . since i have already shown ( stanek 2008 ) that longer papers are cited more ( again , in the median sense , and as always with a large dispersion in number of citations )
, maybe what we are seeing in fig.@xmath103 is simply that having more authors also means writing more pages ? however , as can be seen in fig.@xmath104 , that is definitely not the case , in fact the median number of pages does not seem to depend at all on the number of authors and it is always from 8 to 10 pages .
additional authors clearly do not seem to be actually contributing much in terms of additional paper length , but they are clearly adding ( logarithmically ) to a paper s impact
. this additional impact could be simply due to more self - citations , but that is only a guess , which i have not investigated here .
finally , in fig@xmath105 i plot the cumulative distributions of the fraction of the total citations vs. number of authors for each of the four journals .
the curves for apj , a&a and mnras are rather similar , while the cumulative curve for aj is significantly different , meaning papers with more authors contribute a significantly larger fraction of total citations for aj vs. the other journals .
i will let the reader amuse herself / himself by speculating on or investigating the possible reasons for this difference .
in stanek ( 2008 ) i concluded with some semi - humorous publishing advice , directed especially towards first - year graduate students .
i stand by that advice , also because of the mostly positive feedback i have received after posting that study : `` i enjoyed your well written article about citations and the length of papers .
perhaps surprisingly , the longest of the @xmath11n00 papers that i have written was often cited , but not much read .
so perhaps one should look for ways to try to measure the impact of a paper rather than the number of times it is cited - clearly a difficult task . perhaps the truly great papers are the ones that still get cited after many decades .
it is a bit frightening to note that zwicky s great paper on the discovery of dark matter only started to get cited a quarter of a century after it was written . '' ; `` i wanted to say that i read your paper and really enjoyed it .
in particular , the last section where you gave ` semi - humorous ' advice .
that was the best advice i ve heard in a long time ! '' ; brilliant astro - ph submission !
i think this will be a mandatory read for any students that want to go on to grad school. ; `` thanks for your insightful and very funny paper on astro - ph .
it made my morning . '' ; but also `` cute paper on astro - ph today , but do nt you find this whole series of meta - papers to be rather cynical ?
you call the advice to first - year graduate students like myself ` semi - humorous , ' but the message seems to be telling us how to game the system . ''
i do not have enough new semi - humorous material to put in this paper , but if you send me something funny and relevant , i might add you and the advice to the paper , and perhaps together we will be cited more ( logarithmically speaking ) .
it would perhaps be of some interest to investigate how the effect found in this paper changes with time , and more generally if the statistics of number of authors has evolved in the last few decades .
this might or might not be a subject of a future study .
i thank hsiao - wen chen and chris kochanek for suggesting this particular line of inquiry .
i also thank chris kochanek for locating all these `` the ' 's and `` a ' 's that i have lost , misplaced or applied randomly .
i thank the people whose e - mails i am quoting ( i will keep them anonymous ) .
the data and plots presented in this paper are real , despite the ( mostly intended ) humorous style of writing .
this paper will not be submitted to any journal .
|
following my previous study of paper length vs. number of citations in astronomy ( stanek 2008 ) , some colleagues expressed an interest in knowing if any correlation exists between citations and the number of authors on an astronomical paper .
at least naively , one would expect papers with more authors to be cited more .
i test this expectation with the same sample of papers as analyzed in stanek ( 2008 ) , selecting all ( @xmath0 ) refereed papers from a&a , aj , apj and mnras published between 2000 and 2004 . these particular years
were chosen so that the papers analyzed would not be too `` fresh '' , but number of authors and length of each article could be obtained via ads .
i find that indeed papers with more authors published in these four major astronomy journals are on average cited more , but only weakly so : roughly , the number of citations doubles with ten - fold increase in the number of authors .
while the median number of citations for a 2 author paper is 17 , the median number of citations to a paper with 10 to 20 authors is 32 .
however , i find that most papers are written by a small number of authors , with a mode of 2 authors and a median of 3 authors , and 92% of all papers written have fewer than 10 authors . perhaps surprisingly
, i also find that papers with more authors are not longer than papers with fewer authors , in fact a median number of 8 to 10 pages per paper holds for any number of authors .
for the same sample of papers , a median number of citations per paper grew from 15 in june 2008 ( stanek 2008 ) to 19 in november 2009 . unlike stanek ( 2008 )
, i do not conclude with any career advice , semi - humorous or otherwise .
|
the miner@xmath0a collaboration emerged as a result of a joint effort between the high - energy physics and medium energy nuclear physics communities .
the main goal of the experiment is to measure neutrino - nucleus interaction cross - section with a high degree of precision .
this is an important issue for both present and future neutrino oscillation experiments .
the current available cross - section measurements have large uncertainties due to low statistics and poor knowledge of the incoming neutrino flux as can be seen on figure [ figure1 ] .
miner@xmath0a will collect large event samples for various interaction channels and will measure the cross - sections with negligible statistical errors and with the well - controlled beam systematic errors .
+ another important goal of the experiment is to study nuclear effects in neutrino interactions .
these effects such as final state interactions ( as seen on figure [ figure2 ] ) can modify the distribution of outgoing particles which will lead to the incorrect estimation of the incoming neutrino energy .
@xmath1 from @xmath0@xmath2 + fe56 interactions at 1 gev @xcite.,width=302 ] various nuclear targets incorporated into the body of the detector as well as placed in front will allow to understand processes which are absent in neutrino scattering on free nucleons .
the numi beamline at fermilab provides high - intensity neutrino beam with @xmath335e12 protons per spill and @xmath3350 kw beam power @xcite .
the layout of the facility is shown on figure [ figure2 ] .
main injector produces 120 gev proton beam that is incident on graphite target .
the resulting secondary hadrons are focused by the set of two magnetic horns . following the horns
, there is a decay pipe where the hadrons produce the neutrinos that enter the detector hall .
proton , hadron and muon monitors are installed throughout the beamline to have better control over the beam . by changing the horn polarity one can switch between neutrino and anti - neutrino modes .
beam energy can be tuned by moving the target with respect to the first horn . as a result , pions and kaons of different energies reach the focusing region .
figure [ figure4 ] illustrates this technique where different target - horn geometries are shown on the left with the resulting p@xmath4/p@xmath5 distributions of the produced hadrons on the right .
it is known that the largest source of uncertainty in the predicted neutrino flux at the detector comes from the hadron production at the target .
our studies show that there is about 10% spread in the peak of neutrino spectrum when the hadron production models are varied while keeping the geometry the same .
miner@xmath0a plans to use several techniques to reduce the flux - associated errors .
first approach is to use the latest , more accurate external hadron production data .
however , care must be taken since most of the available experimental data is for thin targets while numi target is two interaction lengths .
in addition , one needs to take into account downstream interactions of the produced hadrons .
we will also use the data from three muon monitors ( shown on figure [ figure4 ] ) that are located at different distances from the target and thus sample different regions of neutrino energy spectrum . finally , to better constrain hadron production at the target ,
a series of beam runs were taken with varying target position and horn current .
then , hadron production yields in mc are tuned to match the real data .
the result is a set of weights in each of p@xmath4-p@xmath5 bins that is applied to mc pion yields .
the outcome of this technique is shown on figure [ figure5 ] where pre - fit errors are shown in blue and post - fit are in red .
miner@xmath0a is a fine - grained scintillator - strip detector with electromagnetic and hadronic calorimetry regions .
there are approximately 30,000 scintillator channels grouped into inner and outer regions . for details on general detector structure ,
see @xcite . + for studying nuclear effects in neutrino interactions , various targets are incorporated into the body of the detector with interspersed layers of active scintillator .
there is a liquid helium target located in front of the detector . the location of nuclear targets in shown on figure [ figure6 ] .
all the targets are installed except for water target which is planned for installation later this year .
also , the collaboration has prepared a proposal for introducing additional nuclear target : deuterium for high - precision nuclear - to - d ratio measurements and comparison with the available charged lepton data as well as various theoretical models - that will replace the he .
a detector.,width=510 ] the complete list of the targets with fiducial volumes and event estimates ( with genie 2.6.2 ) is given in table [ table1 ] .
.miner@xmath0a nuclear targets with event estimates . [ cols="<,^,^ " , ] [ table1 ]
for the initial analysis , the most downstream target ( fe / pb ) was chosen due to the ease of event reconstruction ( no passive material in track s path ) .
charged current inclusive @xmath0@xmath2 events were selected with one muon track matched to the minos near - detector that we use to measure the muon momentum and charge . low - energy neutrino mode ( see fig.4 ) was used with 0.9e20 and 11.2e20 protons - on - target for data and mc , respectively .
the fiducial volume was a hexagon with 85 cm apothem with z position of muon vertex ( reconstructed origin of muon track ) in the nuclear target or the first module downstream .
there is also a requirement of no muon - like activity upstream of the target .
the resulting muon tracks are extrapolated backward into the target region and based on the projected location the event is classified as the one from iron or lead . at this analysis stage , the events were not background subtracted ( the background that comes from the inability to resolve whether the event originated in the nuclear target or the first scintillator module downstream ) so the samples are named `` iron - enriched '' and `` lead - enriched '' .
figure [ figure7 ] shows the event misidentification due to incorrect reconstruction of the vertex where red triangles are true events in lead that were reconstructed in iron and vice versa for lead with blue triangles .
as can be seen from the figure , vertex reconstruction works properly with only a few events being incorrectly vertexed .
one of the scientific goals of nuclear target program of miner@xmath0a experiment is to compare various reconstructed quantities ( track multiplicities , etc ) between nuclear targets and the active scintillator volume of the detector . for that purpose , a region with 4 scintillator modules ( plastic reference target )
was selected downstream of all the targets with the imaginary divide as in the last fe / pb target .
this analysis is in progress and as a cross - check , the ratio of lead to iron plastic reference targets is shown on figure [ figure9 ] .
the reason this ratio is less than one is due to the fact that these two reference volumes occupy different regions in x - y plane and thus have different acceptance values for muons propagating into minos detector .
the ratios among actual nuclear targets will be available in the nearest future .
final analysis will be done with approximately four times more protons - on - target and four times more fiducial mass which will provide a highly - precise measurement .
the support for this work was provided by the fermi national accelerator laboratory operated by the fermi research alliance , llc , under contract no . de - ac02 - 07ch11359 with the united states department of energy .
construction support also was granted by the united states national science foundation under nsf award phy-0619727 and by the university of rochester .
|
the miner@xmath0a experiment is aimed at precisely measuring the cross - sections for various neutrino interaction channels .
it is located at fermilab in the underground cavern in front of minos near detector .
miner@xmath0a is a fine - grained scintillator with electromagnetic and hadronic calorimetry regions .
there are various nuclear targets located inside and in front of the detector for studying nuclear medium effects in neutrino - induced interactions .
the installation was completed in march 2010 and since then the detector has been collecting data . in this paper , the method for determining the neutrino flux is described in detail with the associated uncertainties as well as the techniques for their reduction .
the general structure of the detector is given with the emphasis on the nuclear targets region .
preliminary results related to nuclear effects studies are presented followed by their discussion and future plans .
|
the cold dark matter model with a cosmological constant ( @xmath1cdm ) is well established enough ( see e.g. , @xcite,@xcite , @xcite,@xcite ) that individual large - scale simulation efforts can be carried out that focus on just this one cosmology .
we have also reached the point at which supercomputers enable numerical modeling of cosmological volumes with enough resolution to study the properties of individual galaxies . in this paper
we report on a p - gadget hydrodynamic simulation of @xmath10 cubic volume , the massiveblackii simulation .
it has @xmath11 mass resolution , cooling , star formation , black holes and feedback , and represents the evolution of a @xmath1cdm universe to redshift @xmath0 .
numerical simulations ( see reviews by @xcite , @xcite , ) are the tool of choice to address many questions in cosmology , as galaxy formation is a complex non - linear problem .
two criteria which must be satisfied for accurate results are : \(1 ) a large enough simulation volume that fourier density modes on the largest scales are evolving independently .
the volume simulated must be a representative region of the universe , otherwise inferences drawn from it regarding such questions as clustering and the mass function of objects will be incorrect ( @xcite , @xcite , @xcite ) . when @xmath1cdm models are evolved to redshift @xmath0 , then a volume of at least @xmath12 on a side becomes necessary to predict the overall star formation rate , for example ( @xcite ) .
\(2 ) high enough mass and spatial resolution that the properties of the objects of interest have converged .
this requires many resolution elements ( particles or grid cells ) . if we focus on particle based simulations relevant to the current work , at the very least for identification of objects there must enough particles to overcome shot noise .
if we require detailed properties of the objects such as galaxy spectra , or angular momenta then this can require many more ( e.g. , @xcite ) .
these criteria of large simulation volume and high resolution are more straightforward to address in the context of more restricted physical modeling . as a result , dark matter and gravity - only simulations
have long been used to make cosmological predictions that cover both large and small scales in the same volume ( e.g. , @xcite , @xcite , @xcite ) .
semi - analytic modeling has been used to process dark matter simulations , resulting in many studies of the galaxies and their properties in the @xmath1cdm model ( see e.g. , @xcite , @xcite and references therein ) .
baryonic physics including hydrodynamics obviously plays an important role in the formation of luminous objects and structure .
this has lead to the inclusion of the relevant equations in simulation codes in many forms , smoothed particle hydrodynamics ( sph , @xcite ) , eulerian grid solvers ( e.g. , @xcite , @xcite ) and hybrid lagrangian / eulerian schemes ( e.g. , @xcite ) .
although previous work has not simultaneously reached the combination of large volume and high resolution that we present here , research has progressed using many methods , including making use of zoom simulations of smaller volumes inside a representative one ( @xcite , @xcite ) , simulations that stop at high redshifts before large - scale modes become non - linear ( e.g. , @xcite ) , or by tackling problems which require lower mass resolution ( e.g. , @xcite ) . the advent of large - scale computing facilities with 100,000 compute cores or more ( such as the cray xt5 , `` kraken '' on which the current simulation was run ) and the development of highly efficient distributed memory simulation codes ( such as gadget2 , @xcite ) means that simulations which satisfy both criteria ( 1 ) and ( 2 ) are now possible .
we have run one such simulation as part of the nsf petascale applications in cosmology program , using the code p - gadget ( see e.g. , @xcite ) .
our aim was to simulate and analyze a large , representative volume of the @xmath1cdm model with the most important physical processes previously included in zoom runs or simulations with smaller boxes .
these are hydrodynamics ( using sph ) , cooling , a subgrid multiphase model for star formation ( @xcite ) and subgrid black hole modeling ( @xcite,@xcite ) , both with feedback .
our use of the physical modeling and algorithms used in previous work such as @xcite , @xcite , @xcite and @xcite enables continuity and therefore comparison with this previous work .
our aim is to see what this `` fiducial '' model ( @xmath1cdm + gadget sph + the particular subgrid algorithms employed ) predicts about the properties of galaxies , their halos and their clustering at redshifts extending down to the present day .
we have not adjusted methods , algorithms and parameters used in previous work ( e.g. , @xcite ) to try to tune to observational results .
our goal is to see how this model performs , now that there is a large volume at high resolution .
we naturally expect both regions of agreement and disagreement with observations and we aim that our work will offer guidance to future work to address the problems . in this paper , we make the first use of the massiveblackii simulation evolved to @xmath0 , and use it to explore some topics in structure formation . here we choose to focus on topics relevant to galaxy and agn formation and large - scale structure , including mass functions , galaxy and halo properties and clustering .
our emphasis is on lower redshifts ; the simulation at redshifts @xmath13 has been explored by @xcite , @xcite , and @xcite .
topics that we leave to future work include the intergalactic medium , absorption lines , galaxy clusters and x - ray emission .
our plan for the paper is as follows . in section [ sec_methods ]
we briefly describe the numerical methods and algorithms used to run the simulation , select galaxies and carry out stellar population synthesis . in section [ sec_vis ] , we describe visualization of the simulation . in sections [ sec_massfn ] and [ sec_hod ]
respectively we present the mass function and halo occupation distribution and in section [ sec_gal_clustering ] we examine the clustering of dark matter and galaxies . the properties of galaxies and supermassive black holes are examined in sections [ sec_galaxy ] and [ sec_bh ] and we derive some conclusions from our work in section [ sec_conc ] .
we have used p - gadget , an upgraded version of gadget3 ( see @xcite for an earlier version ) which we are developing for use at upcoming petascale supercomputer facilities .
this code was also used to run the massiveblack ( mb ) simulation ( @xcite . both mb and mbii are cosmological simulation of a @xmath1cdm cosmology .
the major differences between mb and mbii are resolution and volume .
however there are minor differences in cosmology between the two .
the initial conditions for mbii were generated with the cmbfast transfer function at @xmath14 and the simulation was evolved to @xmath0 .
the cosmological parameters used were : amplitude of mass fluctuations , @xmath15 , spectral index , @xmath16 , cosmological constant parameter @xmath17 , mass density parameter @xmath18 , baryon density parameter @xmath19 and @xmath20 ( hubble s constant in units of @xmath21 ) .
these are consistent with the wmap7 cosmology @xcite .
.basic simulation parameters for the simulation .
the columns list the size of the simulation box , @xmath22 , the number of particles ( dark matter + gas ) used in the simulation , @xmath23 , the mass of a single dark matter particle , @xmath24 , the initial mass of a gas particle , @xmath25 , and the gravitational softening length , @xmath26 .
all length scales are in comoving units . [ cols="^,^,^,^,^ " , ]
in this paper we have examined a variety of standard predictions using the recently completed mbii simulation . in particular , the results presented here range from the clustering of halos and galaxies , hod , the mass function of halos to basic properties of galaxies and agns .
our main conclusions are : * we find that halos with masses @xmath27 have a baryon fraction close to 80%-90% of the cosmic mean .
the baryon fraction decreases steadily with decreasing halo mass .
our results are broadly consistent with @xcite for cluster sized halos but differ significantly for lower mass halos .
the discrepancy can be attributed to feedback both from agn and supernovae which were not been modeled in @xcite .
* we find that the fof halo mass function ( where halo mass includes both dark matter and baryonic components ) in mbii can be fit with a universal form ( valid for all redshifts ) at the @xmath28 level .
* one of the most striking results predicted by mbii is the behavior of the halo mass function which shows a strong suppression in the abundance of halos below the knee of the mass function ( @xmath29 which corresponds to @xmath30/h at @xmath0 and @xmath31/h at @xmath32 ) when compared to dark matter only predictions of the halo mass function .
this is due to the significant impact of baryonic processes which tend to suppress the mass of the halo by up to 30% @xcite .
* fits to the mass function from dark matter only simulations overpredict the mass function at the 20 - 35% level below the knee of the mass function at @xmath33 .
* we have quantified the scale dependence of bias and stochasticity in the simulation .
we find that scale of the deviation from a linear fit for both bias and stochasticity reaches a minimum at @xmath34 .
the galaxy and mass density field can in this sense be said to trace each other most closely at this redshift . *
the mbii overall matches most observed measurements of galaxy clustering with stellar mass .
there are however some discrepancies in the bias at the lowest masses and redshifts probed .
the simulation is consistent with observed measurements of stochasticity from combinations of galaxy clustering and weak lensing .
* we find that the hod is well described by a power law behavior ( see equation [ occfit ] ) .
we find a modest evolution for the power law slope however the normalization mass for the distribution exhibits an exponentially decaying behavior with redshift . *
the location of the peak and the width for the radial distribution of satellite galaxies decrease with decreasing redshift irrespective of the host mass of the halo , showing that satellite galaxies cluster more strongly around the central galaxy with time . *
we find that the shape of the csed in mbii is consistent with observed data .
the amplitude is however much lower and can be attributed to incompleteness at low galaxy masses in mbii . *
the gsmf predicted by mbii is consistent with observations out to @xmath35 . at lower redshifts
however the mbii gsmf is much steeper at lower masses whereas mbii over produces stars at larger masses at @xmath0 .
* we find black hole mass and agn luminosity to be broadly correlated to host @xmath36 but with very large scatter , indicating a wide range of black hole properties for a given halo mass .
however , the local @xmath37 relation is tighter and consistent the observed one .
our results also imply very moderate evolution for the @xmath37 from @xmath38 .
the global star formation rate density and black hole accretion rate density are also similar , but with a peak for the latter shifted to later times . *
the bolometric luminosity function of the simulated agn population is broadly consistent with observational constraints ( * ? ? ?
* see also ) .
we note however that simple thermal coupling for agn feedback appears to quench the bright end of the agn lf too fast around @xmath39 whilst not being sufficient to fully quench the brightest agn at @xmath0 .
the best agreement with observations is at @xmath40 and @xmath41 .
our results show a weak dependence of agn clustering with luminosity .
( note that we defer to future work we to carry out detailed analysis of agn population ) .
the mbii simulation is the largest simulation of its kind run to date with sufficient resolution to resolve @xmath42/h halos .
we have found that the properties , such as hod and clustering are consistent with previous work and observations .
however we find that feedback from agns is still not sufficient at lower redshifts to properly account for the properties of large galaxies and agns since it is unable to quench star formation in massive galaxies and the abundance of luminous quasars .
however it describes the current state of standard sph simulations of galaxy formation and should be used as a testbed for improving models of galaxy formation .
the parameters used in mbii were based on much smaller sph simulations which missed these large halos in their volumes and thus could not find discrepancies with observations at the large mass end . at smaller masses
the gsmf in mbii is very steep compared to observational constraints .
a variable wind model based on the galaxy velocity dispersion @xcite has been shown to flatten the gsmf at smaller masses to better reproduce observational constraints at @xmath0 .
however at higher redshifts these models still predict a steep slope . on the other hand we find that agn feedback is still insufficient to quench star formation in the most massive galaxies .
the effect of agns is important in regulating the growth of massive galaxies ; e.g. the results of @xcite ( which do not include agn feedback ) and those of @xcite ( which include agn feedback ) clearly illustrate how agns affect the stellar content of massive galaxies .
however these simulations are still small in volume and miss out on massive halos .
it would be interesting to see their predictions when run on boxes similar in size to mbii . in subsequent work we will look at improving on the feedback models employed in mbii .
one of the important effects we have quantified here is the effect of baryonic processes on the halo mass function .
although we find that the universality of the mass function in mbii holds at the 13% level consistent with previous work , we find considerable differences in the abundance of halos below the knee of the mass function when compared to dark matter simulations .
one of the natural questions which arise are : how do different models for star formation , black hole growth and feedback affect the mass function ? how strong is the redshift dependence ? what can be said of halo abundance matching techniques and their predictions when done with mass functions such as mbii ?
these questions can only be addressed with simulations with different galaxy formation models and corresponding dark matter simulations which we will address in future work .
the simulations were run on the cray xt5 supercomputer kraken at the national institute for computational sciences .
this research has been funded by the national science foundation ( nsf ) petaapps programme , oci-0749212 and by nsf ast-1009781 .
nk would like to aknowledge useful discussions with michael boylan - kolchin and erin sheldon .
|
we investigate the properties and clustering of halos , galaxies and blackholes to @xmath0 in the high resolution hydrodynamical simulation massiveblack - ii ( mbii ) .
mbii evolves a @xmath1cdm cosmology in a cubical comoving volume of @xmath2 and is able to resolve halos of mass @xmath3/h .
it is the highest resolution simulation of this size which includes a self - consistent model for star formation , black hole accretion and associated feedback .
we provide a simulation browser web application which enables interactive search and tagging of halos , subhalos and their properties and publicly release our galaxy catalogs to the scientific community .
our analysis of the halo mass function in mbii reveals that baryons have strong effects , with changes in the halo abundance of 20 - 35% below the knee of the mass function ( @xmath4/h at @xmath0 ) when compared to fits based on dark matter only simulations .
we provide a fitting function for the halo mass function valid for the full range of halo masses in mbii out to redshift @xmath5 and discuss how the onset of non - universal behavior in the mass function limits the accuracy of our fit .
we examine the halo occupation distribution of satellite galaxies and present results valid over 5 orders of magnitude in host halo mass .
we study the clustering of galaxies , and in particular the evolution and scale dependence of stochasticity and bias .
comparison with observational data for these quantities for samples with different stellar mass thresholds yields reasonable agreement .
using population synthesis , we find that the shape of the cosmic spectral energy distribution predicted by mbii is consistent with observations , but lower in amplitude . the galaxy stellar mass function ( gsmf )
function is broadly consistent with observations at @xmath6 . at @xmath7 ,
observations probe deeper into the faint end and the population of passive low mass ( for @xmath8 ) galaxies in the simulation makes the gsmf too steep . at the high mass end ( @xmath9 )
galaxies hosting bright agn make significant contributions to the gsmf .
the quasar bolometric luminosity function is also largely consistent with observations .
we note however that more efficient agn feedback ( beyond simple thermal coupling used here ) is likely necessary for the largest , rarest objects / clusters at low redshifts .
methods : numerical cosmology : theory cosmology : large - scale structure of universe galaxies : formation galaxies : evolution quasars : general
|
the large scale structure of the universe was arisen from initial fluctuations through the nonlinear evolution of gravitational instability .
gravitational interaction is of long range , and therefore , the evolution of cosmic clustering is not localized in physical space .
the typical processes of cosmic clustering , such as collapsing and falling into potential wells , the fourier mode - mode coupling and the merging of pre - virialized dark halos , are generally _ non - local_. these processes lead to a correlation between the density perturbations at different positions , even if the perturbations at that positions initially are statistically uncorrelated . for instance , in the zeldovich approximation ( zeldovich 1970 ) , the density field @xmath3 at ( eulerian ) comoving position @xmath4 and time @xmath5 is determined by the initial perturbation at ( lagrangian ) comoving position , @xmath6 , plus a displacement @xmath7 : @xmath8 the displacement @xmath9 represents the effect of density perturbations on the trajectories of self - gravitating particles .
the intersection of particle trajectories leads to a correlation between mass fields at different spatial positions .
thus , the gravitational clustering is non - local even in weakly non - linear regime .
on the other hand , spatial locality has been employed in the gaussianization technique for recovery of the primordial power spectrum ( narayanan & weinberg 1998 ) .
underlying this algorithm is to assume that the relation between the evolved mass field and the initial density distribution is local , i.e. the high(low ) initial density pixels will be mapped into high(low ) density pixels of the evolved field ( narayanan & weinberg 1998 ) . obviously , the localized mapping is difficult in reconciling the initially gaussian field with the coherent non - linear structures , such as halos with scaling behavior .
it has been argued that the locality assumption may be a poor approximation to the actual dynamics because of the non - locality of gravitational evolution ( monaco & efstathiou 2000 ) .
nevertheless , the localized mapping is found to work well for reconstructing the initial mass field and power spectrum from transmitted flux of the ly@xmath10 absorption in qso spectra ( croft et al .
1998 . ) these results , joint with the data of wmap , have been used to determine the cosmological parameters ( spergel et al 2003 ) .
however , the dynamical origin of the locality assumption remains a problem .
it is still unclear under which condition the localized mapping is a good approximation .
this problem has been studied in weakly non - linear regime under the zeldovich approximation .
the result showed that the cosmic gravitational clustering evolution is spatially quasi - localized in phase ( @xmath0-@xmath1 ) space made by the dwt decomposition ( pando , feng & fang 2001 ) . in this approach
, each perturbation mode corresponds to a cell in the @xmath0-@xmath1 space , ( @xmath0 to @xmath11 , @xmath1 to @xmath12 ) with @xmath13 , and the density perturbation of the mode @xmath14 is @xmath15 .
they demonstrated that , in the zeldovich approximation , if the initial density perturbations in each cells are statistically uncorrelated , i.e. @xmath16 , where @xmath17 denotes for kronecker delta function , the evolved @xmath18 and @xmath19 will keep approximately spatially uncorrelated always , @xmath20 , which is just what we call spatial quasi - locality in the dynamics of cosmic clustering . the spatial quasi - locality implies a significant local mode - mode coupling ( different scales @xmath1 at the same position @xmath0 ) , but very weak non - local coupling between the modes ( different positions @xmath0 ) .
the non - linear dynamical evolution is well developed along the direction of @xmath21-axis , rather than @xmath4-axis in the phase space .
this quasi - locality has been justified in weak nonlinear samples such as the transmitted flux of qso ly@xmath10 absorption spectrum ( pando , feng & fang 2001 ) .
it places the dynamical base of recovering initial power spectrum from corresponding weakly evolved field via a localized mapping in phase space ( feng & fang 2000 ) .
this paper is to extend the concept of quasi - locality to fully nonlinear regime .
we try to show that the quasi - locality of the cosmic clustering in phase space holds not only in the weak nonlinear regime , but also in nonlinear evolution .
since the nonlinear cosmic density field can be expressed by the semianalytical halo model ( e.g. cooray & sheth 2002 , and references therein ) , our primary interest is to study whether the quasi - locality could be incorporated in the halo model .
we will first analytically derive the quasi - locality from the halo model , and then make a numerical test using high resolution @xmath22-body simulation samples .
the outline of this paper is as follows .
2 presents the statistical criterion of the quasi - local evolution of a density clustering in the @xmath0-@xmath1 phase space .
3 shows that the density field evolution might be spatially quasi - localized in the phase space if the cosmic density field can be described by the halo model .
numerical tests on these predictions with n - body simulation samples are made in 4 .
finally , the conclusions and discussions will be given in 5 .
in physical space ( @xmath4 ) , the mode is dirac delta function @xmath23 , and cosmic mass density field variable is @xmath24 , while in scale space ( @xmath21 ) , the mode is the fourier bases @xmath25 , and the field variable is @xmath26 , which is the fourier transform of @xmath24 . in hybrid
@xmath0-@xmath1 phase space , one can use the complete and orthogonal bases of the discrete wavelet transform ( dwt ) as the mode function . the mass density field
is then described by the dwt variables . without loss of generality
, we introduce the dwt variables by considering a density field @xmath24 in a cubic box of @xmath27 , @xmath28 and volume @xmath29 .
we first divide the box into cells with volume @xmath30 , where @xmath31 . for a given @xmath32
, there are @xmath33 cells labelled by @xmath34 , and @xmath35 . the cell @xmath36 occupies the spatial range @xmath37 , @xmath28 .
accordingly , indexes @xmath38 and @xmath39 denote for , respectively , the scale and the position of the cells . in each dimension
, we have @xmath40 and @xmath41 , i.e. @xmath42 , or the volume of all cells in the @xmath0-@xmath1 space is @xmath43 .
each cell @xmath44 supports two compact functions : the scaling function @xmath45 and the wavelets @xmath46 ( daubechies 1992 , fang & thews , 1998 , fang & feng , 2000 ) .
both @xmath45 and @xmath46 are localized in cell ( @xmath47 ) .
the scaling functions @xmath45 are orthogonormal with respect to index @xmath39 as @xmath48 the scaling function @xmath45 is a low pass filter at cell @xmath44 . the scaling function coefficient ( sfc ) of the density field is defined by @xmath49 which is proportional to the mean density in cell @xmath44 .
the wavelets @xmath46 are orthogonormal with respect to both indexes @xmath38 and @xmath39 @xmath50 the wavelets @xmath51 form a complete and orthogonal base ( mode ) in the phase space .
therefore , the density field can be described by the wavelet function coefficients ( wfcs ) defined as @xmath52 the wfcs @xmath53 are the dwt variables of the density field .
the dwt variables @xmath53 is the fluctuation of the density field around scales @xmath54 , @xmath55 located at the cell @xmath39 with size @xmath56 . since the wavelet @xmath46 is a band pass filter , in each dimension , @xmath53 is a superposition of fluctuations filtered in the waveband @xmath57 , where @xmath58 , and @xmath59 .
this decomposition of the fluctuation is optimized in sense that the size of cell @xmath60 is adaptively chosen to match with the perturbations at a given wavenumber .
the dwt bases generally have vanishing moments , i.e. @xmath61 where @xmath62 and @xmath63 .
@xmath64 is dependent on wavelet .
for the wavelet daubechies @xmath65 , we have @xmath66 .
thus , the fourier transform of wavelet @xmath67 has a compact support in the wavenumber space @xmath68 . from eq.(6 )
we have @xmath69 for all @xmath47 , and thus , @xmath70 where the density contrast @xmath71/\bar{\rho}$ ] , @xmath72 is the mean density . since @xmath73 , we have also @xmath74
because the set of wavelets is complete , @xmath53 give a complete description of the density field @xmath24 , i.e. one can reconstruct @xmath24 or @xmath75 in terms of variables @xmath53 as @xmath76 the initial density perturbation of the universe @xmath77 is believed to be a gaussian random field with correlation matrix of the fourier variables @xmath78 @xmath79 and all higher order cumulant moments of @xmath78 vanish . the function @xmath80 of eq.(10 )
is the initial power spectrum .
the kronecker delta function @xmath81 in eq.(10 ) indicates that the initial perturbation for each mode @xmath21 is independent , or localized in @xmath1-space .
generally , if the initial fourier power spectrum , @xmath80 , is colored , i.e. @xmath1-dependent , the correlation matrix of variables other than the fourier mode will no longer be diagonal . for instance
, correlation function in @xmath0-space will be @xmath82 , which is the fourier counterpart of @xmath80 .
however , the correlation matrix of the dwt variables of a gaussian field is always diagonal or quasi - diagonal , regardless the fourier power spectrum @xmath83 is white or colored .
that is , the correlation function of @xmath53 is localized with respect to @xmath44 as @xmath84 where @xmath85 in eq.(11 ) is the dwt power spectrum of the field .
the reason of the diagonality of eq.(11 ) is as follows .
first , the wfc @xmath53 is given by a linear superposition of the fourier modes @xmath86 in the waveband around @xmath87 ( @xmath28 ) , @xmath53 with different @xmath38 consist of the fourier modes @xmath86 in different @xmath21 bands .
while for gaussian fields , the fourier modes in different wave bands are uncorrelated in general [ eq.(10 ) ] , and therefore , there might be no correlation between the dwt modes of @xmath38 and @xmath88 if @xmath89 .
this yields the quasi - locality of @xmath90 .
second , the phases of the fourier modes of gaussian field are independent and random . for a superposition of
the random phased fourier modes @xmath86 in the band from @xmath21 to @xmath91 , the spatial correlation length can not be larger than that given by the uncertainty relation @xmath92 .
moreover , the non - zero regions of two dwt modes @xmath93 and @xmath94 with @xmath95 have spatial distance @xmath96 .
consequently , all off - diagonal elements @xmath95 vanish or are much smaller than diagonal elements , i.e. , @xmath97 thus , the correlation function of the dwt variables of a gaussian field is rapidly decaying when @xmath98 and @xmath99 .
the diagonal correlation function described by eq.(11 ) is a generic feature of a gaussian field in the dwt representation .
it should be pointed out that the wfc correlation function eq.(11 ) is different from the ordinary two point correlation function @xmath100 .
the former is the correlation between two modes in @xmath2 phase space , while the later is for two modes in @xmath0-space .
explicitly , eq.(11 ) describes the correlation of perturbation modes in the waveband @xmath101 between positions @xmath39 and @xmath102 , and so , it is sensitive to the phases of modes .
the ordinary two point correlation is not sensitive to the phase of perturbations . in the dwt analysis ,
an analogue of the ordinary two point correlation function is defined by by the correlation between sfcs , i.e. @xmath103 . since the scaling function @xmath104 is a low - pass filter on scale @xmath38 and at position @xmath39 , the correlation function @xmath105 behaves in a similar way as @xmath106 , where @xmath107 is a filtered density field smoothed on the scale @xmath108 .
however , as @xmath93 is a high - pass filter , the wfc covariance @xmath109 shows quite different statistical features from the sfc correlation , e.g. , it is always quasi - diagonal or even fully diagonal for a gaussian field .
generally , it has been shown for many analytically calculable random fields that the sfc correlation is significantly off - diagonal , while the wfc correlation is _ exactly _ diagonal ( greiner , lip & carruthers , 1995 ) moreover , within a given volume in the @xmath0-@xmath1 space , such as @xmath110 , the number of the dwt modes @xmath111 is the same as that of the fourier modes @xmath21 .
accordingly , @xmath112 can be expressed as a linear superposition of @xmath113 , and vice versa .
equivalently , the fourier power spectrum can be replaced by the dwt power spectrum ( fang & feng , 2000 ) .
if the evolution of the comic mass field is localized , the evolved density field @xmath75 at a given spatial point is determined only by the initial density distribution @xmath114 at the same point .
as emphasized in 1 , this locality is inconsistent with the non - local behavior of gravitational clustering .
however , the evolution of the comic mass field can be quasi - localized in sense that the correlation between the dwt variables of the evolved field is always spatial diagonal if the initial correlation function is diagonal , such as eq.(11 ) . for perturbation modes in a waveband @xmath101 ,
the quasi - localized range is @xmath115 .
a quasi - localized evolution means that the auto - correlation function of the dwt variables is always diagonal , or quasi - diagonal @xmath116 when @xmath117 , if it is diagonal initially .
thus , one may place a statistical criterion for the quasi - locality as @xmath118 where @xmath119 this is a normalized correlation function of the dwt modes , i.e. @xmath120 . the auto - correlation function of the dwt variables , @xmath121 , measures the correlations between the perturbation modes on scales @xmath122 and @xmath123 at two cells with a vector distance given by @xmath124 . in the case of @xmath125 , @xmath126 gives the correlation between fluctuations on scale @xmath127 and @xmath128 at the _ same _ physical area .
therefore , if condition ( 13 ) holds for all redshifts , the dynamical evolution of the mass field is basically spatial localized in the dwt bases . comparing the condition eq.(13 ) with eq.(12 ) , we see that the cosmic field undergoing a local evolution is different from its gaussian predecessor by the factor @xmath90 , but not @xmath129 . in other words , the evolution leads to the significant scale - scale @xmath130 coupling , rather than modes at different locations @xmath117 .
one can also construct the criterions for the quasi - locality using higher order correlations among the dwt variables @xmath53 .
for instance , a @xmath131 order statistical criterion is given by @xmath132 where @xmath133 where @xmath134 and @xmath135 can be any even number .
obviously , @xmath136 for gaussian fields .
@xmath137 corresponds to a local scale - scale correlation , while @xmath138 a nonlocal scale - scale correlation .
the quasi - local evolution of cosmic mass field requires that nonlocal scale - scale correlation is always small .
it should be pointed out that the statistical conditions eqs.(13)-(16 ) are not trivial because the dwt basis does not subject to the central limit theorem .
if a basis subjects to the central limit theorem , the corresponding variables will be gaussian even when the random field is highly non - gaussian . in this case ,
eqs.(13 ) and ( 15 ) may be easily satisfied , but it does not imply that the evolution is localized , or quasi - localized . statistical measure
subjected to the central limit theorem is unable to capture non - gaussian features of the evolution .
we will show , in this section , that the statistical criterions of 2.3 are fulfilled if the cosmic mass field can be described by the halo model .
the cosmic clustering is self - similar and hierarchical , as the dynamical equations of collisionless particles ( dark matter ) do not have preferred scales , and admits a self - similar solution as well as the initial density perturbations are gaussian and scale free .
the halo model further assumes that all mass in a fully developed cosmic mass field is bound in halos on various scales ( neyman & scott 1952 , scherrer & bertschinger 1991 ) .
thus , the cosmic mass field in non - linear regime is given by a superposition of the halos @xmath139 where @xmath140 is the density profiles of halo with mass @xmath141 at position @xmath142 , and @xmath143 is the density profile normalized by @xmath144 .
there are several different versions of the halo density profiles , such as @xmath145^{\beta}$ ] with @xmath146 ( navarro , frenk & white 1996 ) , and @xmath147 $ ] with @xmath148 ( moore et al .
a common feature of the halo density profiles is self - similar , which implies that the indexes @xmath10 and @xmath149 should be mass - independent .
the mass dependence is only given by @xmath150 , which characterizes the size of the @xmath141 halo .
the details of the profiles are indifferent for the problem we try to study below .
what is important for us is only that one can set a self - similar upper limit to the normalized halo density profile as @xmath151 where @xmath152 , @xmath153 is a constant and the index @xmath154 is mass - independent
. the @xmath141-dependence of @xmath150 are not stronger than a power law as @xmath155 .
the halo model also assumes that the halo - halo correlation function on scales larger than the size of halos is given by the two - point correlation functions of the linear gaussian field with a linear bias correction , or by the correlation functions of quasi - linear field .
therefore , no assumption about higher order halo - halo correlations on large scales is needed . in this model , the time - dependence of the field is mainly given by the mass function of halos , @xmath156 , which is the number density of the halos with mass @xmath157 at time @xmath5 . in the press & schechter formalism ( 1974 ) ,
the mass function is determined by the power spectrum of the initial gaussian density perturbation . moreover , the self - similarity of the halo density profiles insure that eq.(18 ) holds for all time .
the cosmic evolution only leads to the parameters on the r.h.s of eq.(18 ) to be time - dependent . with eq.(17 )
, the dwt variable of the cosmic mass field in the halo model is given by @xmath158 the auto - correlation function of the dwt variables is then @xmath159 where the first and second terms on the r.h.s .
are usually called , respectively , 1- and 2-halo terms .
they can be written in the explicit form , @xmath160 @xmath161 where @xmath162 is the number density of halos with mass @xmath157 , and @xmath163 is the halo - halo correlation function .
we show now that the dwt correlation function @xmath121 is quasi - diagonal , or fast decaying with respect to the spatial distance @xmath164 .
first , consider the 2-halo terms eq .
according to the halo model , the two - point correlation function @xmath165 is determined by the linearly gaussian density field .
in fact , as having been discussed in 2.2 , the dwt correlation function of a gaussian field is generally diagonal [ eq.(12 ) ] , regardless the fourier power spectrum is white or colored
. therefore , the dwt correlation function @xmath121 contributed from the 2-halo term should be quasi - local .
the dwt integral in the 2-halo term of eqs.(22 ) is not completely the same as eq.(12 ) , as the halo - halo correlation function in eq.(22 ) is @xmath166 , while in eq.(12 ) is @xmath167 . for halos with size
less than the scale @xmath38 considered , the factor @xmath168 in correlation function @xmath169 is smaller than the size of the dwt mode , @xmath170 , and so , the factor @xmath171 can be ignored in comparison with the variable @xmath172 . moreover , by definition of the halo model , the correlation functions @xmath169 does not include the contributions from halos with sizes larger than @xmath173 .
thus , the cases of @xmath174 can always be ignored .
the 2-halo term essentially follows eq.(12 ) , and is always approximately diagonal with respect to spatial indexes @xmath39 and @xmath102 . to analyze the 1-halo term eq.(21 ) , we use the following theorem of wavelets ( tewfik & kim , 1992 ) .
because the dwt bases is self - similar , for any 1-d power law function @xmath175 , we have @xmath176 where @xmath177 is wavelet in 1-d space , and @xmath178 is a constant . therefore , while using wavelets with large enough @xmath64 [ eq.(5 ) ] , the integral eq.(23 ) is quickly decaying with the spatial distance @xmath179 .
in other words , besides two nearest position of @xmath180 , there is no correlation between modes at different position @xmath181 .
the 3-d integral @xmath182 in eq.(21 ) has the similar structure as eq.(23 ) . using the upper limit eq.(18 )
, one can estimate the decaying of the integral @xmath183 with @xmath184 .
when @xmath185 , we can expand the function @xmath186 in terms of @xmath187 .
thus , applying eq.(23 ) term by term , we have the decaying behavior of @xmath188 at least as fast as @xmath189 .
similarly , in the case of @xmath190 , one can expand the function @xmath186 by the factor @xmath191 .
following from eq.(23 ) , the decaying behavior of @xmath188 is then decreasing with @xmath184 as @xmath192 .
accordingly , the 1-halo term in the correlation between modes @xmath193 and @xmath194 will generally decay as @xmath195 or @xmath196 .
the correlation function @xmath197 is then approximately diagonal w.r.t .
the spatial index @xmath39 and @xmath102 .
this result is largely valid due to @xmath150 varying with @xmath141 by a power law .
proceeding in the similar way as above , we can also show the diagonality of the correlation function @xmath198 . in this case , instead of eq.(23 ) , we use the following theorem ( tewfik & kim , 1992 ) @xmath199 consider that the cell @xmath200 has the same physical position as cell @xmath201 if @xmath202 , the theorem eq.(24 ) also yields that the 1-halo term of the two modes @xmath47 and @xmath203 will decline with the physical distance between the two modes as @xmath204 or @xmath205 . based on above discussions , one may draw the conclusion that if the halo model is a good approximation to the cosmic density field @xmath24 in the nonlinear regime at all time , their correlation matrix in the dwt representation remains quasi - diagonal forever , and the evolution is quasi - local .
this result is based on the self - similarity of the density profiles of halos and the weakly nonlinear correlation between halos .
the self - similar scaling ensures that the non - local correlation among the dwt variable is uniformly suppressed , independent of the mass of halos .
mathematically , the correlations between the dwt modes of perturbations at different physical places are uniformly converging to zero with the increasing of @xmath64 if the index @xmath154 is mass - independent . to show the quasi - locality of higher order statistical criterion
, we use the hierarchical clustering or linked - pair relation ( peebles 1980 ) , which is found to be consistent with the halo model . for the 3@xmath206 order correlations ,
the linked - pair relation is @xmath207,\end{aligned}\ ] ] where the coefficient @xmath208 might be scale - dependent . subjecting eq.(25 ) to a dwt by 3@xmath206 order basis @xmath209
, we have @xmath210.\end{aligned}\ ] ] where @xmath211 is given by the 3-wavelet integral , @xmath212 since @xmath46 is localized in the cell ( @xmath47 ) , @xmath213 is significant only if the three cells ( @xmath214 ) , ( @xmath215 ) and ( @xmath216 ) coincide with each other at the same physical area .
thus , by virtue of the locality of correlations @xmath217 and @xmath218 ( 3.2 ) , it is easy to see that @xmath219 is small if the cells ( @xmath214 ) , ( @xmath220 ) and ( @xmath221 ) are disjoint in the physical space .
since @xmath208 does not depend on @xmath4 , the result of locality will keep valid when @xmath208 is scale - dependent , obviously , the 3@xmath222 order result can be generalized to @xmath223 order dwt correlation function .
the integral of @xmath224 wavelets @xmath225 , @xmath226, ...
@xmath227 is zero or very small otherwise the @xmath224 cells ( @xmath214 ) , ( @xmath220 ) ... ( @xmath228 ) coincide in the same physical area .
in addition , all terms on the r.h.s .
of the hierarchical clustering relation consist of linked 2@xmath222 dwt correlation function , the @xmath223 order dwt correlation function @xmath229 should be localized .
thus , the criterion eq.(15 ) and other higher order criterions will be satisfied in general . in summary ,
if the cosmic density field is evolved self - similarly from an initially gaussian field , the spatial quasi - locality is true at all time , i.e. * the second and higher order correlation functions of the dwt variables , @xmath53 of the evolved field is quasi - diagonal with respect to the position index @xmath39 . for those type of fields , the possible non - gaussian features with the dwt variables are mainly * non - gaussian one - point distribution of the dwt variables @xmath53 ; * local scale - scale correlation among the dwt variables . above three points
are the major theoretical results of this paper .
to demonstrate the quasi - locality of the evolved cosmic density field , we use samples produced by jing & suto ( 2002 ) .
the samples are given by high resolution n - body simulation , running with the vectorized - parallelized p@xmath230 m code .
the cosmological model was taken to be lcdm model specified by parameters @xmath231 .
the primordial density fluctuation is assumed to obey the gaussian statistics ( this is important for us ) , and the power spectrum is of the harrison - zeldovich type . the linear transfer function for the dark matter power spectrum is taken from bardeen et al ( 1986 ) .
the simulation was performed in a periodic , cubical box of size 100h@xmath232mpc with a 1200@xmath230 grid points for the particle - mesh ( pm ) force computation , and 512@xmath230 particles .
the short - range force is compensated for the pm force calculation at the separation less than @xmath233 , where @xmath234 is the mesh cell size .
the simulations are evolved by 1200 time steps from the initial redshift @xmath235 .
the force resolution is 20 h@xmath232 kpc for the linear density softening form .
it is noted that our statistic tests are performed on scales @xmath236 h@xmath232 mpc .
we have one realization . in the practical computations ,
we divide the 100 h@xmath232 mpc simulation box into 8 subboxes each with size @xmath237 50h@xmath232 mpc .
accordingly , the ensemble average and 1@xmath238 variance are obtained from those 8 subboxes . before showing the quasi - locality ,
we first calculate the correlation function of the sfcs , i.e. @xmath239 . as has been discussed in 2.2 , the correlation function @xmath239 is actually analogue to the ordinary two - point correlation function @xmath240 , where @xmath241 .
since @xmath242 is a filtered density field smoothed by the scaling function on the scale @xmath38 , it is expected that the correlation function @xmath239 will display similar feature as @xmath240 .
figure 1 presents the @xmath243-dependence of @xmath239 at @xmath244 , corresponding to the smoothed field filtered on the linear scale @xmath245 h@xmath232 mpc .
the spatial distance between the cells @xmath39 and @xmath102 is @xmath246 h@xmath232 mpc . in this calculation , we applied wavelet daubechies 4 ( daubechies , 1992 ) .
as expected , the @xmath243- or @xmath184-dependence of the correlation function @xmath239 shows the standard power law , @xmath247 with the index @xmath248 .
that is , this correlation function is not localized .
meanwhile in fig.1 , the contributions of the 1-halo and 2-halo terms are also plotted , respectively .
the 1-halo term is calculated from eq.(21 ) with the nfw density profiles of halos , and the 2-halo term is given by eq.(22 ) in which the correlation function @xmath240 is given by the linear power spectrum .
the nonlinear clustering is largely due to the 1-halo term .
therefore , one can see that the nonlinear evolution of gravitational clustering cause correlations on scales of several mpc .
we now study the quasi - locality of the clustering with the correlation function @xmath121 , which is used in the criterion @xmath249 [ eq.(13 ) ] .
first , we take the same parameter as fig
. 1 , @xmath250 , and @xmath251 h@xmath232 mpc , and also we used wavelet daubechies 4 ( daubechies , 1992 ) , which has @xmath252 .
the result of dwt correlation function @xmath253 .
is shown in fig .
2 . the solid circle at @xmath254 in fig . 2 corresponds to @xmath255 , or @xmath256 , and other solid circles from left to right correspond , successively , to @xmath257 , 2 , 3 ... from fig .
2 , we can see immediately that the shape of the @xmath243- or @xmath258-dependence of @xmath253 is quite different from the `` standard '' power law .
the correlation function @xmath253 is non - zero mainly at point @xmath254 or @xmath259 . at @xmath260 , the correlation function @xmath261 drops to tiny values around @xmath262 . for @xmath263 ,
the correlation function basically is zero .
the correlation length in terms of the position index @xmath39 is approximately zero , namely , the covariance @xmath253 is diagonal .
this is the spatial quasi - locality . in fig .
2 , we also plot the 1- and 2-halo terms , @xmath197 and @xmath264 .
although 1-halo term is dominated by massive halos , the covariance @xmath197 is also perfectly quasi - localized because of the self - similarity of density profiles of massive halos .
the 2-halo term is zero at @xmath254 , because , by definition , eq.(22 ) does not contain the contribution of autocorrelations of halos .
figure 3 presents @xmath265 vs. @xmath243 , for @xmath266 , @xmath267 and @xmath268 . the physical distance
@xmath243 is the same as fig .
2 , given by @xmath269 h@xmath232 mpc . the solid circle at @xmath254 corresponds to @xmath270 , at which , by definition , we have the normalization @xmath271 . other points from small to large values of @xmath243 correspond to @xmath272 1,2 , .... successively .
clearly , all @xmath273 for @xmath274 are less than @xmath275 , which is actually from the noises of sample .
the result implies that for all calculable points of @xmath276 , the correlation is negligible , and satisfies the criterion eq.(13 ) .
we also calculated @xmath265 for modes of @xmath277 , but @xmath278 .
most of these cases shows @xmath279 if @xmath280 .
only exception is for the cases of @xmath281 , and @xmath282 . as an example , fig .
4 presents @xmath283 for modes @xmath284 , and @xmath285 .
it shows that @xmath286 if @xmath287 .
the non - zero value at @xmath288 is about 20% of that at @xmath289 .
this result is consistent with the theorem ( 23 ) , which only requires the suppression for modes @xmath290 , but may not work for @xmath291 .
it implies that the clustering may give rise to the correlations between nearest neighbor cells in phase space .
however , the cell resolved by @xmath281 is a rectangle in the physical space , and the shortest edge is given by @xmath292 , the physics distance of @xmath293 is still less than whole size of the rectangle .
thus , it could be concluded that the covariance @xmath253 is always quasi - diagonal in the sense that all members with @xmath39 and @xmath102 are almost zero if the distance @xmath243 between @xmath39 and @xmath102 is larger than the size of the cell @xmath294 considered . in fig .
4 , we show also a result calculated with wavelets daubechies 6 ( d6 ) , for which @xmath295 .
it yields about the same results as d4 . for the correlation between modes
@xmath44 and @xmath296 with @xmath89 , we use the criterion @xmath297 of eq.(14 ) . in this case
, the physical distance between two cells is @xmath298 and @xmath299 h@xmath232 mpc .
figure 5 plots @xmath249 vs. @xmath243 for modes @xmath300 , @xmath301 , and @xmath302 and 6 .
all the values of @xmath249 in fig .
5 are not larger than @xmath303 , and are much less than the diagonal terms @xmath304 , or @xmath305 . we found this result is generally true for all the cases of @xmath306 .
that is , the second order correlation between two modes with different scales @xmath38 and @xmath88 are always negligible , regardless the indices @xmath39 and @xmath102 . in other words ,
the covariance of the wfc variables @xmath121 generally are quasi - diagonal . as for the high order statistics by criterion eqs.(15 ) and
( 16 ) , we can cite some previous calculations of the non - local scale - scale correlation defined by @xmath307 which is the criterion eq.(15 ) with @xmath308 .
it has been shown that either for the apm bright galaxy catalog ( loveday et al .
1992 ) or mock samples of galaxy survey ( cole et al .
1998 ) , the non - local scale - scale correlation always yields @xmath309 if @xmath310 ( feng , deng & fang 2000 ) .
although this work was not for addressing the problem of the quasi - locality , the result did support the quasi - locality up to the 4@xmath311 order statistics .
as discussed in 2.3 , it is necessary to show that the random variables @xmath53 are non - gaussian for evolved fields .
the possible non - gaussian features with the dwt variables are ( 1 ) non - gaussian one - point distribution of @xmath53 , and ( 2 ) local scale - scale correlations ( 3.3 ) . in fig .
6 , we plot the one - point distribution on scales @xmath312 and @xmath313 5 , 6 , 7 and 8 .
it illustrates that the kurtosis of the one - point distribution is high .
the pdf ( probability distribution function ) is approximately lognormal .
the 4@xmath311 order local scale - scale correlation @xmath314 is plotted in fig .
7 . it shows @xmath315 on small scales ( @xmath313 5 , 6 and 7 ) , while random data gives @xmath316 on all scales .
the evolved field is highly non - gaussian , although it is always quasi - localized .
we showed that the cosmic clustering behavior is quasi - localized .
if the field is viewed by the dwt modes in phase space , the nonlinear evolution will give rise to the coupling between modes on different scales but in the same physical area , and the coupling between modes at different position is weak .
the quasi - local evolution means that , if the initial perturbations in a waveband @xmath57 and at different space range @xmath60 is uncorrelated , the evolved perturbations in this waveband at different space range @xmath60 will also be uncorrelated , or very weakly correlated . in this sense
, the nonlinear evolution has memory of its initial spatial correlation in the phase space .
this memory is essentially from the hierarchical and self - similar feature of the mass field evolution .
the density profiles of massive halos obey the scaling law [ eq.(18 ) ] , and therefore , the contributions to the non - local correlation function from various halos are uniformly suppressed .
it has been realized about ten years ago that some random fields generated by a self - similar hierarchical process generally shows locality of their auto - correlation function in the phase space , if the initial field is local , like a gaussian field ( ramanathan & zeitouni , 1991 ; tewfik & kim 1992 ; flandrin , 1992 ) .
later , this result are found to be correct for various models of structure formations via hierarchical cascade stochastic processes ( greiner et al .
1996 , greiner , eggers & lipa , 1998 ) .
these studies implies that the local evolution and initial perturbation memory seems to be generic of self - similar hierarchical fields , regardless the details of the hierarchical process .
it has been pointed out that models for realizing the self - similar hierarchical evolution of cosmic mass field , such as the fractal hierarchy clustering model ( soneira & peebles 1977 ) , the block model ( cole & kaiser 1988 ) , merging cell ( rodrigues & thomas 1996 ) , have the same mathematical structures as hierarchical cascade stochastic models applied in other fields ( pando et al 1998 , feng , pando & fang 2001 ) .
obviously , the local evolution can be straightforward obtained in those models .
the dwt analysis is effective to reveal the quasi - locality in phase space .
such quasi - locality is hardly described by the fourier modes @xmath86 , as the information of spatial positions is stored in the phases of all fourier modes . moreover , the fourier amplitudes @xmath317 subject to the central limit theorem , and are insensitive to non - gaussianity .
the wavelet basis , however does not subject to the central limit theorem ( pando & fang 1998 ) , which enable us to measure all the quasi - local features with the statistics of @xmath53 .
the quasi - locality of the dwt correlation is essential for recovery of the primordial power spectrum using a localized mapping in phase space .
such mapping has been developed in recovering the initial gaussian power spectrum from evolved field in the quasi - linear regime ( feng & fang 2000 ) . by virtue of the quasi - locality in fully developed fields
, we would be able to generalize the method of localized mapping in phase space to highly non - linear regime .
the quasi - local evolution may also provide the dynamical base for the lognormal model ( bi , 1993 ; bi & davidsen 1997 , jones 1999 ) .
the basic assumption of the lognormal model is that the non - linear field can be approximately found from the corresponding linear gaussian field by a local exponential mapping .
the local mapping is supported by the quasi - local evolution .
we see from fig . 6
that the pdf of evolved field is about lognormal .
therefore , in the context of quasi - local evolution , a local ( exponential ) mapping from the linear gaussian field to a lognormal field might be a reasonable sketch of the nonlinear evolution of the cosmic density field .
we thank dr .
y.p . jing for kindly providing his n - body simulation data .
llf acknowledges supports from the national science foundation of china ( nsfc ) and national key basic research science foundation .
|
we show that the nonlinear evolution of the cosmic gravitational clustering is approximately spatial local in the @xmath0-@xmath1 ( position - scale ) phase space if the initial perturbations are gaussian .
that is , if viewing the mass field with modes in the phase space , the nonlinear evolution will cause strong coupling among modes with different scale @xmath1 , but at the same spatial area @xmath0 , while the modes at different area @xmath0 remain uncorrelated , or very weakly correlated .
we first study the quasi - local clustering behavior with the halo model , and demonstrate that the quasi - local evolution in the phase space is essentially due to the self - similar and hierarchical features of the cosmic gravitational clustering .
the scaling of mass density profile of halos insures that the coupling between @xmath2 modes at different physical positions is substantially suppressed . using high resolution n - body simulation samples in the lcdm model
, we justify the quasi - locality with the correlation function between the dwt ( discrete wavelet transform ) variables of the cosmic mass field .
although the mass field underwent a highly non - linear evolution , and the dwt variables display significantly non - gaussian features , there are almost no correlations among the dwt variables at different spatial positions
. possible applications of the quasi - locality have been discussed .
|
in this supplemental material , we perform additional analysis that support the results obtained in the main article . in this section ,
we explicitly derive the link operators in terms of schwinger bosons . following @xcite
, we define the operators @xmath25 , @xmath26 as _ right _ and _ left _ generators , acting on the finite - dimensional hilbert space of the link , @xmath90 where @xmath91 are bosons that implement the schwinger representation of the @xmath0 algebra , and the pauli matrices @xmath92 the bosonic operators act on two different sites @xmath93 on the link between the two adjacent sites @xmath2 and @xmath10 and obey the @xmath0 commutation rules @xmath94=i \sum_{c } \epsilon^{abc}r^{c } , ~ \ ,
~[r^{a},l^{b}]=0 , ~ \ , ~ [ l^{a},l^{b}]=&i \sum_{c } \epsilon^{abc}l^{c } , \\ \end{split}\ ] ] taking into account that they commute on different links ( i.e. , different @xmath2 , @xmath95 ) . with the schwinger representation , there are two multiplets with well - defined @xmath0 commutation relations @xmath96 = \sum_{\gamma } \frac{\sigma^{a}_{\alpha \gamma}}{2 } c_{\gamma l } , ~ \ , ~ & \left [ \left ( \sum_{\beta } \sigma^{y}_{\alpha \beta}c^{\dagger}_{\beta
l } \right ) , l^{a}\right ] = \sum_{\gamma } \frac{\sigma^{a}_{\alpha \gamma}}{2 } \left(\sum_{\beta } \sigma^{y}_{\gamma \beta } c^{\dagger}_{\beta l }
\right),\\ \left[c^{\dagger}_{\alpha r } , r^{a}\right ] = - \sum_{\gamma } c^{\dagger}_{\gamma r } \frac{\sigma^{a}_{\gamma \alpha}}{2 } , ~ \ , ~ & \left[\left ( \sum_{\beta } c_{\beta r } \sigma^{y}_{\beta \alpha } \right ) , r^{a}\right ] = -\sum_{\gamma } \left(\sum_{\beta } c_{\beta l } \sigma^{y } _ { \beta \gamma }
\right ) \frac{\sigma^{a } _ { \gamma \alpha}}{2}. \end{split}\ ] ] given these two multiplets , we can define the link operator as @xmath97 we stress that these quantum link operators , or gauge fields , are associated with links between lattice points , transporting color information between the two points they connect .
the two indices of the link operator @xmath98 are identified with the two ends of the link @xmath99 .
the left ( right ) index is associated with the beginning ( end ) of the link as shown in fig .
[ gaugecartoon ] .
the gauge transformation acts on @xmath19 according to @xmath100 since in general @xmath101 and @xmath102 are different , our gauge - invariant equations will require invariance under right multiplication and left multiplication separately .
an arbitrary gauge transformation can be built from individual gauge transformations at the point of the lattice .
therefore , it is sufficient to consider just a gauge transformation at the lattice site @xmath103 .
every link emanating from the point is affected by the gauge transformation generated by @xmath104 and the total color @xmath105 carried by a link is just the difference : @xmath106 .
the sum over non - abelian electric fields emanating from a single point of the lattice is the lattice analog of the electromagnetic divergence of the electric field , @xmath107 .
however , because the electric field itself varies from @xmath108 to @xmath109 along the link , there is an additional contribution given by the total color carried by the link .
thus , the generator is recast into : @xmath110 .
the gauge invariance of the physical sector is defined by @xmath111 , i.e. , at every vertex the state transforms locally as a singlet of @xmath0 .
if we identify the local color density @xmath112 , then @xmath113 is the usual gauss law .
these are properties of non - abelian lattice gauge theories , independent of their representation via quantum link models .
quantum link .
below , the two different gauge sectors in the hilbert space of two quantum links , i.e. the singlet and the triplet sector . [ gaugecartoon ] ] the local degrees of freedom in a @xmath0 link are encoded in the quantum state of four different bosonic modes @xmath37 .
different representations of this local hilbert space are given by the occupation of these modes .
the hilbert space with one excitation in total is four - dimensional , and it can be decomposed in a spin degree of freedom with the @xmath0 symmetry group and a position degree of freedom . a @xmath0 gauge invariant interaction commutes with the local generator of the @xmath0 at any vertex . as a consequence ,
the total angular momentum at any vertex of the lattice is a conserved quantity of the dynamics . in the simplest example of two quantum links , i.e. just one vertex ,
there are only two possible gauge sectors , the singlet sector where the total angular momentum around the vertex is zero , and the triplet sector where the total angular momentum is equal to one .
in the main text , we have considered the local four - dimensional hilbert space spanned by the occupation of four different modes @xmath37 . by choosing the subspace with one excitation per link
, we can faithfully represent its quantum state with two qubits , the position and spin qubits , @xmath114 .
the excitations of these qubits are related to the left and right modes through @xmath115 and @xmath116 . with these definitions ,
the left and right operators defined in the main text are recast into @xmath117 \end{split}\ ] ] with @xmath47 , @xmath48 .
the hamiltonian for one plaquette can then be written down as @xmath118.\ ] ] in order to identify all the terms of the trace one can use the condition @xmath119 \neq 0 $ ] .
the condition is met if and only if @xmath120 the first condition gives @xmath69 term , the second @xmath121 terms , and @xmath122 terms for the last three equations , for a total of @xmath123 terms @xmath124=\frac{1}{4 } \left\ { \gamma^0_{12}\gamma^0_{23}\gamma^0_{31}+ \sum_{abc}\epsilon_{abc}\gamma^a_{12}\gamma^b_{23}\gamma^c_{31 } -\sum_{a}\big[\gamma^0_{12}\gamma^a_{23}\gamma^a_{31 } + \gamma^a_{12}\gamma^0_{23}\gamma^a_{31}+\gamma^a_{12}\gamma^a_{23}\gamma^0_{31}\big ] \right\ } , \end{split}\ ] ] where we have used the subscripts @xmath125 to represent the links @xmath126 , respectively .
the pure - gauge hamiltonian can then be written in terms of @xmath127 matrices as @xmath128 \right\}. \end{split}\ ] ] then , one can explicitly write this hamiltonian in terms of up to six - body interactions between qubits , for instance @xmath129 in this section we provide a brief overview of the experimental setups considered for the implementation of the proposed simulator . the first setup considered in the main text
is comprised of six transmon qubits provided with tunable coupling to a single microwave resonator .
a transmon qubit is a special instance of a cooper pair box @xcite , operating at specific regimes .
a cooper pair box is made of a piece of a superconducting island connected to a voltage source through a capacitor , and grounded via a josephson junction .
the temperature of these superconducting systems is low enough so that the quantum state of the setup can be defined by the number of cooper pairs on the superconducting island . in order to add tunability and control to this quantum system ,
the single josephson junction of the cooper pair box is substituted with a squid loop , threaed with magnetic fluxes .
these superconducting systems can be coupled to superconducting microwave resonators , which can be used to induce , via electromagnetic interactions , tunnelling of cooper pairs in and out the superconducting island , as well as reading out the state of the system @xcite .
the hamiltonian of the setup comprising a cooper pair box coupled to a microwave resonator reads @xmath130 here we have defined the quantized charge @xmath131 on the superconducting island , which is proportional to the number of cooper pairs , the offset charge @xmath132 and the quantized flux @xmath54 that threads the squid loop .
the @xmath133(@xmath134 ) are bosonic operators that act upon the resonator field , of frequency @xmath135 .
@xmath136 stands for the charging energy of the superconducting island , while @xmath137 is the josephson energy of the dc - squid loop .
the coupling coefficient @xmath9 is a function of the circuit capacitances , @xmath138 is the root mean square voltage of the resonator , and @xmath139 is the electron charge .
one of the main problems that arise in using cooper pair boxes as controlled quantum systems is the high sensitivity of their quantum state to noise on the offset charge @xmath132 . by using large capacitive pads shunting the superconducting island
, one can achieve insensitivity with respect to this noise .
these cooper pair boxes provided with large capacitive pads are known as transmon qubits @xcite . in this parameter regime ,
the hamiltonian of eq .
( [ cpb ] ) models a transmon qubit .
it can be rewritten in a diagonal basis for the part which is not coupled to the resonator , @xmath140 where we have included the first three levels @xmath141 of this device . higher levels are usually excluded from the quantum dynamics , as being more energetic and hard to access .
the first three levels can be approximately modeled as the first three levels of an anharmonic oscillator , with a relative anharmonicity factor of @xmath142 .
if @xmath143 is big enough , one can consider the transmon qubit device a two - level system @xcite .
the setups envisioned in this work use two variants of the transmon design . each of the six devices sketched in fig .
1a in the main text is made of three superconducting islands , connected via two squid loops , marked with different colors in the figure .
the magnetic fluxes acting on these loops modify the quantum level structure of the system , and an appropriate manipulation of these fluxes in time allows for a fast modulation of the effective coupling @xmath144 to the resonator ( see eq .
[ hamcircuitqed ] ) . this modulation does not affect the bare energy levels @xmath145 of the device @xcite and gives the possibility of performing red and blue sidebands upon the coupled qubit - resonator system , which , in turns , allows to generate many - body interactions @xcite .
the setup considered in fig .
1b in the main text describes transmon qubits with improved design , the so called xmon qubits @xcite . in this case , the shape of the capacitance pads is designed in order to improve coherence times of the device , by reducing coupling to material defects and radiative losses .
the interactions among these qubits are mediated by direct capacitive couplings , not making use of a common superconducting resonator .
for this reason , the many - body interactions are reduced in this case to nearest - neighbor two - qubit gates . [
[ many - body - interactions - and - gauge - invariance - for - different - sectors ] ] many - body interactions and gauge invariance for different sectors ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ in order to simulate the hamiltonian associated with the non - abelian model in eq .
( [ explicitham ] ) , one can simulate stepwise the single many - body monomials that compose the total interaction .
each of them has the form of an @xmath52-qubit string term @xmath146 where the set of superindices can assume values @xmath147 .
up to a maximum of @xmath70 local rotations , this general interaction can be mapped back to , e.g. , @xmath148 . in what follows
, we will show how to obtain this effective interaction in two different scenarios . in one case
, one can use a set of @xmath52 tunable - coupling transmon qubits coupled to a single microwave resonator . in the second scenario , a set of @xmath52 charge - like qubits , e.g. , transmon or xmon qubits , are coupled to an additional ancillary qubit , for a total of @xmath149 qubits .
we consider the sequence of collective and single qubit gates @xcite , as @xmath150 and obtain the set of interactions @xmath151 the collective interaction @xmath152 in eq .
( [ ums ] ) can be realized within a set of @xmath52 three - island transmon qubits , by threading the two squid loops embedded in each device with fast - tunable magnetic fluxes , tuned at appropriate frequencies @xcite .
the general @xmath52-body term of eq .
( [ genericnbody ] ) can be simulated with this approach with a total of @xmath73 collective gates , and a number of single qubit gates which is upper bounded by @xmath74 .
the hamiltonian in eq .
( [ explicitham ] ) is composed of @xmath123 monomials , @xmath69 three - body term , @xmath121 six - body terms , and @xmath122 five - body terms .
therefore , the simulation for one digital step will amount to implement @xmath76 collective gates and a number of single - qubit gates which is bounded by @xmath77 .
converges to the expected invariant , @xmath153 for a ) and @xmath154 for b ) , as the number of digital steps increases .
the initial states @xmath79 chosen in a ) and b ) are represented in the main text , infig .
2 a ) and c ) , respectively .
the overlap @xmath80 witnesses the periodic oscillations of the initial state on the triangular plaquette for the first case , while it remains invariant in the second one.[sintrip ] ] one can implement the interaction @xmath155 in a set of capacitively - coupled charge - like qubits , by decomposing it into a set of @xmath83-phase and single - qubit gates . first , let us consider the sequence of gates in eq .
( [ ums ] ) between @xmath52 qubits and an additional ancillary one , labeled by @xmath156 .
we consider also that the gates are rotated about the @xmath157 axis of an angle @xmath158 . the sequence of gates in eq .
( [ ums ] ) can then be rewritten in the following way @xmath159 this sequence of gates is equivalent to the following interactions @xmath160 if the ancillary qubit is initialized in an eigenstate of @xmath161(@xmath162 ) , @xmath163 ( @xmath164 ) , then the effective dynamics on the @xmath52-qubit system would be equivalent to the interaction @xmath165 .
the interaction @xmath166 can be realized as a sequence of @xmath52 @xmath167 gates between the ancillary qubit and all the other ones , @xmath168 .
each of these two - body gates can be realized as a sequence of two local @xmath169-rotations and a @xmath83-phase gate between the qubits @xmath69 and @xmath67 , @xmath170 .
in fact , defining @xmath171 one has that @xmath172 in this way , an arbitrary @xmath52-body term can be encoded , up to a number of local rotations bounded by @xmath70 , with a sequence of @xmath70 @xmath83-phase gates and a number of single - qubit gates bounded by @xmath173 .
thus the quantum simulation of one digital step of the hamiltonian in eq .
( [ explicitham ] ) will amount to realize @xmath174 @xmath83-phase gates and a number of single - qubit rotations which is bounded by @xmath84 .
the @xmath83-phase gate between two qubits can be performed by adiabatically changing the frequency of one qubit , bringing the level @xmath175 close to the anharmonic excitated level @xmath176 , to induce state - dependent phase accumulations .
unwanted errors due to leakage to the @xmath176 state are exponentially suppressed in the gate duration . for finite gate times errors
can be minimized upon optimization in the fourier spectrum of the control signal @xcite .
this optimization has been carried out for @xmath177 gates in @xcite . for a generic quantum simulation involving @xmath83-phase gates
, one should define the set of angles to be used in the experiment , and optimize over the duration and fourier spectrum of the gate , to avoid unwanted leakage .
this has been done in @xcite by increasing the gate time from @xmath178 ns to @xmath179 ns , not observing significant state leakages up to phases of @xmath180 rads .
additional errors on the fidelity of the gate come from standard decoherence of the qubits , and control errors , see ref .
@xcite for a discussion of the error budget . finally , to complete the information given in the main text
, we show in this section , in fig .
[ sintrip ] , additional plots that complement fig . 3 in the main text .
the simulated gauge invariance is shown to converge for the singlet and triplet configurations on a single triangular plaquette .
the proposed experiment can be extended to larger simulated lattices , which are implementable by considering setups of superconducting qubits with enhanced connectivity .
the required connectivity can be achieved by using several microwave resonators , acting as non - local quantum buses , or by considering superconducting pads that can provide with multiple capacitive couplings , as schematized in fig .
[ largelattice ] .
for instance , we can envision , as shown in the main text , a ladder architecture that can support a singlet gauge invariant subspace , where non - abelian string dynamics could be studied .
it has been shown @xcite that the scaling of the number of steps @xmath181 needed to achieve a digital error @xmath182 in a suzuki - lie - trotter decomposition is almost linear in the simulated phase of the dynamics and sub - polinomial in @xmath182 , according to the formula @xmath183 in the formula @xmath184 is the number of terms that compose a generic hamiltonian @xmath185 , and @xmath186 is the fractal degree of approximation @xcite , which we set to @xmath187 .
for one plaquette one has @xmath188 , @xmath69 three - body term , @xmath121 six - body terms , and @xmath122 five - body terms , all present in the system with the energy @xmath189 .
considering that the norm of an arbitrary product of @xmath190 pauli matrices is @xmath191 , @xmath192 , one has , using a triangular inequality , that the norm of the @xmath0 hamiltonian for one plaquette is bounded by @xmath193 . as a consequence , the number of steps necessary to implement a dynamics with error @xmath182 for the time @xmath194 on a lattice of @xmath52 plaquettes is bounded by @xmath195 notice that this formula gives a very loose bound which is not dependent on the underlying structure of the hamiltonian .
nevertheless , it shows that the scaling of the number of steps is efficiently solvable by a quantum simulator , not being exponential in the resources used .
the exact scaling of the number of steps can not be known ( up to an exact solution of the problem , which is not addressable numerically for larger lattices ) , however it is useful to extract the dependence @xmath196 from the upper - bound formula in eq .
( [ errbound ] ) . assuming independent experimental errors on the single trotter steps
, one can relate the total number of steps @xmath181 to the required experimental fidelity for the implementation of a single hamiltonian term .
we impose an accumulated experimental error @xmath197 of the same order of magnitude of the trotter error @xmath198 .
this enforces the error on the single step to be of the order @xmath199 .
besides these issues on error scaling and accumulation , the accesible simulation time has to be smaller than @xmath200 and @xmath201 of the single qubits , defining a system of necessary conditions for the simulation of a lattice of @xmath52 plaquettes @xmath202 in the main text we have focused on pure gauge dynamics with plaquette interactions , where usual perturbative approaches do not get energy scales which are big enough . in this section ,
we point out the basic elements needed for the simulation of the simplest @xmath0 gauge model coupled to a matter field in @xmath203-dimensions .
the basic ingredients and the validity of the different energy - scale regimes have been already discussed in the simulation of abelian gauge models in trapped ions and superconducting circuits @xcite .
this section can be seen as a generalization to non - abelian setups .
the ideal interaction hamiltonian that we would like to simulate is given by @xmath204 + \sum_{\mu \nu } \left [ \sigma^{y}_{\alpha \mu } c^{\dagger}_{\mu l}(i ) c_{\nu r } ( i+1 ) \sigma^{y}_{\nu \beta } \right ] \right\ } b_{\beta}(i+1 ) + \text{h.c .
} , \end{split}\ ] ] where @xmath205 and @xmath206 are hard - core bosonic modes , with @xmath23 , @xmath207 and @xmath208 .
one way to obtain the previous hamiltonian is to start with operators that respect one of the symmetries at the fundamental level and introduce an energy penalty for the combinations of operators that do not fulfill the second symmetry . in general
, it is simpler to enforce the abelian @xmath210 symmetry , since it only requires density - density interaction , and start with interactions that fulfill the non - abelian symmetry from the beginning .
the basic building blocks are given by a strong interaction that enforces the abelian symmetry , i.e. , @xmath212 + \omega \sum_{i } \left [ g(i , i+1 ) - n_{0 } \right]^{2}.\ ] ] here , the @xmath213 frequency of the bosonic modes splits the hilbert space into subspaces with an equal number of excitations . performing a rotating - wave approximation , interactions that conserve the number of particles
are energetically allowed , while the term with energy scale @xmath214 enforces @xmath215 excitations per link .
hence , in second - order perturbation theory , up to a renormalization of the bare frequencies @xmath213 of the modes , the effective hamiltonian is given by @xmath217 + \sum_{\mu \nu } \left [ \sigma^{y}_{\alpha \mu } c^{\dagger}_{\mu l}(i ) c_{\nu r } ( i+1 ) \sigma^{y}_{\nu \beta } \right ] \right\ } b_{\beta}(i+1 ) + \text{h.c . } \\
& -2 \frac{j^{2}_0}{\omega } \sum_{i } \left\ { \sum_{\alpha } b^{\dagger}_{\alpha}(i ) b_{\alpha}(i ) \sum_{\beta } \left [ c^{\dagger}_{\beta r}(i ) c_{\beta r}(i ) + c^{\dagger}_{\beta l}(i ) c_{\beta l}(i ) \right ] \right\ } , \end{split}\ ] ] which reduces to the ideal hamiltonian @xmath218 , up to a density - density term that does not break any of the symmetries and it could be reduced or even eliminated , allowing a counter - term of the same type in the initial hamiltonian @xmath219 .
r. barends , j. kelly , a. megrant , d. sank , e. jeffrey , y. chen , y. yin , b. chiaro , j. mutus , c. neill , p. omalley , p. roushan , j. wenner , t. c. white , a. n. cleland , and j. m. martinis , phys .
lett . * 111 * , 080502 ( 2013 ) .
r. barends , j. kelly , a. megrant , a. veitia , d. sank , e. jeffrey , t. c. white , j. mutus , a. g. fowler , b. campbell , y. chen , z. chen , b. chiaro , a. dunsworth , c. neill , p. omalley , p. roushan , a. vainsencher , j. wenner , a. n. korotkov , a. n. cleland , and j. m. martinis , nature * 508 * , 500 ( 2014 ) .
r. barends , l. lamata , j. kelly , l. garca - lvarez , a. g. fowler , a. megrant , e. jeffrey , t. c. white , d. sank , j. y. mutus , b. campbell , yu chen , z. chen , b. chiaro , a. dunsworth , i .-
c . hoi , c. neill , p. j. j. omalley , c. quintana , p. roushan , a. vainsencher , j. wenner , e. solano , and j. m. martinis , nat .
* 6 * , 7654 ( 2015 ) .
|
we propose a digital quantum simulator of non - abelian pure - gauge models with a superconducting circuit setup . within the framework of quantum link models ,
we build a minimal instance of a pure @xmath0 gauge theory , using triangular plaquettes involving geometric frustration .
this realization is the least demanding , in terms of quantum simulation resources , of a non - abelian gauge dynamics .
we present two superconducting architectures that can host the quantum simulation , estimating the requirements needed to run possible experiments .
the proposal establishes a path to the experimental simulation of non - abelian physics with solid - state quantum platforms .
gauge invariance is a central concept in modern physics , being at the core of the standard model of elementary particle physics . in particular , invariances with respect to @xmath0 and @xmath1
gauge symmetries characterize the weak interaction and quantum chromodynamics @xcite . in this sense ,
gauge theories represent a cornerstone in our understanding of the physical world and lie at the heart of diverse phenomena , such as the quark - gluon plasma or quantum spin liquids . in condensed matter physics
, @xmath0 gauge fields can also emerge dynamically in relation to exotic many - body phenomena , like quantum hall systems , frustrated magnets , or superconductors @xcite .
lattice gauge theories ( lgt ) are non - perturbative discrete formulations that contribute to the analysis of key features of these models , such as color confinement or chiral symmetry breaking . starting from the seminal work by wilson in 1974 @xcite , lgt have attracted a significant attention across several branches of theoretical physics . in the last decades
@xcite , quantum montecarlo simulations have achieved unprecedented accuracies in determining the whole hadronic spectrum of the standard model . however , understanding its full phase diagram from first principles , or simulating dynamical processes , remain out of reach of current numerical computations . quantum simulators @xcite provide a new approach to solve complex long - standing problems in quantum physics . in a quantum simulator ,
a proper encoding of lgt can allow for the retrieval of information about ground state and dynamics , in a wide range of regimes .
previous works have considered abelian @xcite and non - abelian lgt in optical lattices @xcite ( see also @xcite and references therein ) , both as analog and digital simulations @xcite . in these implementations ,
matter - gauge interactions are modeled as a second - order process from hubbard - like interactions , while the simulation of pure - gauge dynamics remains more demanding . in the last years
, superconducting circuits have proven to be reliable devices that can host quantum information and simulation processes @xcite . the possibility to perform quantum gates with high fidelities , together with high coherence times ,
makes them ideal devices for the realization of digital quantum simulations @xcite , previously considered in ion - trap systems @xcite . triangular plaquette model schematized in c ) , where the left and right gauge degrees of freedom are explicitly depicted.[scheme ] ] in this letter
, we propose a digital quantum simulation of a non - abelian dynamical @xmath0 gauge theory in a superconducting device .
we start by building a minimal setup , based on a triangular lattice , that can encode pure - gauge dynamics .
the degrees of freedom of a single triangular plaquette of this lattice are encoded into qubits .
we propose two implementations of this quantum simulator , using two different superconducting circuit architectures , as depicted in fig .
[ scheme ] .
we consider a setup in which six tunable - coupling transmon qubits are coupled to a single microwave resonator , and a device where six capacitively - coupled xmon qubits stand on a triangular geometry , coupled to a central auxiliary one .
we compute experimental requirements necessary to perform the simulation on one plaquette , and provide arguments for scaling to large lattices .
_ lattice gauge theories.- _ lgt are discretized versions of a gauge theory . in a conventional approach of lattice gauge theories , space - time
is discretized while ensuring a covariant formulation of the theory . in quantum simulations , there is no direct access to the time direction , that is fixed and continuous .
hence , the equivalent hamiltonian formulation of lattice models is used . in this case , the space is discretized and the action of a local gauge invariant hamiltonian characterizes a continuous dynamical evolution .
this fictitious asymmetry of the time direction forces us to define a physical hilbert space or gauss law of the model .
we make use of a set of discretized space points @xmath2 , and @xmath3 operators @xmath4 associated to each link , connecting two adjacent sites @xmath5 .
a single @xmath4 operator has two implicit color indices @xmath6 , @xmath7 , that connect the @xmath8-th component of the fermionic field at position @xmath2 with the @xmath9-th component of the field at position @xmath10 . in order to construct a non - abelian @xmath3 theory built out of these @xmath4 operators
, one has to build a hamiltonian that is invariant under a generic gauge transformation @xmath11 where @xmath12 are the @xmath13 generators of the @xmath3 algebra in the corresponding representation , and @xmath14 the relative phase angles .
it is straightforward to verify that the following hamiltonian on a square lattice is gauge invariant @xmath15 + \text{h.c .
} , \label{wilsonloop}\ ] ] where the trace is performed over the color indices , @xmath16 has energy units , and the directions @xmath17 and @xmath18 span the two - dimensional lattice .
this hamiltonian is a pure gauge operator , as it does not involve any fermionic operator and it is the minimum instance of a _ wilson loop _ around a plaquette @xcite .
we focus now on the construction of a non - abelian @xmath0 quantum link model that mimics this gauge - invariant behavior in a finite - dimensional hilbert space , suitable for quantum simulations .
notice that the framework of the quantum link models is valid for any compact lie group @xcite .
we use a four - dimensional hilbert space or , equivalently , two qubits associated with each link .
one can define link operators @xmath19 acting on this four - dimensional hilbert space with the associated gauge generators @xmath20 that satisfy the @xmath0 algebra @xmath21=i\delta_{{\vec x}{\vec y } } \sum_{c } \epsilon_{abc}g^c(\vec{x}).\ ] ] to define the set of operators , we use a quantum link formulation @xcite , where the hilbert space of a link is given by a set of bosonic modes @xmath22 that implement the schwinger representation of the @xmath0 algebra , with @xmath23 , acting on two different sites @xmath24 on the link between the two adjacent vertices @xmath2 and @xmath10 . in this space , we build two sets of _ right _ and _ left _
generators @xmath25 , @xmath26 , on this finite - dimensional hilbert space , using the bosonic modes @xmath22 @xcite .
we define the gauge generators as @xmath27 , where the sum over @xmath28 is taken among all the links of the lattice converging to a single vertex @xmath2 . upon them .
d ) extended lattice with periodic boundary conditions .
this lattice allows for the @xmath29 sector.[states ] ] the representations of @xmath0 are quasi - real , therefore , ordinary and their dual representation ( i.e. , particle and antiparticle ) are equivalent .
therefore , there are two multiplets with well - defined @xmath0 transformations @xcite .
one can finally define the link operators , acting on the finite - dimensional hilbert space of one link , in terms of schwinger bosons on a given link @xcite , @xmath30 these operators @xmath31 have the following commutation rules with the left and right operators @xmath32&= - u({\vec y},{\hat \mu})\frac{\sigma^{a}}{2 } \delta_{{\vec x}{\vec y}}\delta_{{\hat \nu}{\hat \mu}},\\ [ u({\vec y},{\hat \nu } ) , l^{a}({\vec
x},{\hat \mu})]&=\frac{\sigma^{a}}{2}u({\vec y},{\hat \mu})\delta_{{\vec x}{\vec y}}\delta_{{\hat \nu}{\hat \mu } } , \end{split}\ ] ] where we have defined the usual pauli matrices @xmath33 , see also @xcite , while @xmath34 is defined as the identity operator .
the commutation rules for a general gauge transformation follow in a straightforward way , _
ye^-i^a(y)g^a(y)&u(x , ) _
ze^i^a(z)g^a(z ) + & = e^i^a(x ) u(x , ) e^-i^a(x+ ) , [ commug ] where the products over @xmath35 and @xmath36 are extended over the whole lattice .
this equation shows that quantum link models are formulations of gauge - invariant models .
in fact , one can notice how the action of a gauge transformation in eq .
( [ commug ] ) mimics eq .
( [ gauge ] ) .
a generic state in the local hilbert space @xmath37 of a quantum link is given by the occupation of the operators @xmath38 .
since the total occupation per link , @xmath39 , is a constant of motion [ see eq .
( [ udef ] ) ] , we restrict to the subspace with one occupied mode per link .
notice that matter - gauge interactions in @xmath40 dimensions can be derived @xcite in a second - order perturbation theory , considering extensions of previous works @xcite . in the following
, we will focus instead on pure - gauge two - dimensional interactions . _
triangular lattice.- _ the continuum limit of quantum link models and their thermodynamical properties are topics under active research . in related condensed matter models
, the importance of the lattice geometry has been shown in fundamental aspect of the phase diagram , as the existence of confinement and de - confinement phases @xcite .
quantum simulations on large lattices may provide insights on the continuum limit of lgt , and its dependence on the geometry of the underlying lattice . in this spirit
, we consider a minimal implementation of a pure @xmath0 invariant model in a triangular lattice , by using triangular plaquettes , as depicted in fig .
[ scheme]c . in this case
, the pure - gauge hamiltonian on a single plaquette reads @xmath41.\ ] ] this interaction corresponds to the magnetic term of a gauge invariant dynamics , which acts on closed loops .
we focus on the magnetic term , since the representation of the electric counterpart is trivial . since this hamiltonian commutes with gauge generators , @xmath42=0 , \ ,
~ \forall \vec{x},a$ ] , these are constant of motion , and one can define different gauge sectors with the initial values of the generators .
due to the triangular geometry and having one occupied mode per link , there are no global states that have zero eigenvalue for the gauge generators @xmath43 at every vertex in a single plaquette . in general , the absence of the zero - eigenvalue sector depends on the topology of the lattice , and is avoided , for example , in the periodic lattice of fig .
[ states]d .
the different gauge sectors can be characterized by the eigenvalues of the operator @xmath44 ^ 2 $ ] , as in fig .
[ states]a , [ states]b , and [ states]c , see the supplemental material for additional details @xcite .
the local hilbert space of a link is four dimensional , and it can be faithfully spanned by two qubits , called `` position '' @xmath45 and `` spin '' qubit @xmath46 . in this subspace , it is useful to define the operators @xmath47 , @xmath48 , such that the total hamiltonian is written as @xcite @xmath49 \big\ } \nonumber.\end{aligned}\ ] ] ) , versus the ideal value ( @xmath50 ) , @xmath51 , as a function of the digital steps @xmath52 and the simulated phase @xmath53 .
the error decreases with large @xmath52 , and small @xmath54 .
the contour plot is interpolated from numerical data at @xmath55 .
the initial state is chosen as depicted in fig .
[ states]a.[contourgauge ] ] _ superconducting circuit implementation.- _ in order to simulate the interaction of eq .
( [ hg ] ) , one can decompose its dynamics in terms of many - body monomials , and implement them sequentially with a digitized approximation @xcite . in a digital approach ,
one decomposes the dynamics of a hamiltonian @xmath56 by implementing its components stepwise , @xmath57 ( here and in the following @xmath58 ) , for a total of @xmath59 gates , with an approximation error that goes to zero as the number of repetitions @xmath52 grows . in a practical experiment
, each quantum gate @xmath60 will be affected by a given error @xmath61 . by piling up sequences of such gates , for small gate errors @xmath62 ,
the total protocol will be affected by a global error , which is approximately the sum @xmath63 .
this model for error accumulation has been proved in recent experiments @xcite .
the effects of digitization in the simulation of lgt can be observed in fig .
[ contourgauge ] .
we have numerically integrated a schrdinger equation regulated by the hamiltonian in eq .
( [ hg ] ) , choosing as initial state one of the @xmath64 states shown in fig .
[ states]a , for different simulated phases @xmath53 @xcite .
we compute both the evolution with the ideal hamiltonian and with a digital sequence , in which we act with a single monomial at a time , repeating the protocol @xmath52 times . as a figure of merit
, we compute the relative deviation @xmath51 from the ideal value of the gauge invariant , where @xmath65 stand for average values computed on the ideal ( digitally simulated ) state .
the deviation goes to zero for large @xmath52 and small phases , defining surfaces at given error tolerance in the @xmath52-@xmath54 space . to simulate the pure - gauge interaction in a single triangular plaquette ,
we first consider a setup in which six tunable - coupling transmon qubits are coupled to a single microwave resonator @xcite .
each tunable - coupling qubit is built using three superconducting islands , connected by two squid loops .
acting on these loops with magnetic fluxes , one can modify the coupling of the qubits with the resonator , without changing their transition frequencies . for
further clarifications of the experimental setup involving tunable - coupling transmon qubits , see the supplemental material @xcite . by threading with magnetic fluxes at high frequencies
, one can drive simultaneous red and blue detuned sidebands , and perform collective gates @xcite .
each many - body operator can be realized as a sequence of collective and single - qubit gates , @xmath66 where the indices @xmath67 and @xmath68 run from @xmath69 to @xmath52 . in this way
, one can implement a generic ( up to @xmath70 local rotations ) @xmath52-body evolution operator @xmath71 , with @xmath72 , using @xmath73 collective gates , and a number of single qubit gates which is upper bounded by @xmath74 , counting the single qubit gate in eq .
( [ un ] ) and the @xmath70 rotations necessary to map @xmath75 to any @xmath52-body operator .
the simulation for one digital step will amount to implementing @xmath76 collective gates and a number of single qubit gates which is upper bounded by @xmath77 @xcite . and
b ) @xmath78 digital steps .
the initial state @xmath79 is chosen as in fig .
[ states]b .
periodic oscillations between the position and qubit degrees of freedom at the vertices are witnessed by @xmath80 .
the theoretical fidelity @xmath81 increases with the number of digital steps , together with the width of the fidelity bands due to experimental imperfections .
[ simulationerr ] ] we consider now the architecture of fig . [ scheme]b , in which six xmon qubits in a triangular geometry are capacitively coupled with an additional central ancillary qubit @xcite . in this case , the collective interactions @xmath82 can be decomposed and performed with pairwise @xmath83-phase gates , using the central ancillary qubit to mediate non - nearest interactions . in this way , the quantum simulation of one digital step of the hamiltonian in eq .
( [ hg ] ) will amount to realize and a number of single - qubit rotations which is upper bounded by @xmath84 .
for additional details on the experimental setup and gate counts , see @xcite . to estimate the effects of experimental imprefections on the simulations , we plot in fig .
[ simulationerr ] the dynamics of a single plaquette , starting from a state .
the observed periodic dynamical oscillations between the two paired states , depicted in fig .
[ states]b , are signaled by the periodic behavior of the overlap @xmath80 .
the simulation is run for @xmath85 and @xmath86 steps , necessary to observe one flip between the two states and one full oscillation , respectively .
on top of these oscillations , we plot fidelity bands that estimate the fidelity cap due to accumulated gate errors in the digital protocol , in the case of the two presented setups in fig . [ scheme ] , with collective and two - qubit gates . in order to see coherent oscillations ,
the fidelity of the two - body gates has to be one order of magnitude better with respect to the collective ones , with an error window of approximately @xmath87 in the collective gate case and a @xmath88 for the @xmath83-phase gate setup .
we assume a factor @xmath89 for the error loss on a single qubit gate . from the plots
, it can be seen that , while the digital fidelity @xmath81 increases , the fidelity bands due to experimental errors broaden , affecting the simulation .
the intersection of the digital fidelity line with the bands marks a regime dominated by experimental imperfections , for small @xmath54 , and one in which the error of the digital expansion prevails , for large @xmath54 .
the proposed simulation can be extended to larger lattices , where one could analyze dynamical properties of the confinement - deconfinement phase transition , which can be traced back to the breaking of the global center symmetry @xcite in groups with a non - trivial center group , like the @xmath0 case .
moreover , such a lgt simulator can be used to perform quench experiments in a ladder configuration , analyzing breaking of gluon strings between pairs of particles and anti - particles @xcite . in order to scale the quantum simulation to large qubit lattices
, one has to consider that the accumulated gate error does not depend on the size of the lattice @xcite , and only on the number of gates .
the quantum resources necessary to run the simulation scale polynomially in the lattice size , and sub - polynomially in the digital error @xcite , making the whole protocol efficient . with a current total number of gates
that exceeds 1000 @xcite , we believe that in the near future simulations of lgt on large scales will be feasible .
furthermore , the possibility of performing quantum error correction for digital quantum simulations may drastically increase its effectiveness @xcite .
am and ll acknowledge useful discussions with rami barends .
er thanks d. banerjee , t. calarco , m. dalmonte , s. montangero , u .- j .
wiese , p. zoller for discussions , in the framework of quantum technologies for lattice gauge theories .
we acknowledge financial support from basque government grants it472 - 10 and it559 - 10 , spanish mineco fis2012 - 36673-c03 - 02 , ramn y cajal grant ryc-2012 - 11391 , upv / ehu project no .
ehua14/04 , upv / ehu ufi 11/55 , promisce and scaleqit european projects . 99 i. montvay and g. mnster , _ quantum fields on a lattice _ , ( cambridge university press , cambridge , uk , 1994 ) .
h. j. rothe , _ lattice gauge theories _ , ( world scientific , singapore , 2012 ) .
i. affleck , z. zou , t. hsu , and p.w .
anderson , phys .
rev .
b * 38 * , 745 ( 1988 ) .
e. dagotto , e. fradkin , and a. moreo , phys .
rev .
b * 38 * , 2926 ( 1988 ) . x.g .
wen , _ quantum field theory of many - body systems : from the origin of sound to an origin of light and electrons _ , ( oxford graduate texts , oxford , 2004 ) .
k.g .
wilson , phys .
rev .
d * 10 * , 2445 ( 1974 ) .
j. kogut and l. susskind , phys .
rev .
d * 11 * , 395 ( 1975 ) .
s. durr , z. fodor , j. frison , c. hoelbling , r. hoffmann , s.d .
katz , s. krieg , t. kurth , l. lellouch , t. lippert , k.k .
szabo , and g. vulvert , science * 322 * , 1224 ( 2008 ) .
r. p. feynman , int .
j. theor
. phys .
* 21 * , 467 ( 1982 ) .
h. weimer , m. mller , i. lesanovsky , p. zoller , h. p. buchler , nat.phys .
* 6 * 382 ( 2010 ) .
e. kapit , e. mueller , phys .
rev .
a * 83 * , 033625 ( 2011 ) .
e. zohar , b. reznik , phys .
rev .
lett . , * 107 * , 275301 ( 2011 ) .
e. zohar , j. i. cirac , b. reznik , phys .
rev.lett . * 109*,125302 ( 2012 ) .
d. banerjee , m. dalmonte , m. mller , e. rico , p. stebler , u .- j .
wiese , p. zoller , phys .
rev .
lett . *
109 * , 175302 ( 2012 ) .
e. zohar , j.i .
cirac , b. reznik , phys .
rev .
lett . * 110 * , 055302 ( 2013 ) .
l. tagliacozzo , a. celi , a. zamora , m. lewenstein , ann . phys . * 330 * , 160 ( 2013 ) .
e. zohar , j. i. cirac , b. reznik , phys .
rev .
a * 88 * , 023617 ( 2013 ) .
d. marcos , p. widmer , e. rico , m. hafezi , p. rabl , u .- j .
wiese , p. zoller , ann .
phys . * 351 * , 634 ( 2014 ) .
k. stannigel , p. hauke , d. marcos , m. hafezi , s. diehl , m. dalmonte , p. zoller , phys .
rev .
lett .
* 112 * , 120406 ( 2014 ) .
a.w .
glaetzle , m. dalmonte , r. nath , i. rousochatzakis , r. moessner , p. zoller , phys . rev .
x * 4 * , 041037 ( 2014 ) .
s. notarnicola , e. ercolessi , p. facchi , g. marmo , s. pascazio , f.v .
pepe , j. of phys . a : math
. theor . * 48 * , 30ft01 ( 2015 ) .
a. bazavov , y. meurice , s .- w .
tsai , j. unmuth - yockey , j. zhang , e - print arxiv:1503.08354 ( 2015 ) .
l. tagliacozzo , a. celi , p. orland , and m. lewenstein , nat .
comm . * 4 * , 2615 ( 2013 ) .
d. banerjee , m. bgli , m. dalmonte , e. rico , p stebler , u .- j .
wiese , and p. zoller , phys .
rev .
lett . * 110 * , 125303 ( 2013 ) .
e. zohar , j. i. cirac , and b. reznik , phys .
rev .
lett . * 110 * , 125304 ( 2013 ) .
u .- j .
wiese , nucl .
phys .
a * 931 * , 246 ( 2014 ) .
e. zohar , j. cirac , and b. reznik , arxiv:1503.02312 ( 2015 ) .
i. m. georgescu , s. ashhab , and f. nori , rev .
mod .
phys . * 86 * , 153 ( 2014 ) .
m. h. devoret and r. schoelkopf , science * 339 * , 1169 ( 2013 ) .
u. las heras , a. mezzacapo , l. lamata , s. filipp , a. wallraff , and e. solano , phys .
rev .
lett .
* 112 * , 200501 ( 2014 ) .
u. las heras , l. garca - lvarez , a. mezzacapo , e. solano , and l. lamata , epj quantum technology * 2 * , 8 ( 2015 ) .
r. barends , l. lamata , j. kelly , l. garca - lvarez , a. g. fowler , a. megrant , e. jeffrey , t. c. white , d. sank , j. y. mutus , b. campbell , yu chen , z. chen , b. chiaro , a. dunsworth , i .-
c . hoi , c. neill , p. j. j. omalley , c. quintana , p. roushan , a. vainsencher , j. wenner , e. solano , and j. m. martinis , nat .
comm .
* 6 * , 7654 ( 2015 ) .
y. salath , m. mondal , m. oppliger , j. heinsoo , p. kurpiers , a. potonik , a. mezzacapo , u. las heras , l. lamata , e. solano , s. filipp , and a. wallraff , phys .
rev .
x * 5 * , 021027 ( 2015 ) .
r. barends , a. shabani , l. lamata , j. kelly , a. mezzacapo , u. las heras , r. babbush , a. g. fowler , b. campbell , yu chen , z. chen , b. chiaro , a. dunsworth , e. jeffrey , e. lucero , a. megrant , j. y. mutus , m. neeley , c. neill , p. j. j. omalley , c. quintana , p. roushan , d. sank , a. vainsencher , j. wenner , t. c. white , e. solano , h. neven , and john m. martinis , arxiv:1511.03316 .
b. p. lanyon , c. hempel , d. nigg , m. mller , r. gerritsma , f. zhringer , p. schindler , j. t. barreiro , m. rambach , g. kirchmair , m. hennrich , p. zoller , r. blatt , and c. f. roos , science * 334 * , 57 ( 2011 ) .
j. casanova , a. mezzacapo , l. lamata , and e. solano , phys .
rev .
lett . * 108 * , 190502 ( 2012 ) .
a. mezzacapo , j. casanova , l. lamata , and e. solano , phys .
rev .
lett . * 109 * , 200501 ( 2012 ) .
r. brower , s. chandrasekharan , and u .- j .
wiese , phys .
rev .
d * 60 * , 094502 ( 1999 ) .
see supplemental material , which includes refs .
@xcite .
v. bouchiat , d. vion , p. joyez , d. esteve , and m. h. devoret , phys .
scr .
* t76 * , 165 ( 1998 ) .
a. blais , r .- s .
huang , a. wallraff , s. m. girvin , and r. j. schoelkopf , phys .
rev .
a * 69 * , 062320 ( 2004 ) .
a. wallraff , d. i. schuster , a. blais , l. frunzio , r .- s .
huang , j. majer , s. kumar , s. m. girvin , and r. j. schoelkopf , nature * 431 * , 162 ( 2004 ) .
j. koch , t. m. yu , j. gambetta , a. a. houck , d. i. schuster , j. majer , a. blais , m. h. devoret , s. m. girvin , and r. j. schoelkopf , phys . rev .
a , * 76 * , 042319 ( 2007 ) . m. mller , k. hammerer , y. l. zhou , c. f. roos , and p. zoller , new j. phys .
* 13 * , 085007 ( 2011 )
. j. m. martinis and m. r. geller , phys .
rev .
a * 90 * , 022307 ( 2014 ) .
m. suzuki , phys .
lett . a * 146 * , 319 ( 1990 ) . d. horn , phys .
lett .
b * 100 * , 149 ( 1981 ) .
p. orland and d. rohrlich , nucl .
phys .
b , * 338 * , 647 ( 1990 )
. m. mathur , j. phys .
a : math .
gen . * 38 * , 10015 ( 2005 ) .
p. hauke , d. marcos , m. dalmonte , and p. zoller , phys .
rev .
x * 3 * , 041018 ( 2013 ) .
d. marcos , p. rabl , e. rico , and p. zoller , phys .
rev .
lett . * 111 * , 110504 ( 2013 ) .
d.s .
rokhsar and s.a .
kivelson , phys .
rev .
lett . * 61 * , 2376 ( 1988 ) .
r. moessner and s.l .
sondhi , phys .
rev .
lett . * 86 * , 1881 ( 2001 ) .
s. lloyd , science * 273 * , 1073 ( 1996 ) .
j. m. gambetta , a. a. houck , and a. blais , phys .
rev .
lett . * 106 * , 030502 ( 2011 ) .
s. j. srinivasan , a. j. hoffman , j. m. gambetta , and a. a. houck , phys .
rev .
lett . * 106 * , 083601 ( 2011 ) .
a. mezzacapo , l. lamata , s. filipp , and e. solano , phys .
rev .
lett . * 113 * , 050501 ( 2014 ) .
r. barends , j. kelly , a. megrant , d. sank , e. jeffrey , y. chen , y. yin , b. chiaro , j. mutus , c. neill , p. omalley , p. roushan , j. wenner , t. c. white , a. n. cleland , and j. m. martinis , phys .
rev .
lett . * 111 * , 080502 ( 2013 ) .
j. greensite , _ an introduction to the confinement problem _ , springer , lecture notes in phys .
* 821 * ( 2011 ) .
m. pepe and u .- j .
wiese , phys .
rev .
lett . * 102 * , 191601 ( 2009 ) .
s. khn , j. i. cirac , and m. c. bauls , e - print arxiv:1505.04441 ( 2015 ) .
t. pichler , m. dalmonte , e. rico , p. zoller , and s. montangero , e - print arxiv:1505.04440 ( 2015 ) .
d. w. berry , g. ahokas , r. cleve , and b. c. sanders , comm .
math .
phys . * 270 * , 359 ( 2007 ) .
r. barends , j. kelly , a. megrant , a. veitia , d. sank , e. jeffrey , t. c. white , j. mutus , a. g. fowler , b. campbell , y. chen , z. chen , b. chiaro , a. dunsworth , c. neill , p. o. malley , p. roushan , a. vainsencher , j. wenner , a. n. korotkov , a. n. cleland , and j. m. martinis , nature * 508 * , 500 ( 2014 ) .
|
the kerr - newman rotating black hole solution has gyromagnetic ratio @xmath0 as that of the dirac electron .
the measurable parameters of the electron : spin , mass , charge and magnetic moment determine the gravitational and electromagnetic field of electron as field of the over - rotating kerr - newman ( kn ) solution .
the corresponding space - time has topological defect the naked kerr singular ring , which forms a branch line of the kerr space into _ two sheets : _ the sheet of advanced and sheet of the retarded fields .
the kerr - schild form of metric @xciteg_=_+ 2h k_k_,[ksh ] in which @xmath1 is metric of auxiliary minkowski space @xmath2 and @xmath3 is a null vector field , @xmath4 forming the principal null congruence ( pnc ) @xmath5 the retarded and advanced sheets are related by a smooth transfer of the kerr pnc via disk @xmath6 spanned by the kerr singular ring @xmath7 ( see fig.1 ) , where @xmath8 is the kerr ellipsoidal radial coordinate . the surface @xmath9 represents a disklike `` door '' from negative sheet @xmath10 to positive one @xmath11 .
the null vector fields @xmath12 differs on these sheets , and two different null congruences @xmath13 create two different metrics g_^=_+ 2h k_^k_^[kspm ] on the same minkowski background @xmath14 twosheetedness of the kerr geometry caused search for different models of the source of kn solution , without mystery of the negative sheet .
singular metric conflicts with basic principles of quantum theory which is settled on the flat space - time and negligible gravitation .
resolution of this conflict requires `` regularization '' of space - time , which has to be done _ before quantization ,
i.e. on the classical level_. singular region has to be excised and replaced by a regular core with a flat internal metric @xmath15 matching with external kn solution .
long - term search for the models of regular source ( h. keres ( 1966 ) , w. israel ( i970 ) , v. hamity ( 1976 ) , c. lpez ( 1984 ) at al .
see refs in @xcite ) resulted in appearance of the gravitating soliton model @xcite which represents a domain - wall bubble , or a bag confining the higgs field in a superconducting false - vacuum state .
such a matter regulates the kn electromagnetic ( em ) field pushing it from interior of the soliton to domain wall boundary and results in consistency of the internal metric with flat space required by quantum theory .
the higgs mechanism of broken symmetry approaches this field model with the famous mit and slac bag models @xcite .
however , the typical quartic potential v(||)=g(|- ^2)^2 , [ phi4 ] used for the self - interacting higgs field @xmath16 , in the most of the soliton and bag - like models can not provide consistency with external kn solution .
eg , in the mit bag model , the vacuum expectation value ( vev ) of the higgs field @xmath17 vanishes inside the bag , @xmath18 and takes non - vanishing value @xmath19 _ outside the bag , _
@xmath20 therefore , the bag forms a well or cavity in the space - time filled by the higgs condensate in a superconducting phase , and this condensate turns the external electromagnetic fields in the short - range ones , spoiling the external kn solutions .
similar problem appears also in the slac bag model .
to ensure the correct external gravitational and electromagnetic ( em ) fields , the symmetry breaking mechanism should be turned inside out : the higgs condensate must be localized inside the bag , @xmath21 leaving the unbroken external vacuum state .
formation of the corresponding potential turns out to be a very non - trivial problem , and to realize it , we used in @xcite a supersymmetric scheme of phase transition with three chiral fields @xmath22 @xcite .
one of which , say @xmath23 has the required radial dependence , and we chose it as the higgs field @xmath24 setting the new notations in accord with @xmath25 the required potential @xmath26 is obtained from the superpotential @xmath27 where @xmath28 and @xmath29 are real constants .
the condition @xmath30 determines two vacuum states separated by a spike of the potential @xmath31 by @xmath32 : \(i ) external vacuum , @xmath33 , @xmath34 with vanishing higgs field @xmath35 , and \(ii ) internal state of false vacuum , @xmath36 , @xmath34 with broken symmetry , @xmath37 \(iii ) intermediate region of the domain wall phase transition @xmath32 _ the boundary of phase transition is unambiguously determined _ by matching the external kn metric @xmath38 where h= with flat internal metric @xmath39 it fixes the boundary at @xmath40 which yields r = r_e = . [ re ] since @xmath8 is the kerr oblate coordinate , the bag forms an oblate disk of the radius @xmath41 with thickness @xmath42 so that @xmath43 the choose of lpes s boundary for regularization of the kn source allows us to neglect gravity inside the source and in the zone of phase transition and consider there the space - time as flat , while outside the source the gravity and em field are unbroken .
inside the source we have only the part of lagrangian which corresponds to self - interaction of the complex higgs field and its interaction with vector - potential of the kn electromagnetic field @xmath44 in the flat space - time . the field model in zone ( iii )
is reduced to the nielsen - olesen model for a vortex string in superconducting media @xcite , _
no= -14 f_f^+ 12 ( d_)(d^)^ * + v(|h| ) , [ lno]where @xmath45 is a covariant derivative , @xmath46 and @xmath47 is reduced to derivative in flat space with a flat dalembertian @xmath48 for interaction of the complex higgs field ( x ) = |h ( x)| e^i(x ) with the maxwell field we obtain the following complicated systems of the nonlinear differential equations d_d^ & = & _ |h v , [ phiin ] + a_= i_&= & e |h|^2 ( , _ + e a _ ) .
[ main ] analysis of the equation ( [ main ] ) in @xcite showed two remarkable properties of the kn rotating soliton : * ( a ) * the vortex of the kn vector potential @xmath49 forms a quantum wilson loop placed along the border of the disk - like source , which leads to _ quantization of the angular momentum _ of the soliton , * ( b ) * the higgs condensate should _ oscillate _ inside the source with the frequency @xmath50 .
the kn vector potential has the form @xcite a_dx^= - re [ ( e r+ia ) ( dr - dt - a ^2 d ) [ am ] .
maximum of the potential is reached in the equatorial plane , @xmath51 , at the lpez s boundary of the disk - like source ( [ re ] ) , @xmath52 which plays the role of a cut - off parameter , a^max_dx^= - e r_e ( dr - dt - a d ) [ amax ] . according to ( [ main ] ) vector field
is regularized , since it can not penetrate deeply inside the higgs condensate for the phase incursion of the higgs field .
one sees , that the @xmath53 component of the kn vector potential @xmath54 forms a circular flow ( wilson loop ) near the source boundary . according to ( [ main ] ) , this flow does not penetrate inside the source beyond @xmath55 since it is compensated by gradient of the higgs phase @xmath56 integration of this relation over the closed @xmath53loop under condition @xmath57 leads to the result * ( a)*. similarly , the result * ( b ) * follows from ( [ main ] ) under condition @xmath57 for the time component of the vector potential @xmath58
the bag models give significant progress in the fermionic sector of the extended particle - like solutions . in the slac bag theory , @xcite ,
the dirac equations interacting with the classical vev of the higgs field @xmath59-field take the form ( i^_- g ) = 0 [ dir - sigma ] , where @xmath60 is a dimensionless coupling parameter .
one sees that the dirac field @xmath61 acquires from the higgs field effective mass @xmath62 which takes maximal value @xmath63 outside the bag and vanishes inside the bag .
the quarks are confined , getting inside the bag a more favorable energetic position , and this is the principal idea of the confinement mechanism . for gravitating bag model consistent with the kn solution , we follow this field mechanism , but must apply it `` inside out '' to have unbroken vacuum for gravitational and em kn solution in external region . thus , as agreed with the external gravitational and em field of the kn solution , the dirac equation will have inside the bag a nonzero bare mass @xmath64 and splits in the weyl basis into the left and right chiral parts , for spinors of the dirac bispinor @xmath65 ^ _ i _ |^= m _ , |^ i _ _ = m |^ , [ dir ] .outside the bag , @xmath66 the mass vanishes and these equations become independent ^ _ i _ |^=0 , |^ i _ _ = 0 .
[ dir0 ] it must be consistent with spinor structure of the external kn solution which is determined by the kerr theorem @xcite . the kerr - schild ( ks ) form of kn solution may be represented in via the both kerr congruences outgoing @xmath67 or ingoing @xmath68 but not via the both ones simultaneously .
taking for the _ physical sheet _ of the kn solution the outgoing kerr congruence @xmath69 corresponding to the retarded em field , we obtain the metric @xmath70 , and note that the advanced fields becomes inconsistent with this physical sheet , and being aligned with another congruence @xmath68 they should be positioned on the separate sheet which different metric @xmath71 .
this problem disappears inside the bag , where @xmath72 , and the space is flat , @xmath73 and the difference between two metrics disappears . due to
consistency conditions the gravitational interaction drops out , and the dirac equations take the ordinary flat space - time form .
the consistency conditions @xmath74 turn into equations for eigenfunctions of the helicity operator ( ) = , ( ) = - .
[ align ] one sees that the spinors @xmath75 and @xmath76 have opposite helicity , forming the `` left - handed '' @xmath75 and `` right - handed '' helicity states , aligned with outgoing direction @xmath77 and ingoing direction @xmath78 correspondingly .
the requirement of the consistency of the dirac solutions with kn solution enforces us to return the removed twosheetedness of the kerr geometry , this time to restore it outside the kn source .
therefore , two massless weyl spinors @xmath79 and @xmath80 living on the different sheets of the external kn solution are joined at the flat space inside the bag and , obtaining mass from the yukawa coupling , they form the dirac bispinor .
one more novelty for the dirac equation in the bag models is appearance of the variable mass term , which is determined by higgs condensate and turns out to be different for different regions of space - time .
the dirac wave function , solution of the dirac equation with variable mass term , avoids the region with a large bare mass , and tends to get an energetically favorable position . in the slac bag model @xcite this problem
is solved by variational approach .
the corresponding hamiltonian is h(x ) = ^(1i + g ) [ ham ] , and the energetically favorable wave function has to be determined by minimization of the averaged hamiltonian @xmath81 under the normalization condition @xmath82 it yields ( 1i + g ) = e [ h - e ] , where @xmath83 appears as the lagrangian multiplier enforcing the normalization condition . similar to results of the slac - bag model , one expects that the dirac wave function will not penetrate deep in the region of large bare mass @xmath84 and will concentrate in a narrow transition zone at the bag border @xmath85 . as it motivated in @xcite , narrow concentration of the dirac wave function
is admissible for scalar potential which does not lead to the klein paradox .
the exact solutions of this kind are known only for two - dimensional case , and the corresponding variational problem should apparently be solved numerically by using the ansatz @xmath86 in which @xmath87 is a variable factor for the dirac solution based on the weyl spinors @xmath88 consistent with the corresponding outgoing and ingoing kerr congruences .
taking the bag model conception , we should also accept the dynamical point of view that the bags are soft and may easily be deformed @xcite . by deformations bags may form stringy structures .
the typical deformation of the bag models are radial and rotational excitations with formation of the string - like open flux - tubes . the old dirac s model of an `` extensible '' spherical electron @xcite
may also be considered as a prototype of the bag model with spherically symmetric radial excitations .
the bag - like source of the kn solution without rotation , @xmath89 turns into the dirac spherical `` extensible '' electron model with classical electron radius @xmath90 the disk - like bag of the kn rotating source may be considered as the dirac spherical bag stretched by rotation to the disk of compton radius , @xmath91 which corresponds to the zone of vacuum polarization of a `` dressed '' electron .
the degree of oblateness of the kn bag is approximately @xmath92 in equatorial plane the boundary of the disk is very close to position of the kerr singular ring , field structure of which agrees with the structure of fundamental string , obtained by sen solution to low energy string theory @xcite .
regularization of the kn source , gives to this string the cut - off parameter determined by ( [ re ] ) . in accord with @xcite
this string may carry traveling waves created by the em excitations of the kn solution , and these waves must deform the bag surface .
em excitations of the kerr background @xcite are determined by analytic function a= ( y,)/p^2 [ a ] where @xmath93 is a complex projective angular variable , @xmath94 is a complex retarded - time parameter and @xmath95 for ks geometry at rest .
the ks vector potential is determined by function @xmath61 as follows @xcite = _
dx^ + = - re [ ( ) e^3 + d ] , [ alpha ] where @xmath96 the simplest solution @xmath97 is just the kn stationary solution , which creates a frozen circular em wave , which is locally plane and `` propagates '' along the kerr singular ring with the constant cut - off parameter @xmath98 , see fig.2a .
it is a zero - mode excitation of the regularized kerr string . along with many other possible stringy waves ,
interesting effect shows the simple wave solutions = e(1 + 1 y e^i ) .
[ psi12 ] function @xmath61 acts on the metric through the function @xmath99 h = .
[ hpsi ] as we have seen in sec.2 , the condition @xmath100 determines the boundary of disk @xmath101 which acts as the cut - off parameter for em field .
one sees that solution ( [ psi12 ] ) takes in equatorial plane @xmath102 the form @xmath103 and the cut - off parameter @xmath104 is not constant and vanishes at @xmath105 , creating a singular pole circulating along the kerr singular ring , @xmath106 this pole may be interpreted as a point - like bare electron in the state of zitterbewegung , or as a confined quark in the conception of the bag models .
the mysterious problem of the source for two - sheeted kerr geometry leads to a gravitating soliton model , which has to retain the external kn solution .
this condition can not be realized by the usual quadratic scheme of self - interacting higgs field and requires a supersymmetric model of phase transition , in which the higgs condensate breaks the gauge symmetry only in the core of the particle - like solution , leaving unbroken the external electromagnetic field of the kerr - newman solution .
the resulting soliton model has much in common with the famous mit and slac bag models .
in particular , the kn bag is compliant to deformations , and takes under rotations an oblate disk - like form with a circular string placed along the border of disk . for parameters of an electron the disk acquires the compton radius , indicating that the kn bag may be related to the compton region of vacuum polarization of a dressed electron .
electromagnetic excitations of the kn bag , generated by stringy traveling waves , deform the bag and create a bag - string - quark system which gives explanation to mystery of zitterbewegung of the dirac electron .
this research is supported by the rfbr grant 13 - 01 - 00602 .
|
gravitational and electromagnetic fields of an electron correspond to over - rotating kerr - newman ( kn ) solution , which has a naked singular ring and two - sheeted topology .
this solution is regularized by a solitonic source , in which singular interior is replaced by a vacuum bubble filled by the higgs field in a false - vacuum state .
field model of this kn bubble has much in common with the famous mit and slac bag models , but the geometry is `` dual '' ( turned inside out ) , leading to consistency of the kn bag model with gravity .
similar to other bag models , the kn bag is compliant to deformations , and under rotations it takes an oblate ellipsoidal form , creating a circular string along the border .
electromagnetic excitations of the kn bag generate stringy traveling waves which deform the bag , creating a traveling singular pole , included in a general bag - string - quark complex .
the dressed electron may be considered in this model as a coherent excitation of this system , confining the point - like electron ( as a quark ) in state of zitterbewegung . 2
|
theories of modified gravity ( such as @xmath0 theory , lovelock gravity , gauss - bonnet theory etc ) are constructed by adding curvature correction terms in the usual einstein - hilbert action through which the cosmic accelerated expansion might be explained @xcite ( see also @xcite for reviews on modified gravity ) .
such correction terms give rise to solutions of the field equations without invoking the concept of dark energy . to find the dynamical equations one can vary the action according to the metric .
there is no restriction on the gravitational lagrangian to be a linear function of ricci scalar @xmath1 @xcite .
recently some authors have taken into serious consideration the lagrangians that arestochastic " functions with the requirement that it should be local gauge invariant @xcite .
this mechanism was adopted in order to treat the quantization on curved spacetime .
the result was that corrective term in the einstein hilbert lagrangian arises due to either background geometry and interactions among quantum fields or gravitational self interaction @xcite .
furthermore , it is also realized that such corrective terms should be incorporated if one wants to obtain the effective action of quantum gravity on planck scale @xcite .
besides fundamental physical motivation , these theories have acquired a huge interest in cosmology as they exhibit inflationary behaviors and that the corresponding cosmological model seem very realistic @xcite . in this article
, our focus will be on the scalar - tensor - vector theory ( will be referred as mog ) and the schwarschild - mog black hole ( mog ) @xcite .
black holes can accelerate particles to arbitrarily high energy if the angular momentum of the particle is fine - tuned to some critical value ( see @xcite and references therein ) .
this phenomenon is robust as it is founded on the basic properties of geodesics around a black hole @xcite . studying the dynamics of a particle ( either massive or massless ) around the gravitational source such as a black hole ( bh )
is important because it is responsible for understanding the geometrical structure of spacetime near the bh .
geodesics may display a rich structure and convey a very reliable information to understand the geometry of the bh .
there are many types of geodesic motion but the circular geodesics are specially important .
the exponential fade - out of a collapsing star s luminosity can be explained by the circular geodesics as given in @xcite .
the motion of test particles helps to study the gravitational fields of objects experimentally and to compare the observations with the predictions about observable effects ( light deflection , gravitational time delay and perihelion shift ) . in the surrounding of the bh ,
a magnetic field is generally present @xcite , due to the presence of plasma in the vicinity of the bh . the accretion disk or a charged gas cloud is primarily responsible for the magnetic field field @xcite .
the magnetic field is stronger in the vicinity of bh s event horizon however it does not effect the geometry of the bh , but the motion of the charged particle moving around a bh is effected @xcite .
the magnetic coupling process is likely responsible for the stability of black hole with its accretion disc @xcite . according to this process
, angular momentum and energy are transferred from the black hole to its surrounding disc .
the magnetic field plays an important role in transferring sufficient energy to the surrounding particles for escaping to spacial infinity @xcite .
other interesting processes around bhs include evaporation and phantom energy accretion onto bhs @xcite . in this article , we revisit the model of zahrani et al @xcite for a mog black hole and explore the effects of modified gravity .
it involves the collision of a bounded particle with an unbounded particle in the vicinity of the black hole .
the main interest lies in finding the conditions of escape of a bounded particle after the collision .
the outline of the paper is as follows : in section ii we develop the basic equations and then derive an expression for escape velocity of a neutral particle . in section iii
we discuss the strength of magnetic field , and the equations of motion of the charged particle moving around weakly magnetized mog and escape velocity for particle also calculated .
force acting on the particle is studied in section iv and geodesics of the particles moving around the mog are discuss in section v. lyapunov exponent is explained in section vi . in section vii and viii ,
trajectories for effective potential and escape velocity of the particle are presented respectively .
conclusion is given in last section .
we will study the motion of particle in the equatorial plane to simplify the calculations . throughout this work we use the following metric signature @xmath2 and assume @xmath3 .
motion of particles around a central massive object under the affect of a central force is a well - studied problem of classical mechanics ( or rather newtonian mechanics ) .
in particular we can think of the following problem in the present context : consider a particle of mass @xmath4 moving in a circular orbit around another object of mass @xmath5 such that @xmath6 .
now for the particle to escape from the gravitational field of @xmath5 , particle s initial velocity must be more than the escape velocity .
the particle can gain the escape velocity either from an external force acting on it or by hitting ( or colliding ) a test particle with the particle in circular orbit .
since the collision leads to transfer of energy as well , the particle will escape from the circular orbit if its energy after collision is more than a critical energy or escape energy .
if however , the energy of the particle after collision is small than the critical energy , the particle falls towards @xmath5 .
the relativistic version of the above scenario was investigated by zahrani et al @xcite .
they studied the motion of a charged particle in the vicinity of a weakly magnetized schwarzschild black hole and focused on the bounded trajectory lying in the black hole equatorial plane . for the charged particle to escape from the innermost circular orbit , another particle ( which is neutral and coming from sufficiently far distance ) hits the charged particle .
the authors obtained the corresponding conditions of escape velocity and escape energy in the resulting process .
they also predicted that the motion of charged particle after collision will be chaotic due to the presence of magnetic field and strong gravitational field . although the chances of collision between two particles around black hole , in general are feeble .
however the process itself is important to describe the ejection of particles from the vicinity of black holes .
later on hussain et al @xcite investigated a similar scenario for a slowly rotating kerr black hole and discussed the conditions of escape for the particle .
jamil et al @xcite investigated a similar scenario for a schwarzschild black hole surrounded by quintessence .
recently , moffat @xcite obtained both static and non - static black hole solutions in the scalar - tensor - vector modified gravity , the theory which he himself proposed @xcite .
the theory fairly describes several astronomical and cosmological observations such as galaxy rotation curves @xcite and gravitational lensing @xcite .
the modified gravitational field equations are given by @xcite : @xmath7 where @xmath8 and @xmath9 where @xmath10 is a vector field with the source charge @xmath11 ( see details below ) .
the role of this vector field is to produce large scale repulsive gravity which can cause accelerated cosmic expansion .
further the vacuum field equations are @xmath12}=0\ ] ] where @xmath13 denotes the covariant derivative operation . to solve the above system of equations , an ansatz for the static and spherically symmetric solution is assumed of the form @xmath14 the calculation yields @xcite @xmath15 and therefore we have @xmath16 note that @xmath17 is a free parameter of the theory , hence it yields a variable gravitational _
constant_. here @xmath5 and @xmath18 is respectively mass and electric charge of the black hole and @xmath19 is the newton s gravitational constant . in metric ( [ 1 ] ) , positive value of @xmath18 is chosen to maintain repulsive gravitational force as it is necessary to describe the stable star .
( for a stable star , the gravitational attraction should be balanced by the repulsive gravity ) .
this is a static , spherically symmetric , point particle solution of a electrically charged body like reissner - nordstrom black hole @xcite . for @xmath20 ,
metric ( [ 1 ] ) reduces to the schwarzschild metric ( which is also the general relativistic limit ) . like kerr @xcite and reissner nordst@xmath21rm metrics it has two horizons : @xmath22 eq .
( [ m2 ] ) corresponds to @xmath23 and will reduce to schwarzschild event horizon for @xmath20 .
the metric ( [ 1 ] ) is invariant under time translation and rotation around symmetry axis .
thus the killing vectors equations are @xcite : @xmath24 which will give the constants of motion , where @xmath25 and @xmath26 .
the conserved quantities corresponding to these killing vectors are the total energy ( per unit mass ) @xmath27 and azimuthal angular momentum @xmath28 ( per unit mass ) of the moving particle at infinity .
motion of a neutral particle moving in mog background is described by the lagrangian density @xcite , @xmath29 from ( [ metric ] ) and ( [ m17 ] ) we can say that @xmath30 and @xmath31 are the cyclic coordinates .
there exists constants of motion corresponding to these cyclic coordinates i.e. total energy and azimuthal angular momentum .
we have calculated these integral of motion by using the euler - lagrange equation @xmath32 therefore , using equation ( [ m18a ] ) for @xmath30 and @xmath31 we have @xmath33 @xmath34 the over dot denotes the differentiation with respect to proper time @xmath35 throughout the calculations .
considering the planar motion of the particle i.e. for @xmath36 the normalization condition @xmath37 , gives @xmath38 the extreme values of the effective potential correspond to @xmath39 .
it occurs at @xmath40 for a schwarzschild black hole @xcite .
the point where the isco exists is the convolution point of the effective potential @xcite . in the present case ,
the isco occurs at @xmath41{2 } \left(l ( l-3 g m)+\alpha g m^2 g_n\right ) \left(l ( 3 g m+l)+\alpha g m^2 g_n\right)\right ) \nonumber\\ & & \bigg[3gm\bigg(-2 \alpha ^3 g^3 m^6 g_n^3 + 3 g l^4 m^2 \left(9 g-2 \alpha g_n\right)-2 l^6 - 3 \alpha g^2 l^2 m^4 g_n \left(2 \alpha g_n+9 g\right ) \nonumber\\ & & + 3 \sqrt{3}\big(g^3 l^2 m^4\bigg(-9 g l^6 + 108 g^3 l^4 m^2+\alpha g_n\big(8 l^6 - 126 g^2 l^4 m^2 \nonumber\\ & & + \alpha g m^2 g_n\big(24 l^4 - 9 g^2 l^2 m^2 + 8 \alpha g m^2 g_n \left(\alpha g m^2 g_n+3 l^2\right)\big)\big)\bigg)\big)^{\frac{1}{2 } } \bigg)^{\frac{1}{3}}\bigg]^{-1 } \nonumber\\ & & -\frac{1}{3 \sqrt[3]{2 } g m } \bigg(-2 \alpha ^3 g^3 m^6 g_n^3 + 3 g l^4 m^2
\left(9 g-2 \alpha g_n\right)-2 l^6 - 3 \alpha g^2 l^2 m^4 g_n \left(2 \alpha g_n+9 g\right ) \nonumber\\ & & + 3 \sqrt{3}\big(g^3 l^2 m^4\bigg(-9 g l^6 + 108 g^3 l^4 m^2+\alpha g_n\big(8 l^6 - 126 g^2 l^4 m^2 \nonumber\\ & & + \alpha g m^2 g_n\big(24 l^4 - 9 g^2 l^2 m^2 + 8 \alpha g m^2 g_n \left(\alpha g m^2 g_n+3 l^2\right)\big)\big)\bigg)\big)^{\frac{1}{2 } } \bigg)^{\frac{1}{3}}\end{aligned}\ ] ] the critical energy and the azimuthal angular momentum of the particle corresponding to isco are @xmath42 @xmath43 we consider the case that an incoming particle collides with the orbiting particle at the isco , so that after collision it will move within a new plane tilted with respect to previous equatorial plane . however , to study the dynamics of a particle it is convenient to use the fact that if the initial position and the tangent vector of the trajectory of the particle lies on a plane that contain the center of the body , then the entire trajectory lies on this plane . after collision
, there are three possible cases depending on the collision mechanism : ( i ) bound motion ( ii ) capture by bh ( iii ) escape to infinity . for a little change in energy and angular momentum , orbit of the particle alters very slightly . while for large changes it may escape to infinity or capture by bh depending upon the nature of change in path . after collision ,
the particle will no longer remain in the same equatorial plane , so further discussion would be dealt with respect to new plane . but note that due to spherical symmetry all equatorial planes are equivalent . due to collision particle should have new constants of motion @xmath27 , @xmath44 and @xmath28 . for simplification of our problem
we consider the case of collision when @xmath45 the azimuthal angular momentum is invariant during collision @xmath46 initial radial velocity does not change .
these conditions imply that only energy of the particle will change hence its motion would be determined by considering only the change in the energy .
after collision particle acquires an escape velocity @xmath47 , in orthogonal direction of the equatorial plane as explained in @xcite . the angular momentum and energy of the particle after collision becomes ( at the equatorial plan @xmath48 ) @xmath49 here @xmath50 and @xmath51 is the initial polar angular velocity of the particle and @xmath52 is the radius of isco .
this velocity should be in orthogonal direction to the equatorial plane @xcite .
further @xmath53 where @xmath54 is defined in eq . ( [ m3 ] )
it is clear that these values of angular momentum ( [ m10 ] ) and energy ( [ m5])are larger than their values ( [ m3 ] ) and ( [ cir2 ] ) before collision .
( [ m5 ] ) shows that as @xmath55 , @xmath56 .
so for @xmath57 particle will have unbound motion . in other words
the particle can not escape to infinity if @xmath58 .
hence the necessary condition for a particle to escape to infinity after collision is @xmath59 or @xmath60 the last expression for velocity @xmath47 is obtained by solving equation @xmath61 at @xmath62 .
the magnetic coupling ( mc ) process is responsible for attraction of black hole with its accretion disc @xcite . according to this process angular momentum and energy
are transferred from a black hole to its surrounding disc .
the process of mc provides the relation between the strength of magnetic field at the black hole horizon and its mass @xmath5 and the rate of accretion @xmath63 @xcite .
this relation is as follows : @xmath64 here black hole horizon @xmath65 is given by ( [ m2 ] ) and @xmath66 is the magnetization parameter which indicates the relative power of process of mc with respect to disc accretion .
if disc accretion dominant over mc process then @xmath67 and if mc process is dominant over disc accretion then @xmath68 while @xmath69 correspond to equipartition of these two processes .
the magnetic field expression is given by @xcite @xmath70 magnetic field strength at the horizon of mog - bh is @xmath71 gauss for @xmath69 , @xmath72 , @xmath73 and @xmath74 .
similar effects of particle collision with high center of mass energy in the vicinity of black hole can also possible if the black hole is non rotating provided there exists a magnetic field in its surrounding . since , there exists theoretical and experimental evidence that such a magnetic field should exist in the surrounding of black hole @xcite-@xcite .
we assume that this magnetic field is weak and its energy and angular momentum does not effect the background geometry of the black hole .
the above mentioned condition satisfy for a black hole of mass @xmath5 if magnetic field strength holds the condition @xcite @xmath75 these kind of black holes are called weakly magnetized . in this section
we explore how does the presence of a magnetic field in the bh exterior stimulate the motion of a charged particle .
firstly , we will calculate the magnetic field in the vicinity of black hole by following the procedure as given by @xcite the killing vector equation is @xcite @xmath76 where @xmath77 is a killing vector
( [ 34 ] ) corresponds to the maxwell equation for 4-potential @xmath78 in the lorentz gauge @xmath79 . defining @xmath78 as @xcite @xmath80 here magnetic field
is defined as @xcite @xmath81 where @xmath82 here @xmath83 is the levi civita symbol , and @xmath84 is the maxwell tensor , defined as @xmath85 for a local observer at rest , the component of four velocity are @xmath86 here we assume @xmath87 to signify the ` forward - in - time ' condition . by equations ( [ m7]-[m18 ] )
the magnetic field @xmath88-vector becomes @xmath89 the magnetic field is along with the @xmath90-axis at spatial infinity , and we assume that it is directed upward @xcite . at equatorial plane
only @xmath91 component will survive . in figure
[ fmog17 ] we have plotted @xmath91 as a function of @xmath92 .
it is decreasing initially and then become almost constant for large @xmath92 ( away from the black hole ) .
so , it is homogeneous at @xmath55 .
vs @xmath92 for @xmath93.,width=226 ] the lagrangian of the moving particle of rest mass @xmath4 and electric charge @xmath94 in the vicinity of mog - bh is @xmath95 using euler - lagrange equation ( [ m18 ] ) for @xmath30 and @xmath31 we get @xmath96 here , @xmath97 is the specific charge of a particle and @xmath98 where @xmath99 using the normalization condition @xmath100 , we obtain @xmath101 by using equations ( [ 11 ] ) and ( [ 11a ] ) in ( [ 15a ] ) and choosing @xmath48 we have @xmath102 and then effective potential is @xmath103}.\ ] ] according to equation ( [ m10 ] ) , after collision @xmath104 . hence the effective potential reduces to @xmath105}.\ ] ] putting the value of @xmath106 from equation ( [ m10 ] ) and @xmath107 in equation ( [ m8 ] ) and then solving for @xmath47 we get @xmath108.\end{aligned}\ ] ] for a particle to escape from the black hole s vicinity , its velocity should be greater or equal to @xmath47 .
a charged particle moving in an external electromagnetic field @xmath109 obeys the equation of motion : @xmath110 the dynamical equations for @xmath111 and @xmath92 respectively are @xmath112 where @xmath113 . for @xmath48 the equation ( [ 13 ] ) is satisfied obviously and equation ( [ 14 ] ) becomes @xmath114 we have solved equation ( [ 14a ] ) numerically by using the built in command of mathematica ndsolve for @xmath93 , @xmath115 and @xmath116 and plotted the solution in figure [ fmog14 ] . in this figure
the upper curve represents @xmath117 , middle is for @xmath118 ( radial velocity ) and the lower one is for @xmath119 .
@xmath118 is the radial velocity which represents the escape behavior as it is increasing with the increase of @xmath92 , and @xmath119 is the radial acceleration which is analogous to the force exerted on the particle in the radial direction after the collision . as the solution of eq .
@xmath120 for @xmath116 , @xmath115 , and @xmath93 . here
the large bold dashed curve represents @xmath121 , solid curve is for @xmath122 and short dashed curve correspond to @xmath123.,width=377 ]
as we have already calculated the effective potential for mog - bh , we can also compute the effective force on a neutral and a charged particle by @xcite , @xmath124 @xmath125 it can be seen from equation ( [ ma ] ) that the force due to scalar tensor - vector gravity is repulsive if @xmath126 .
@xmath127 equation ( [ ma1 ] ) represents the effective force for rn - bh and the force due to charge of bh is repulsive with out any condition .
we are studying the dynamics of a neutral and a charged particle in the surrounding of mog - bh where the scalar tensor - vector field produces a repulsive gravitational force which prevents a particle to fall into singularity @xcite . in figure [ fmog15 ]
we are comparing the effective force on a particle in the vicinity of mog - bh with schwarzschild black hole .
we deduce from this figure that the repulsion to reach the singularity is more for @xmath93 as compared to @xmath20 . for different values of @xmath17.,width=377 ] to study the behavior of force against magnetic field @xmath128 we have plotted the force @xmath129 as a function of @xmath92 for different value of @xmath128 in figure [ fmog16 ] .
it can be seen from this figure , due to large magnetic field strength @xmath128 , force on the particle is more attractive as compared to small @xmath128 . for different values of magnetic field @xmath128.,width=377 ]
* geodesics of a neutral particle moving around a schwarzschild bh : * + geodesics of a particle approaching toward or away from bh could be obtained by using eqs .
( [ 2 ] ) and ( [ 4 ] ) together .
we have : @xmath130 here @xmath131 correspond to a neutral particle given by ( [ 4 ] ) , where positive root gives the path of the particle out - going from the bh , and negative root gives the path of an ingoing particle .
let us consider the particle which is coming from infinity , initially at rest , approaches the bh , so setting @xmath132 , @xmath133 , @xmath20 and @xmath116 in eq .
( [ g1 ] ) we plot the geodesics in @xmath134 coordinates , see figure [ fmog18 ] .
( [ g1 ] ) against @xmath92 for @xmath132 , @xmath135 , @xmath136.,width=377 ] + * geodesics of a charged particle moving around a mog - bh : * + geodesics of the particles approaching the mog - bh could be obtained by using eqs .
( [ 11 ] ) and ( [ 15b ] ) together we obtain : @xmath137 here @xmath131 corresponds to a charged particle given by equation ( [ m8 ] ) where positive and negative signs give the path of the outgoing and ingoing particle respectively .
setting @xmath132 , @xmath135 , @xmath136 , in eq .
( [ geo1 ] ) we get the geodesics which are bounded by the boundaries @xmath138 and the outer horizon of the bh , plotted in figure [ fmog13 ] . in this figure , for thick curve we consider parameter @xmath72 , magnetic field strength @xmath139 , and charge of a particle @xmath140 and thin curve correspond to @xmath141 , @xmath142 and @xmath143 .
figure [ fmog13 ] shows that if the strength of magnetic field @xmath128 is higher , then charged particle can reach arbitrary close to black hole as compared to smaller value of @xmath128 .
( [ geo1 ] ) against @xmath92 for @xmath132 , @xmath135 , @xmath136 .
, width=377 ]
lyapunov exponents are the measurements of the rate of convergence or divergence of the nearby trajectories in the phase space .
it is highly sensitive to the initial conditions .
its positive value is the indication of divergence among the nearby trajectories .
therefore we can check the stability of orbits by the lyapunov exponent @xmath144 @xcite .
it is given by @xmath145 where @xmath146 is the isco of the particle moving around bh . in @xcite a critical component for the instability is defined as @xmath147 here @xmath148 is the instability time scale .
time period for a circular orbits can be obtained from equation ( [ 3 ] ) in the absence of magnetic field as @xmath149 and in the presence of magnetic field , form equation ( [ 11a ] ) we have @xmath150 here @xmath52 is the radius of circular orbit .
we have plotted lyapunov exponent as a function of magnetic field @xmath128 in figure [ fmog11 ] .
it shows that the larger the magnetic field strength the lesser will be @xmath144 . from figure [ fmog11 ] and equation ( [ m12 ] )
one can say that the instability of circular orbits is more for a schwarzschild black hole in comparison with the black hole immersed in a magnetic field . for @xmath151 , @xmath143 , @xmath132 , and @xmath141.,width=377 ] in figure [ fmog12 ]
we have also plotted @xmath144 against @xmath17 it decreases by the increase of @xmath17 .
hence , the stability of circular orbits is less for the schwarzschild black hole as compared to the black hole with non zero @xmath17 .
effective potential largely depends on @xmath152 component of metric . therefore , before discussing the behaviour of effective potential we will compare the @xmath152 component of mog - bh metric with rn - bh metric as it look similar . for rn - bh metric , @xmath153 for mog - bh from equation ( [ metric ] )
we have @xmath154 one can see that equation ( [ mg1 ] ) contain only @xmath155 term while equation @xmath156 contain @xmath18 and @xmath155 .
this difference leads to large change between the behaviour of effective potential of these two black holes .
the @xmath152 component of mog - bh metric also contain a parameter @xmath17 which will create a main difference between the behaviour of potential . in this section
we plot the effective potential and graphically explain the conditions on the energy of the particle required for escape to infinity or for bound motion around mog - bh . in figure [ fmog6 ] and figure [ fmog7 ]
we have plotted @xmath157 and @xmath158 respectively correspond to equation ( [ m8 ] ) . here
we will discuss the long distance and short distance behaviour of equation ( [ m8 ] ) .
one can see that for small value of @xmath92 the square root term will dominate but for large @xmath92 the term which is proportional to @xmath159 will dominate .
for @xmath160 we get some maximum value which we have mentioned in figure [ fmog7 ] as @xmath161 . for a particle to fall into the black hole its energy should be greater then @xmath161 otherwise it will bounce back to infinity or to some stable orbit . in case of @xmath162 , particle can never cross the barrier as shown in the figure @xmath163 .
hence it can not fall into the singularity . against radial
coordinate @xmath92 .
, width=377 ] as a function of @xmath92 .
, width=377 ] in figure [ fmog3 ] different regions of effective potential which correspond to escape and bound motion of the particle are shown . in figure
[ fmog3 ] , @xmath164 corresponds to region of stable orbits .
if the particle has energy equal or greater then @xmath165 , then it will fall into the black hole . @xmath161 and
@xmath166 correspond to unstable and isco orbits respectively .
if the energy of the the particle is equal or less than @xmath164 then it will reside in one of the stable orbits .
particle having energy greater than @xmath164 but less or equal to @xmath165 can have two possibilities : either it will escape to infinity or captured by the black hole .
the local minima might be related to the change in behavior of the effective potential corresponds to bh ( black brane transition ) as observed in @xcite .
correspond to stable orbits for @xmath167.,width=453 ] in figure [ fmog1 ] we are comparing the effective potential of schwarzschild - bh , rn - bh and mog - bh .
it can be seen from this figure that for large @xmath92 all the potentials approach to @xmath168 .
hence it can be concluded that the minimum energy required for a particle to escape from black hole vicinity is @xmath132 as we have mentioned before .
further the maxima for the effective potential of rn - bh is greater in comparison with the maxima of effective potential of schwarzschild - bh and mog - bh .
particle will be captured if it has energy greater than this maxima otherwise it will move back to infinity or may be reside in some stable orbit .
therefore , we can say that the possibility for a particle to escape from the vicinity of a black hole or to reside in some stable orbit is high in case of mog - bh with @xmath169 in comparison to rn - bh and mog - bh with @xmath170 .
particle can reach to the singularity depending upon its energy , but it can not reach the singularity for @xmath169 as shown by the curve , in figure [ fmog1 ] . note that @xmath171 , corresponds to a naked singularity . as a function of @xmath92 . in this figure
we are comparing the effective potential of rn - bh and mog - bh.,width=453 ] in figure [ fmog4 ] we are comparing the effective potentials in the presence with the absence of magnetic field . in this figure @xmath172 and
@xmath173 correspond to unstable orbits while @xmath174 and @xmath175 is refers to iscos .
one can notice that in the presence of magnetic field , minima of the effective potential is shifted towards the horizon and width of stable region is also increased in comparison with the case when the magnetic is absent .
this is in agreement with @xcite .
therefore we can say that magnetic field act as to increase the stability of the orbits .
( a comparison).,width=453 ] in figure [ fmog2 ] we have plotted the effective potential as a function of @xmath92 for different value of angular momentum @xmath28 .
one can see that for large value of @xmath28 , the effective potential has local minima and maxima which correspond to stable and unstable circular orbits respectively .
hence we can say that the particle with greater value of @xmath28 would have more possibility to reside in the stable orbits in comparison with lesser value of @xmath28 . as a function of @xmath92.,width=453 ] to study the behaviour of effective potential against @xmath176 , we have plotted effective potential as a function of radial
coordinate @xmath92 for different values of @xmath176 in figure [ fmog10 ] .
it can be seen that larger the value of @xmath176 greater will be the maxima of potential .
therefore , large value of @xmath176 would correspond to easy escape and vice versa . for different values of @xmath176.,width=453 ] it can be seen from the detailed analysis of the effective potential and the black hole physics that the scalar tensor - vector modified gravity differs from the einstein s theory of gravity a lot at shorter distance and it becomes similar at long distance .
for the angular variable we have , @xmath177 if the left hand side of equation ( [ a ] ) is negative then the lorentz force on the particle is attractive @xcite .
the motion of the charged particle is in clockwise direction .
lorentz force is repulsive if left hand side of ( [ a ] ) is positive .
we are not going to the detail here because it is already discussed in @xcite .
our concern is only about the action of magnetic field on the charged particle .
therefore , magnetic field may deform the oscillatory motion , so , the greater the strength of magnetic field , the larger will be the deformation of the orbit .
hence we can conclude that larger the strength of magnetic field , easy for a particle to escape from the isco .
we have also plotted in figure [ fmog8 ] escape velocity as a function of radial coordinate for different values of magnetic field parameter @xmath178 .
it can be seen that the escape velocity of the particle increases as the magnetic field strength increases but it becomes almost constant just like magnetic field , away from the bh . as the magnetic field is strong near the bh
, therefore , we can conclude that presence of magnetic field will provide more energy to particle , so that it might easily escape from the vicinity of bh .
these conclusions are consistent with @xcite . ) as a function of @xmath92 for different values of magnetic field @xmath128 .
, width=377 ] figure @xmath179 represents the escape velocity against radius @xmath92 for different values of angular momentum @xmath28 . from figure [ fmog5 ]
we can say that the possibility for a particle to escape having large angular momentum is small .
we have plotted escape velocity for different values of @xmath17 in figure [ fmog9 ] and [ fmog9 ] , @xmath20 , correspond to s - bh ( schwarzschild black hole ) and @xmath180 correspond to rn - bh . ) against @xmath92 for different values of angular momentum @xmath181.,width=453 ] in figure [ fmog9 ] we are comparing the escape velocity of a particle moving around the s - bh , rn - bh and mog - bh .
note that the difference between the velocities is larger near the black hole ( initially ) and it becomes almost same away from the black hole .
therefore , we can conclude that the effect of the charge of black hole on the motion of the particle is large while it reduces as particle moves away from it .
one can see that for large @xmath92 escape velocity is same for all values of @xmath17 but for small @xmath92 lesser value of @xmath17 correspond to greater value of escape velocity and vice versa . ) against @xmath92 for different values of parameter @xmath17.,width=377 ]
we have investigated and compare the dynamics of a charged and a neutral particle in the vicinity of s - bh , rn - bh and mog - bh .
geodesics of a neutral and a charged particle in the vicinity of a mog - bh are shown in figures [ fmog18 ] and [ fmog13 ] .
we see that for a charged particle there are two boundaries on the geodesics , @xmath182 and @xmath138 unlike the formal case in which a neutral particle comes from infinity and goes back to infinity before reaching the horizon , @xmath182 , of the bh .
we discussed the effective potential behavior in details regarding the stability of the orbits of the particle .
we further discussed the energy condition for the particle , when it will escape or its motion remains bound .
expressions for the escape velocity of the particle moving around mog - bh and for magnetic field , present in the vicinity of bh due to plasma , is derived in this work .
more importantly a comparison is done for effective potentials , obtained in the presence and absence of magnetic field .
it is found that presence of magnetic field enhance the stability of the orbits of the moving particles , due to the presence of it width of stable region in contrast to which we obtained in the case when magnetic field is absent .
we have also done the comparison of the effective potentials among the rn - bh , s - bh and mog - bh .
further we studied the stability by lyapunov exponent against magnetic field and a parameter @xmath17 .
we conclude that stability of orbits would increase due to the presence of vector field considered in mog .
we deduce that the particle have to face more repulsion to reach the singularity due to the presence of vector field as considered in mog .
but the presence of magnetic field might increase the attractive force .
it is found that a particle with large value of angular momentum @xmath28 would have greater possibility to reside in the stable orbits in comparison with a particle with lesser value of it .
therefore , escape velocity is less correspond to particle with large value of @xmath28 .
effects of magnetic field and parameter @xmath17 on escape velocity are also investigated graphically .
it is concluded that presence of magnetic field might provide sufficient energy to particle to escape easily from the surrounding of bh .
99 c. deffayet , phys .
b * 502 * ( 2001 ) 199 ; + j. s. alcaniz , phys .
d * 65 * , 123514 ( 2002 ) ; + s. m. carroll , v. duvvuri , m. trodden and m. turner , phys .
d * 70 * , 043528(2004 ) ; + s. nojiri and s. d. odintsov , phys .
b 576 ( 2003 ) 5 ; + k. atazadeh and h. r. sepangi , int .
d * 16 * , 687 ( 2007 ) ; + s. nojiri , s. d. odintsov and m. sami , phys .
d * 74 * , 046004 ( 2006 ) ; + i. navarro and k. van acoleyen , j. cosmol .
. phys . * 0603 * , 008 ( 2006 ) ; + s. capozziello , int .
d * 11 * , ( 2002 ) 483 ; + s. nojiri and s. d. odintsov , phys.rev .
d * 68 * , 123512 ( 2003 ) .
s. capozziello , m. de laurentis , phys
. rept . * 509 * , 167 ( 2011 ) ; + s. nojiri , s. d. odintsov , phys . rept .
* 505 * , 59 ( 2011 ) .
g. magnano , m. ferraris , and m. francaviglia , gen .
* 19 * , 465 ( 1987 ) .
j. d. barrow , and a. c. ottewill , j. phys .
a : math . gen .
* 16 * , 2757 ( 1983 ) . n. d. birrell and p. c. w. davies , quantum fields in curved space ( cambridge university press , 1982 )
. g. vilkovisky , class .
quantum grav .
* 9 * , 895 ( 1992 ) .
a. a. starobinsky , phys .
b * 91 * , 99 ( 1980 ) .
d. la , and p. j. steinhardt , phys .
62 , 376 ( 1989 ) ; + d. la , p. j. steinhardt and e. w. bertschinger , phys .
b 231 , 231 ( 1989 )
. j. w. moffat , eur .
j. c * 75 * , 175 ( 2015 ) .
a. zakria , m. jamil , jhep * 05 * , 147 ( 2015 ) ; + i. hussain , m. jamil , b. majeed , int . j. theor
. phys . * 54 * , 1567 ( 2015 ) .
t. harada and m. kimura , class .
quantum grav . * 31 * , 243001 ( 2014 ) .
m. a. podurets , astr .
zh . * 41 * , 1090 ( 1964 ) ( english translation in sovet astr.- aj , * 8*,868 ( 1965 ) ) .
w. l. ames and k. s. throne , j. astrophys . * 151 * , 659 ( 1968 ) . c. v. borm , m. spaans , astron .
* 553 * , l9(2013 ) .
j. c. mckinney , r. narayan , mon . not .
astron . soc .
* 375 * , 523 ( 2007 ) .
p. b. dobbie , z. kuncic , g. v. bicknell , and r. salmeron .
proceedings of iau symposium 259 : cosmic magnetic fields : from planets , to stars and galaxies ( tenerife , 2008 ) .
r. znajek , nature * 262 * , 270 ( 1976 ) .
v. p. frolov and p. krtous , phys .
d * 83 * , 024016 ( 2011 ) .
r. d. blandford , r. l. znajek , mon . not .
soc . * 179 * , 433 ( 1977 ) .
s. koide , k. shibata , t. kudoh , and d. l. meier , science * 295 * , 1688(2002 ) .
s. kide , phys .
d * 67 * , 104010 ( 2003 ) .
m. jamil , a. qadir , gen .
43 * , 1069 ( 2011 ) ; + b. nayak , m. jamil , phys .
b * 709 * , 118 ( 2012 ) ; + m. jamil , d. momeni , k. bamba , r. myrzakulov , int .
phys d * 21 * , 1250065(2012 ) ; + m. jamil , m. akbar , gen .
. grav . * 43 * , 1061 ( 2011 ) .
a. m. a. zahrani , v. p. frolov , a. a. shoom , phys . rev .
d * 87 * , 084043 ( 2013 ) .
s. hussain , i. hussain , m. jamil , eur .
j. c * 74 * , 3210 ( 2014 ) .
v. frolov , the galactic black hole ( editors : h. falcke , f. h. hehl ) , iop ( 2003 ) .
v. p. frolov , phys .
d * 85 * , 024020 ( 2012 ) .
g. nordstorm , proc .
* 29 * , 1238 ( 1918 ) .
h. reissner , ann .
physik , * 50 * , 106 ( 1916 ) .
r. p. kerr , phys .
lett . * 11 * , 237 ( 1963 ) .
t.igata , t. koike , and h. ishihara , phys .
d * 83 * , 065027 ( 2011 ) .
l. d. landau and e. m. lifshitz , the classical theory of fields ( pergamon press , oxford , 1975 ) .
s. chandrasekher , the mathematical theory of black holes ( oxford university press , 1983 ) .
b. punsly , black hole gravitohydrodynamics ( springer - verlag , berlin , 2001 ) .
d. x. wang , k. xiao , w. -h .
lei , mnras * 335 * , 655 ( 2002 ) .
wang d .- x .
, ma r .- y . , leiw .- h . , yao g .- z .
, apj * 595 * , 109 ( 2003 ) .
zhang w.m . , lu y. , zhang s. n , chin .
j. astron .
astrophys . , * 5 * , 347 ( 2005 ) , .
m. yu . piotrovich , n. a. silant , yu.n .
gnedin , t.m .
natsvlishvili , arxiv:1002.4948v1 r. m. wald , phys .
d * 10 * , 1680 ( 1974 ) .
a. a. abdujabbarov , b. j. ahmedov and n.b .
jurayeva , phys .
d * 87 * , 064042 ( 2013 ) .
a. n. aliev and n. ozdemir , mon . not .
astron . soc . *
336 * , 241 ( 1978 ) .
q. hong , c. j. hua , w. y. jiu chin .
* 31 * , 060402 ( 2014 ) .
s. fernando , arxiv:1202.1502v4 v. cardoso , a. s. miranda , e. berti , h. witeck and v. t. zanchin , phy . rev . *
d79 * , 064016 ( 2009 ) .
r. emparan and r. c. myers , jhep * 0309 * , 025 ( 2003 ) arxiv : hep - th/0308056 . v.
p. frolov and a. a. shoom , phys .
d * 82 * , 084034 ( 2010 ) .
|
* abstract : * we investigate the dynamics of a neutral and a charged particle around a black hole in modified gravity immersed in magnetic field . our focus is on the scalar - tensor - vector theory as modified gravity .
we are interested to explore the conditions on the energy of the particle under which it can escape to infinity after collision with another neutral particle in the vicinity of the black hole .
we calculate escape velocity of particle orbiting in the innermost stable circular orbit ( isco ) after the collision .
we study the effects of modified gravity on the dynamics of particles .
further we discuss how the presence of magnetic field in the vicinity of black hole , effects the motion of the orbiting particle .
we show that the stability of isco increases due to presence of magnetic field . it is observed that a particle can go arbitrary close to the black hole due to presence of magnetic field .
furthermore isco for black hole is more stable as compared with schwarzschild black hole .
we also discuss the lyapunov exponent and the effective force acting on the particle in the presence of magnetic field .
|
dissociative electron attachment ( dea ) to the water molecule proceeds through a number of channels , each with a different energetic threshold , @xmath4 the production of these species occurs via three metastable born - oppenheimer electronic states of the h@xmath3o@xmath5 system , whose vertical transition energies therefore determine the incident energies at which dea occurs .
those electronic states of the anion are the @xmath0 , @xmath1 , and @xmath2 feshbach resonances , and they are responsible for the three distinct peaks in the dea cross section .
their potential - energy surfaces contain asymptotes corresponding to the product channels listed in eq.([asymptotes ] ) , with the exception of the h+oh@xmath5 channel ; this product is a result of nonadiabatic effects . here
we report the construction of the complex - valued adiabatic potential - energy surfaces associated with these resonance states , which will be used within the local complex potential ( lcp ) model @xcite to calculate the nuclear dynamics leading to dissociation . the present study is followed by a second paper@xcite , to which we will refer as paper ii , in which we present the results of nuclear dynamics calculations under the lcp model using the calculated surfaces .
dissociative electron attachment to water was studied as early as 1930 , in the experiment of lozier@xcite , and as recently as 2006 , in the study by fedor _ _ et al.__@xcite .
these two experiments , along with the rest of the prior theoretical and experimental work on this subject@xcite , have succeeded in characterizing each of the product channels of eq.([asymptotes ] ) and the three feshbach resonances involved in their production . however , prior to our recent theoretical study of dea to water via the lowest - energy @xmath0 resonance@xcite , there had been no complete theoretical treatment of this process , nor , in fact , any _ ab initio _ treatment of dissociative attachment to any molecule , involving more than one nuclear degree of freedom
. we will give a more complete summary of the prior theoretical and experimental results concerning the dynamics of this process in paper ii .
the present treatment supersedes our previous study of dea via the lowest - energy @xmath0 state .
we have also studied the angular dependence of dea to h@xmath3o and h@xmath3s via the @xmath0 state of either anion @xcite .
additionally , we previously presented a qualitative study of the topology of the potential - energy surfaces of these three electronic states @xcite , including the many intersections that these surfaces exhibit . that qualitative study informs the present study ,
in which we construct quantitative surfaces .
two of the most significant features of these surfaces can be seen in fig .
[ c2vfig ] , reproduced here from ref.@xcite .
this figure depicts the behavior of the resonance energies with respect to the h - o - h bending angle @xmath6 , fixing the oh bond lengths at @xmath7=@xmath8=1.81@xmath9 .
the @xmath0 and @xmath1 resonances are degenerate at linear geometry ; this degeneracy will lead to renner - teller coupling between the two states .
in addition , there is a conical intersection between the @xmath1 and @xmath2 surface located at approximately @xmath6=73@xmath10 .
this conical intersection plays a crucial role for the nuclear dynamics of dea via the uppermost @xmath2 resonance state
. in @xmath11 geometry , plotted with respect to bending angle , in degrees . ]
the presence of a conical intersection between the adiabatic @xmath1 and @xmath2 states means that the nuclear dynamics calculations that employ these potential - energy surfaces must either explicitly include the coupling between them , or be performed in a diabatic basis with a smooth coupling term .
we perform an approximate diabatization of the calculated adiabatic @xmath1 and @xmath2 ( 1 and 2 @xmath12 ) states to produce diabatic @xmath1 and @xmath2 surfaces .
this diabatization is based upon the diagonalization of a particular symmetry operation described in sec .
[ sec : diab ] , not the explicit minimization of derivative matrix elements .
a resonance state may be characterized by a width , @xmath13 , and an energy , @xmath14 , which are functions of the internuclear geometry @xmath15 .
these quantities define a complex potential surface @xmath16 , @xmath17 the width @xmath13 is related to the lifetime as @xmath18 ( we use atomic units throughout ) . for a triatomic , the internal coordinates @xmath15 may be the set of bond - angle coordinates ( @xmath7 , @xmath8 , @xmath19 ) .
we use separate techniques to define the two components of the potential - energy surfaces , @xmath20 and @xmath21 .
we employ the complex kohn variational method @xcite to perform scattering calculations which include the effect of the electronic continuum upon the anion state .
these calculations yield both @xmath14 and @xmath13 , but we discard the value of @xmath22 obtained from them . in its place , we use bound - state configuration - interaction ( ci ) calculations to obtain @xmath14 . in the asymptotic regions , the resonances become bound electronic states , and therefore the use of bound - state methods is entirely appropriate in those regions of nuclear geometry . near the franck - condon region where these states are resonances , our ci treatment restricts the included configuration space to eliminate the ground state electronic continuum from the calculation .
thus our ci calculations neglect the shift in @xmath14 due to coupling with that continuum , a well - known effect explained by the feshbach resonance formalism @xcite .
since the shift in @xmath14 by this coupling is generally of the same order as the width , @xmath13 , this is an excellent approximation in regions where the resonance is narrow .
the three resonance states we treat here have as their dominant configuration an electron attached to a singly excited configuration of the neutral target ( feshbach resonances ) , and thus tend to have small widths . however , there are a some geometries where the width of one or the other of the two upper resonances is of the order of a few tenths of an ev , and in those regions the error in our calculated values of @xmath14 for that resonance are larger .
we also define a ground - state neutral h@xmath3o potential surface .
the resulting energetics of the anion surfaces relative to the ground state appear to reproduce the vertical transition energies and all but one of the two - body asymptotes of these resonances very well , without a relative vertical adjustment between the neutral and anion calculations .
however , as we will describe in detail , the ci calculations fail to produce one of the two - body breakup asymptotes on one resonance surface , and we are forced to employ an _ ad hoc _ patching procedure with another ci to correct the unphysical behavior .
a similar patching procedure was necessary to correctly describe the three body breakup asymptotes , but dynamics leading to them is not part of the study we will present in paper ii .
the data points calculated at a large set of nuclear geometries are assembled into global representations of the potential - energy surfaces .
we construct the global representations of the real part , @xmath14 , and the width , @xmath13 , separately .
the global representations of the real parts of the adiabatic @xmath0 and the diabatic @xmath1 and @xmath2 surfaces are defined by a sum of an analytic fit and a spline of the residual error of this fit .
we also define a global representation of the off - diagonal coupling between the diabatic states .
the global representations of @xmath14 for the adiabatic @xmath1 and @xmath2 states are obtained as the eigenvalues of the 2@xmath232 hamiltonian matrix defined by the fitted diabatic surfaces . the adiabatic - to - diabatic transformation angle is applied to the constructed width surfaces to obtain diabatic widths .
the final result is a complete set of complex potential surfaces for the full dynamics calculations we report in paper ii .
the outline of this paper is as follows .
we begin in sec .
[ elecsect ] with a description of the electronic structure of the resonance states and the basic features of their potential - energy surfaces , reviewing the results on the topology of those surfaces of ref .
@xcite . in sec .
[ scatsect ] we discuss the complex kohn scattering calculations from which we obtain @xmath21 , present the results of these calculations , and describe the construction of the global potential - energy surfaces from the individual data points . in sec .
[ cisect ] we do the same for the configuration - interaction calculations defining @xmath20 . in sec . [ surfsect ] we present the constructed potential - energy surfaces and discuss their features .
the electronic states that are primarily involved in dissociative electron attachment to h@xmath3o are the @xmath0 , @xmath1 , and @xmath2 feshbach resonances @xcite . at the equilibrium geometry of neutral h@xmath3o , @xmath24 , @xmath25 , where @xmath9 is the bohr radius 0.529189379 @xmath23 10@xmath26 m , these states are characterized by their dominant electronic configurations @xmath0 : 1@xmath27 2@xmath27 1@xmath28 3@xmath27
1@xmath29 4@xmath27 @xmath1 : 1@xmath27 2@xmath27 1@xmath28 3@xmath30
1@xmath31 4@xmath27 @xmath2 : 1@xmath27 2@xmath27 1@xmath32 3@xmath27
1@xmath31 4@xmath27 such that each is described as [ h@xmath3o]@xmath334@xmath27 .
the parent state of each corresponds to the analagous configuration [ h@xmath3o]@xmath334@xmath30 , which may be a singlet or a triplet .
since the resonance states are doublets , both the singlet and triplet states of the neutral may be considered the parent state to which the incident electron is attached .
the peaks in the dea cross section occur at approximately 6.5 , 8.4 , and 11.8ev , respectively , at the equilibrium geometry of the neutral ; these values approximate the vertical transition energies of the resonances .
the ionization energy of h@xmath3o is 12.621ev .
the vertical transition energy to excite the @xmath2 resonance is close this value and we can not eliminate the possibility that it is physically above the ionization energy .
it is possible , therefore , that the parent state of the @xmath2 resonance is an autoionizing excited state of neutral h@xmath3oat least at some geometries . from the equilibrium geometry of the neutral ,
the resonance states may be followed adiabatically toward the two - body breakup arrangements in which the electron is attached to one of the fragments or h - oh or o - h@xmath3 .
we identified the proper asymptotes in these two arrangements previously @xcite .
in the first of these arrangements , the @xmath0 and @xmath1 states correlate with h@xmath5+oh ( @xmath34 ) , while the @xmath2 state correlates with h@xmath5+oh ( @xmath35 ) , leaving the oh fragment excited . in the second ,
the @xmath0 state correlates with ground - state h@xmath3+o@xmath5 .
the @xmath1 state is found at a much higher energy , as h@xmath3(triplet 1@xmath361@xmath37)+o@xmath5 , and does not have a bound two - body asymptote in this arrangement .
an additional complication is that the @xmath2 surface is inherently double - valued . as described in ref .
@xcite , its lower asymptote is ground - state h@xmath3+o@xmath5 , while the upper asymptote of this surface in this arrangement is not electronically bound , and corresponds to o ( @xmath38 ) plus a resonant state of h@xmath39 . in our previous study
@xcite , we did not determine which resonant state of h@xmath39 is involved .
the double - valuedness of the @xmath2 surface presents a problem for our treatment because we are unable to fully characterize both sheets of this surface for all geometries .
we instead define a single surface which interpolates between the two sheets within the three - body breakup region .
we will attempt to identify any discrepancies between the results here and the experimental ones which we might ascribe to this omission .
however , it is likely that this feature of the adiabatic manifold only plays a significant role for three - body breakup , a channel which we will not consider in paper ii .
we do , however , include the @xmath1-@xmath2 conical intersection in the current treatment , because it is critical for a description of the dynamics of dissociative electron attachment via the @xmath2 state .
this conical intersection is a consequence of the crossing of the 1@xmath40 and 3@xmath41 orbital energies as the bond angle is varied , and has analogs in both the singlet and triplet @xmath42 and @xmath43 states of the neutral , which exhibit conical intersections near the present one . the symmetry labels @xmath0 , @xmath1 , and @xmath2 are appropriate when the h@xmath3o molecule has @xmath11 symmetry , i.e. , when the oh bond lengths are equal .
when the oh bond lengths are unequal , the molecule belongs to the c@xmath44 point group .
the appropriate symmetry labels in those geometries are @xmath45 , 1 @xmath12 , and 2 @xmath12 . due to the conical intersection
, the lower ( 1 ) and upper ( 2 ) @xmath12 states may each correspond to either @xmath1 or @xmath2 , depending on the bond angle .
a complete discussion of the topology of the anion surfaces is given in ref .
the resonance positions and widths are extracted from the results of fixed - nuclei scattering calculations carried out at physical ( real ) energies .
the scattering calculations yield an _ s _
matrix whose energy dependence is analyzed to determine the location of the resonance pole in the complex energy plane as a function of nuclear geometry .
these calculations are fully _ ab initio_. we use the complex kohn variational method @xcite , which provides a stationary principle for the _ s _ matrix . since detailed descriptions of the method have been given elsewhere @xcite , we will limit ourselves here to a brief summary to establish the terminology we will use to describe our numerical calculations .
the ( @xmath46)-electron scattering wavefunction is represented explicitly in this calculation , using the standard methods of quantum chemistry : one - electron molecular orbitals are assembled as sums of multicenter contracted cartesian gaussian functions , products of which define @xmath47-electron configuration state functions used to construct the target states .
this basis , augmented with additional gaussian functions and numerical continuum functions to describe the scattered waves , is also used to expand the ( @xmath46)-electron wave function .
thus , key components of this calculation include the appropriate choice of one - electron orbital and multielectron configuration bases defining the target states and the resonance state . not only must the resonance state be accurately represented , but also the target states into which it decays .
this requirement becomes more difficult with increasing resonance energy , as more target states become energetically accessible as decay channels . in our implementation of the complex kohn variational method
, the ( @xmath46)-electron scattering wavefunction is expanded as @xmath48 .
\end{split } \label{eq : trialwf}\ ] ] the first sum in eq .
( [ eq : trialwf ] ) is over target states @xmath49 explicitly included in a close - coupling expansion , which may be energetically open or closed .
the antisymmetrizer is denoted by @xmath50 , and the scattered wave associated with channel @xmath13 is further expanded as @xmath51 /{k_\gamma^{\frac{1}{2}}r } \end{split } \label{eq : channelfcn}\ ] ] for incoming boundary conditions in the target channel @xmath52 with initial @xmath53 quantum numbers for the incident electron .
the functions @xmath54 in eq.([eq : channelfcn ] ) , which we denote as `` scattering orbitals '' , are gaussian molecular orbitals that are orthogonal to the `` target orbitals '' used to construct the target states @xmath55 .
thus , one of the first steps in the complex kohn calculation is the partitioning of the one - electron hilbert space into the sets of target and scattering orbitals .
the functions @xmath56 and @xmath57 in eq.([eq : channelfcn ] ) are constructed from products of radial functions ( @xmath58 and @xmath59 ) times angular functions that are real - valued combinations of spherical harmonics @xmath60 consistent with the spatial symmetry of the anion .
these functions are then schmidt orthogonalized to the target and scattering molecular orbitals .
the function @xmath58 is the regular ricatti - bessel function , while @xmath61 is a numerically generated continuum function that is regular at the origin and behaves asymptotically like the outgoing riccatti - hankel function : @xmath62 it is obtained by solving the driven radial equation , @xmath63 subject to the stated boundary conditions . for all calculations presented here ,
we use @xmath64 . for energetically closed channels , only the scattering orbitals @xmath54
are included in the sum in eq.([eq : channelfcn ] ) ; the continuum functions @xmath65 and @xmath66 are not included .
thus , the calculation can give _ s _
matrices that are discontinuous across channel thresholds . to minimize this problem
, we include additional diffuse gaussian functions in the scattering orbital basis to represent the wavefunction in barely - bound channels . in any case
, the resonance energies can be formally discontinuous across channel thresholds ; their discontinuities are not due to errors in the calculation , but instead are properties of the true resonance states @xcite . the second sum in eq.([eq : trialwf ] )
is over square - integrable ( @xmath46)-electron configurations @xmath67 constructed exclusively from target orbitals . for convenience we refer to the ( @xmath46)-electron configurations as the `` _ _ q _ _ space '' and to the close - coupling part of the expansion of the wavefunction as the `` _ _ p _ _ space '' of the calculation .
space may be further divided into the bound componentbuilt from ( @xmath46)-electron configurations incorporating only gaussian orbitalsand the free " component , corresponding to the target states times continuum functions .
two approximations are made in calculating the matrix elements of the electronic hamiltonian with respect to the antisymmetrized basis functions of eq.([eq : trialwf ] ) .
the exchange portion of all matrix elements within _ p _ space that involve free " components , as well as the matrix elements between _ q _ space and the free components of _ p _ space , are assumed to be zero .
these approximations follow from the orthogonalization of the free functions to the bound molecular orbitals and the assumed completeness of the combined sets of target + scattering orbitals over the restricted region of space spanned by the target orbitals .
errors associated with these approximations are minimized by keeping the target orbitals compact and by augmenting the set of scattering orbitals with functions that extend beyond the target orbitals so that the orthogonalized continuum functions are separated from the target orbitals by a large region of space .
further details concerning these approximations can be found in refs .
@xcite .
the description of a complex kohn calculation requires the specification of the target states @xmath55 , the _ q _ space configurations @xmath67 , and , for the expansion of the channel eigenfunctions , the scattering orbitals @xmath68 and the @xmath69 pairs included in the asymptotic partial wave expansion .
we first turn our attention to the target orbitals and states . [ cols="^,^,^,^,^,^,^ " , ] the entire 1 @xmath12 surface is shown in fig .
[ 1apsurf ] .
the real part of this surface is similar to that of the @xmath0 surface , having the same asymptotes , but is everywhere higher ( except at linear @xmath6=180@xmath10 where they are degenerate ) , and is highest above the @xmath0 surface along the conical intersection seam , that intersects this figure at @xmath6=75@xmath10 at ( @xmath7=@xmath8=2.3@xmath9 ) , which point is marked with a dot in the figure .
the conical intersection seam is roughly parallel to the grain of the cut , and falls within the plane @xmath7=@xmath8 .
thus , the dot marks the point where the 1 @xmath12 surface meets the 2 @xmath12 surface along this cut .
everywhere else , it is below . near the franck - condon region of the neutral , the behavior of the 1 @xmath12 surface with respect to bending angle is different from the relatively flat @xmath0 surface . in the cut at @xmath19=150@xmath10
these two surfaces are nearly degenerate , being exactly degenerate at linear geometry . as the bending angle is decreased , however , the 1 @xmath12 surface rises in energy , and the contour lines in fig .
[ 1apsurf ] buckle outward ; one such contour line reaches a point at the conical intersection at @xmath19=75@xmath10 . from this angle ,
the contour lines near the franck - condon region relax slightly downward going to @xmath19=45@xmath10 , which behavior traces the small well on the 1 @xmath12 surface apparent in fig . [ bendfig ] around @xmath19=55@xmath10 . at smaller angles , the h@xmath3+o@xmath5 potential well develops .
the imaginary component of this state is much different from that of the @xmath0 state , being in general higher within the franck - condon region , and having the large ( 0.1ev ) peaks in the exit wells just as the resonance becomes bound as h@xmath5+oh .
these peaks are a consequence of the a symmetry of the resonance , as discussed in ref .
@xcite . they will lead to a large rate of autodetachment in this channel , and an isotope effect in the cross sections for h@xmath3o versus d@xmath3o that is disproportionate to the entrance amplitude , as discussed in ref .
they also may portend the breakdown of the lcp model near these geometries , for they indicate that the 1 @xmath12 state may exist as a virtual state there .
we examine this issue more fully in paper ii .
the 2 @xmath12 surface is shown in fig .
[ 2apsurf ] and is different from the other surfaces in every region .
its lowest points are along the well of the the h@xmath5+oh(@xmath35 ) channel where the surface drops to 8ev .
this channel has a broader and shallower potential - energy well than the h@xmath5+oh(@xmath34 ) asymptote of the other resonances , with a minimum at approximately @xmath70=1.95@xmath9 .
this surface exhibits a broad plateau in the @xmath71 plane at @xmath6=75@xmath10 and 10ev . at this angle
there is a dimple near the conical intersection at ( @xmath7=@xmath8=2.3@xmath9 ) , which is again marked with a dot .
there is no h@xmath3+o@xmath5 well evident along the symmetric stretch direction at @xmath6=15@xmath10 because in this arrangement the 2 @xmath12 state is unbound as h@xmath3 ( triplet 1@xmath361@xmath37)+o@xmath5 ; h@xmath3 ( triplet 1@xmath361@xmath37 ) has a dissociative potential - energy curve .
thus , the panel at @xmath6=15@xmath10 shows a cut along the top of this repulsive wall , going up past the last contour line at 16ev ; this repulsive wall extends infinitely in the symmetric stretch direction at ever smaller @xmath6 .
near the franck - condon region , one can see that the real part of the surface is high ( @xmath7215ev ) at @xmath6=150@xmath10 , and slopes downward going from panel to panel to @xmath6=75@xmath10 , as the contour lines near @xmath7=@xmath8=1.81@xmath9 go from being convex to concave
. moving on to the panels at 45@xmath10 , 30@xmath10 and 15@xmath10 , the repulsive wall of h@xmath3(1@xmath361@xmath37)+o@xmath5 then quickly develops .
this surface , constructed from the ci calculation , is single - valued , though the physical surface is not .
the @xmath2 shape resonance curve which intersects that of the @xmath2 feshbach resonance in branch - point fashion@xcite has been omitted from the present treatment .
the physical 2 @xmath12 surface contains not only the h@xmath3(1@xmath361@xmath37)+o@xmath5 asymptote ( overall @xmath1 symmetry ) at small h - h bond distances , but also the metastable h@xmath39(1@xmath73 1@xmath37)+o(@xmath38 ) asymptote ( overall @xmath2 symmetry ) as well ; the two three - body asymptotes of this state are thus h+h@xmath5+o(@xmath38 ) and h+h+o@xmath5 ( degenerate with 1 @xmath12 ) .
we patch the three - body breakup region of the @xmath2 surface to its upper physical asymptote at 10.71ev .
the imaginary component of the 2 @xmath12 state takes a large value ( @xmath720.12ev ) along the symmetric stretch direction for @xmath6=105@xmath10 and 150@xmath10 but otherwise is smaller ; its representation is dominated by the interpolation between the large - valued regions and the small - valued regions . in particular , in the h@xmath5+oh ( @xmath35 ) exit well the imaginary component drops to approximately 0.02ev by ( @xmath7=1.8@xmath9 , @xmath8=3.5@xmath9 ) .
the interpolation is physical when @xmath7 or @xmath8 is small .
however , in the three - body channel , we interpolate between physically distinct sheets of the @xmath12 manifold .
thus , the imaginary component drops from 100mev near the diagonal for @xmath74 going from @xmath6=105@xmath10 to @xmath6=75@xmath10 , but this behavior is unphysical , and corresponds to the interpolation between the two sheets of @xmath2 symmetry .
the imaginary component remains small along the diagonal going from @xmath6=75@xmath10 to 45@xmath10 as the conical intersection is passed and the symmetry of the 2 @xmath12 state changes from @xmath2 to @xmath1 .
we have calculated the potential - energy surfaces necessary for a description of dissociative eletron attachment to @xmath75 . in paper ii , these surfaces are used in a study of the nuclear dynamics in the local complex potential model , and the cross sections for dissociative eletron attachment are calculated .
this work was performed under the auspices of the us department of energy by the university of california lawrence berkeley national laboratory under contract de - ac02 - 05ch11231 and was supported by the u.s .
doe office of basic energy sciences , division of chemical sciences .
|
we present the results of calculations defining global , three - dimensional representations of the complex - valued potential - energy surfaces of the @xmath0 , @xmath1 , and @xmath2 metastable states of the water anion that underlie the physical process of dissociative electron attachment to water .
the real part of the resonance energies is obtained from configuration - interaction calculations performed in a restricted hilbert space , while the imaginary part of the energies ( the widths ) is derived from complex kohn scattering calculations .
a diabatization is performed on the @xmath1 and @xmath2 surfaces , due to the presence of a conical intersection between them .
we discuss the implications that the shapes of the constructed potential - energy surfaces will have upon the nuclear dynamics of dissociative electron attachment to h@xmath3o .
|
it has been a longstanding goal in the field of atom optics to realize an integrated coherent matter wave circuit @xcite .
this concept envisions a matter wave analog of an integrated optical circuit : a single device in which coherent de broglie waves would be created and then launched into waveguides where they can be switched , divided , recombined , and detected as they propagate .
research to develop coherent matter wave circuits is motivated in part by the many potential applications of this technology .
one important aim is the creation of waveguide atom interferometers @xcite , which have applications ranging from fundamental physics to various forms of sensing .
for example , when the interferometer splits the moving matter waves into two wavepackets and then recombines them after the separated wavepackets have traveled along different paths that enclose an area , the device will be sensitive to rotations through the sagnac phase @xcite .
a ring waveguide geometry should allow for a large enclosed area relative to the size of the device and also permit making many round trips in the interferometer .
these are considerable advantages over free space atom gyroscopes @xcite .
a second circuit application lies in the emerging field of atomtronics @xcite , which develops cold atom analogs of electronic devices such as diodes @xcite , transistors @xcite , and batteries @xcite . as with standard electronics , one wishes to connect these building block devices together to create complex circuits with the desired functionality for the current of atoms .
other circuit applications include precisely delivering atoms to a desired point in space with a specified velocity , realizing a kind of `` atom laser pointer '' that may be useful in technologies such as atom lithography . in this paper
we describe experiments that create simple circuits for propagating coherent atomic matter waves .
these circuits demonstrate the basic elements that are needed to engineer arbitrarily complex matter wave circuitry .
the problem of realizing the matter wave equivalent to the laser source of integrated optics , namely creating coherent de broglie waves in a waveguide , was solved in 1995 with the development of techniques to form an atomic bose - einstein condensate ( bec ) @xcite . in the past fifteen years
several of the other elements required to construct a matter wave circuit have been demonstrated individually for incoherent cold atoms that fill many spatial modes of the confining waveguide .
however , extending these techniques to operate in a single spatial mode with coherent matter waves from a bec and then incorporating the elements into integrated circuits have both proven to be difficult .
we show here how to accomplish both of these goals .
( a ) the coherent matter waves of a moving bose - einstein condensate propagate along waveguide - shaped time - averaged optical dipole potentials formed by the combination of a horizontal laser light sheet and a rapidly - moving , tightly - focused vertical laser beam .
the vertical beam paints the desired waveguide geometry , here a y - junction .
( b ) measured time - averaged laser intensity distributions used in the experiments reported here , recorded by imaging the laser intensity at the plane of the circuit onto a camera .
the intensity is rendered into three dimensions using the colour scale shown .
clockwise from top left , with image dimensions in parentheses : straight waveguide ( 114@xmath0m@xmath1114@xmath0 m ) , straight waveguides connected by a circular bend ( 93@xmath0m@xmath193@xmath0 m ) , y - junction ( 114@xmath0m@xmath1114@xmath0 m ) , and square waveguide circuit ( 62@xmath0m@xmath162@xmath0m).,width=566 ] cold atoms can be manipulated using either the magnetic stern - gerlach force or the optical dipole force . using the magnetic approach , currents in suitably - shaped conductors
have been used to guide incoherent laser - cooled atoms in straight lines @xcite , around bends @xcite , and to form y - junction splitters for such atoms @xcite .
also , becs have been launched in toroidal @xcite and linear @xcite guides produced with electromagnets .
the development of atom chips for cold atoms @xcite and bec @xcite opened up possibilities for integration @xcite , complex geometries , and miniaturization .
however , while coherent beamsplitters for stationary becs have been realized on atom chips @xcite , to date all propagation on atom chips has been restricted to linear guides @xcite .
the optical dipole force of laser light propagating inside a hollow - core optical fiber was used in the first demonstration of atom guiding @xcite .
subsequently , incoherent cold atoms from a magneto - optical trap were propagated along a miniature planar optical dipole potential waveguide above a surface @xcite .
propagation of coherent matter waves along a linear optical dipole guide formed by a collimated laser beam has been demonstrated @xcite , and overlapping laser beams have been used to create beamsplitters for guided cold atoms @xcite and for an atom laser @xcite .
micro - optics have been used to create beamsplitter and interferometer optical dipole potentials for cold atoms @xcite . in the limit where the moving bec completely fills the waveguide ,
superfluid flow has been observed in toroidal optical dipole potentials @xcite and atom - squid devices have been demonstrated @xcite . finally , digital holography has created complex optical dipole potentials that might realize an atomtronic or - gate once loading of cold atoms into the potential has been demonstrated @xcite . however , none of the experiments discussed above has demonstrated phase - coherent splitting of propagating matter waves or single - mode matter wave propagation in waveguide sections connected by bends - two essential ingredients of a coherent matter wave circuit . both
are realized in our experiment .
our circuit elements are created with the painted potential technique @xcite , a form of time - averaged optical dipole potential in which a rapidly - moving , tightly - focused laser beam superimposed on a sheet of laser light exerts confining forces on atoms through their electric polarizability .
our device is analogous to an integrated optical circuit with the roles of matter and light being reversed : while the optical circuit uses matter to guide light , here we use light to guide matter .
the matter wave source , analogous to the laser , is a bec .
the painted potential is used to draw waveguides and waveguide structures [ fig.[apparatus](a ) ] , such as bends and y - junctions [ fig.[apparatus](b ) ] .
while the proof - of - principle circuits we present here are simple , the system should be able to create any planar circuit topology that can be represented with the approximately @xmath2 resolvable spots of the two - axis acousto - optic deflector that scans the painting beam @xcite .
further , because the painted potential is dynamic the matter wave circuit can be modified arbitrarily as atoms propagate through it , a degree of flexibility that is not available in atom chips . in the experiments reported here
, we launch becs into painted waveguides that guide coherent matter waves around bends and form switches , phase coherent beamsplitters , and closed circuits . in the following sections we discuss the implementation and performance of each of these circuit elements .
our original painted potential apparatus used for creating and manipulating becs in arbitrary shapes has been described in detail elsewhere @xcite .
briefly , a cloud of cold @xmath3rb atoms is formed in a standard double magneto - optical trapping system .
after the cloud is compressed in a magnetic quadrupole trap , the painted potential overlapping the cloud is switched on . in the configuration used here and illustrated schematically in fig.[apparatus](a ) , the painted potential is formed from a combination of a horizontal light sheet with trapping frequency 433hz and a time - averaged waveguide potential of trapping frequency 1705hz and depth @xmath4k painted by a beam with waist size @xmath5 m at repetition frequency of 10khz .
reducing the intensity of the light sheet allows relatively hot atoms to evaporate from the trap until a condensate is formed in the painted potential .
the significant improvements to the flatness and resolution of the potential described in refs .
@xcite are essential for realizing the experiments reported here .
to permit slow - moving wavepackets to propagate without reflection from imperfections in the waveguide potential it is necessary to flatten the time - averaged potential to better than @xmath6 of its depth .
this is achieved by changing the intensity of the painting beam appropriately as it was scanned .
such smoothing of the potential is not possible with magnetic potentials produced by currents flowing in conductors on atom chips , which may explain why propagating matter wave circuits have not been realized with that technology .
we measure the potential flatness either by observing density variations in a bec created to fill the entire circuit or by observing the propagation dynamics of slowly - moving condensates .
the uncorrected intensity distributions in fig.[apparatus](b ) show the level of smoothness present in the time - averaged potential before this flattening process is applied .
all bec images presented here are taken in absorption , so each image corresponds to a different experimental run .
except where a period of free expansion before imaging is specified , all images depict the bec moving in the trapping potential . in
all cases condensate density increases from black to white , with blue corresponding to background ( regions of zero density ) .
a bec launched with speed 19mm / s into a straight waveguide potential similar to that shown in the top left panel of fig.[apparatus](b ) .
the launch duration was 1.3ms .
the times marked on each image are relative to the end of the launching process .
the dimensions of each panel are @xmath7m@xmath8m.,width=529 ] the experiments begin with the creation of a @xmath3rb bec in a short ( @xmath9 m long ) painted waveguide .
the bec typically contains 4000 atoms in the @xmath10 state , resulting in a chemical potential of approximately @xmath11nk .
the potential is then switched to the circuit geometry under investigation .
the bec is launched with a desired velocity by spatially modulating the intensity of the painting beam to create a linear slope in the waveguide potential for the bec to accelerate down [ fig.[launch ] ] . in this way we create velocities up to 25mm / s .
this simple launching method has a side effect of modulating the transverse trapping potential , because it also depends on the laser intensity .
the result is excitation of a small breathing oscillation in the bec .
numerical simulations with the gross - pitaevskii equation ( gpe ) show , and experiment confirms , that the amplitude of the excitation is reduced to an unimportant level by making the potential slope more gentle and the acceleration time correspondingly longer ( typically 1.5ms ) .
this effect could be avoided in future work by painting a transversely - uniform inclined plane potential on top of the waveguide potential .
we note that the axial expansion of the propagating wave packets seen in fig.[launch ] is mostly due to the repulsive interactions between the atoms of the condensate
. it would be greatly reduced , and the signal to noise ratio correspondingly improved , in a bec of @xmath12k , where the interaction strength can be tuned through zero with a magnetic field that controls a feshbach resonance @xcite .
a bend connecting two straight waveguides is an essential matter wave circuit element .
while incoherent cold atoms filling hundreds of modes of a macroscopic waveguide have been propagated around bends in a waveguide @xcite , there has been no previous demonstration of atoms kept in the ground state of a transverse confining potential while propagating around bends connecting straight waveguide sections .
fig.[bend ] shows the propagation of a bec around such a circuit element , realizing this goal .
a bec propagating with speed 19mm / s around a @xmath13 bend with radius @xmath14 m , a potential similar to that shown in the top right panel of fig.[apparatus](b ) .
the launch duration was 1.3ms .
the times marked on each image are relative to the end of the launching process .
the dimensions of each panel are @xmath15m@xmath16m.,width=529 ] the increase in the width of a bec allowed to expand before and after propagating around a bend can be used to determine the occupation of the ground state after the bend . (
a ) left : a bec allowed to expand freely for 1ms inside the light sheet of the painted potential after propagating at speed 19mm / s inside a straight guide for 2.5ms .
image dimensions are @xmath17m@xmath18 m .
right : gaussian fit to density integrated along axis of condensate .
( b ) left : a bec allowed to expand freely for 1ms inside the light sheet of the painted potential after propagating at speed 19mm / s inside a guide containing a bend of radius @xmath14 m for 3.3ms , sufficient time for the bec to propagate past the bend .
image dimensions are @xmath17m@xmath18 m .
right : gaussian fit to density integrated along axis of condensate .
( c ) results from gpe simulations of propagation of a bec with speed 14mm / s around bends of different radii followed by free expansion inside the light sheet .
the disks show the rms width after 1ms of free expansion inside the light sheet versus bend radius .
the solid line connects the points as a guide to the eye .
( d ) the disks show the condensate excited state fraction after the bend versus the rms width after 1ms of free expansion inside the light sheet for the simulation results in ( c ) .
the dashed line is a linear fit to the result .
the slope of this line is used in the analysis to connect the measured increase in rms width with the decrease in ground state occupation.,width=604 ] an important question for many applications is : to what extent is the bec in the ground state of the waveguide after it emerges from the bend ? to answer this question we image the bec after switching off the waveguide potential and allowing the condensate to expand freely inside the light sheet for 1ms .
fig.[bendanalysis](a ) shows the result obtained before the bec enters the bend .
the same type of measurement made after the bend [ fig.[bendanalysis](b ) ] contains an axial `` wiggle '' .
this structure indicates that negotiating the bend creates a collective excitation of the condensate ( not a thermal gas of non - condensed atoms ) .
numerical simulations with the gpe of the propagation of becs around bends exhibit such coherent excitations .
the simulations also show that the root mean square ( rms ) width of the condensate observed after the free expansion is a good measure of the degree of excitation ( fig.[bendanalysis ] ) . specifically ,
if @xmath19 is a normalized transverse wavefunction of the ground state in the guide , we compute for the normalized gpe solution wavefunction after the bend , @xmath20 , the axially integrated projection , @xmath21 , of @xmath22 onto @xmath23 : @xmath24 the gpe simulations show that a bec propagated through a range of bend radii exhibits varying degrees of excitation after the bend , and hence varying rms widths when the bec is allowed to expand after passing around the bend [ fig.[bendanalysis](c ) ] . for each simulation
we compute as above the excited state fraction in the waveguide , @xmath25 , after the bend .
we find that it is an approximately linear function of the rms width @xmath26 with slope @xmath27 m [ fig.[bendanalysis](d ) ] . for a condensate containing 4000atoms
the width before the bend [ fig.[bendanalysis](a ) ] is @xmath28 m , and the width after the bend [ fig.[bendanalysis](b ) ] , corrected for the slightly longer propagation time , is @xmath29 m . from the difference , 0.14(16)@xmath0 m
, we thus infer that the relative occupation of the ground state after the bend is 0.92(9 ) .
while this represents a relatively small degree of excitation ( less than @xmath30 per bend ) , a significantly lower amount of excitation would be necessary if the matter waves are to negotiate many bends without substantial excitation . as we now discuss
, there are a number of strategies available to achieve this goal .
theoretical studies of the excitation of a wave packet propagating around a circular bend @xcite have found that the degree of transverse excitation depends sensitively on an interplay between bend radius and bec velocity , with the excitation being minimal when the transverse oscillation of the condensate in the curved guide returns it to the center of the waveguide just as it exits the bend [ this effect is responsible for the oscillatory behavior seen in fig.[bendanalysis](c ) ] .
while this prediction could be studied in future experiments , there may be better solutions to reducing excitation than fine tuning the bec velocity for the particular bend radius .
an analogous problem arises in optical waveguides , where it is known that introducing an offset between the center of the bend and the center of the straight section improves coupling into the lowest mode of the curved guide @xcite . in the matter wave circuit implementation of this idea , the bend radius would be decreased slightly so that the bec enters the bend waveguide off - center by just the distance that allows the transverse waveguide potential to provide the required classical centripetal force .
alternatively , it should be possible to reduce excitations if the transition between straight waveguide and bend is made more gradual than it is for the simple circular bend employed above , where there is a discontinuous step increase in curvature at the start of the bend .
one possibility is the clothoid ( euler spiral ) employed in highway and railway transition curves , which has the useful property that the curvature is a linear function of arc length along the curve .
numerical simulations confirm that these techniques significantly reduce excitation out of the ground state at bends @xcite .
the painted potential should be able to realize both the offset and clothoid approaches .
a bec launched with speed 19mm / s into a closed waveguide formed from straight sections and four @xmath13 bends of radius @xmath31 m , forming a potential similar to that shown in the bottom left panel of fig.[apparatus](b ) .
the launch duration was 0.8ms .
the times marked on each image are relative to the end of the launching process .
the dimensions of each image are @xmath32m@xmath33m.,width=529 ] figure[stadium ] shows bends and straight sections connected together to form a closed waveguide circuit , demonstrating that several non - trivial circuit elements can be combined and function well .
such a circuit might be used , for example , to realize a sagnac interferometer by using bragg diffraction to coherently split a bec in a straight section into wave packets counter - propagating around the circuit @xcite , and then recombining them after a number of complete circuits to measure the sagnac phase .
a bec propagating through a y - junction at speed 21mm / s .
the arm separation is @xmath34 m , forming a potential similar to that shown in the bottom right panel of fig.[apparatus](b ) .
the dimensions of each image are 100@xmath0m@xmath35 m .
the propagation times on each panel are relative to the end of the launching process.,width=529 ] ( a ) absorption images showing that the splitting ratio in the setup used for fig.[yjunction ] can be controlled by altering the relative depth of the potentials in the two arms ( labeled on each panel ) .
the dimensions of each image are 86@xmath0m@xmath36 m .
the three graphs at right show the densities integrated axially .
( b ) the dependence of the splitting ratio on the ratio of the potential depths of the arms . the line is a linear fit to the data.,width=529 ] jumping between different intensity ratios in the output arms of the y - junction realizes a switch , sending the front of the wavepacket down one arm and the back of the wavepacket down the other .
the dimensions of each image are 90@xmath0m@xmath8 m .
the propagation times shown in each panel are relative to the end of the launching process.,width=529 ] a second important circuit component is the y - junction or beamsplitter .
this circuit element has not been demonstrated previously for guided coherent matter waves . we found that smoothing of the potential in the junction region was critical to making the y - junction work , which may explain why splitting of a propagating bec has not been observed previously .
further , the painted potential is not subject to the same constraints as potentials produced by static magnetic fields , where a second input waveguide merges with the input waveguide at the splitting point and the transverse confinement at that point is weak @xcite , which can lead to large perturbations when a condensate is split in that way @xcite .
figure[yjunction ] shows a bec traveling through a painted y - junction which smoothly divides it into two pieces .
altering the relative depths of the potentials in the two arms allows the splitting ratio to be tuned controllably ( fig.[beamsplitter ] ) . suddenly changing the intensity ratio from 2:1 to 1:2
while the bec is in the junction region realizes a switch that sends the front of the wave packet down one arm and the back down the other ( fig.[switch ] ) .
it should be possible to extend this scheme to switch multiple input guides between multiple output guides .
this is a simple example of how our approach can realize dynamically reconfigurable circuits . if a y - junction is to serve as a useful beamsplitter in a guided atom interferometer it must preserve the coherence of the condensate .
we demonstrated the phase coherence of the division process by allowing the center of the bec to propagate past the start of the y - junction for 2ms ( rightmost panel of fig.[yjunction ] ) and then releasing the split condensates to observe interference fringes [ fig.[phase](a ) ] after @xmath37ms of free expansion time .
this procedure realizes a traveling bec interferometer somewhat analogous to the configuration of a stationary bec with a time - dependent potential demonstrated in refs.@xcite .
we note that additional challenges arise when splitting a moving bec because of the need to avoid transferring significant forward kinetic energy into transverse excitation as the bec moves through the junction .
this is important because using a propagating bec provides a way to accomplish the applications discussed in the first paragraph of this paper .
the relative phase of the becs in the two arms can be determined by fitting a suitable function to the fringe pattern @xcite [ fig.[phase](c ) ] .
the measured phase distribution shown as blue diamonds in fig.[phase](e ) has a standard deviation of @xmath38 , much less than the standard deviation of @xmath39 expected for a uniform distribution of random relative phases .
calculations predict that tunnel coupling between the split condensates should be negligible .
we verified this by forming condensates directly in two long traps separated by the y - junction arm spacing and analyzing the fringe pattern formed after free expansion [ figs.[phase](b ) , [ phase](d ) , and [ phase](f ) ] . in this case
the phase is indeed randomly distributed over all possible angles . the observed phase coherence therefore must be established by the splitting process .
the measured phase spread is larger than the standard quantum limit @xmath40 for expected for our atom number @xmath41 .
understanding the origin of the phase spread and reducing it will be the subject of future work .
the phase coherence time is longer than we can measure at present , but we note that the fundamental limit from phase diffusion in a bec with @xmath42 atoms and chemical potential @xmath0 , given by @xmath43 @xcite , is approximately 80ms for our conditions .
close examination of the fringe pattern in fig.[phase](a ) shows that the fringes are slightly concave with respect to the central fringe .
gpe simulations confirm that this is due to excitations produced in the splitting process [ fig.[phase](e ) ] .
once these excitations are reduced , for example by fine tuning the shape of the potential at the division point , it will be interesting to connect two such y - junctions back - to - back to form a waveguide mach - zehnder interferometer .
we have shown here how to create simple circuits for coherent atomic matter waves .
the system , which uses time - averaged optical dipole potentials , is able to launch matter waves from a bec into a waveguide at a desired velocity , and then propagate the matter waves almost single - mode around bends connecting straight waveguide sections .
it can also switch propagating matter waves , and divide the moving matter waves phase - coherently to realize a simple atom interferometer .
this demonstration of the basic matter wave circuit elements opens the door to the creation of complex and dynamic matter wave circuits .
in particular , the scalable circuit technology reported here is well - suited to creating guided atom interferometers @xcite and to both realizing atomtronic devices @xcite and wiring them together to create functionality .
a next step is determining the optimum shape for bends and y - junctions to minimize excitation out of the ground state , and then implementing these geometries using the technology described here .
looking to the future , it will be possible to add further circuit functionality by painting potential structures on waveguides to act as partial ( or total ) reflectors for matter waves , extending the analogy with integrated optics .
one can of course use a simple potential barrier with appropriate height to reflect ( or transmit ) matter waves .
however this form of mirror / beamsplitter has the disadvantage for manipulating wavepackets that its reflectance and transmittance vary rapidly with de broglie wavelength .
one solution to this problem is well - known from traditional optics : replace the single potential barrier by a periodic array of weaker potentials .
this is the basic principle of broadband multilayer optical coatings and of distributed bragg reflectors .
such reflectors have been realized for becs by imposing an optical lattice on a collimated red - detuned laser beam serving as a guide @xcite . in our case
we could form a waveguide bragg reflector by painting a periodic modulation in the waveguide potential .
simulations show that with as few as six wells the transmission is almost flat for incident energies from 100% to 200% of the modulation depth .
a further next step is to combine two such reflectors on the same guide to realize a matter wave cavity .
this atom - optical configuration has been studied theoretically for almost twenty years @xcite , beginning with a proposal to use such a cavity as an extremely narrow - band velocity filter @xcite , an application which remains relevant today .
in addition , since atoms , unlike photons , can interact strongly it may be possible to operate such a cavity in a nonlinear regime . in the case of strong nonlinearity
it is predicted @xcite that an atom blockade effect will be seen , where only one atom can occupy the cavity at a time and the de broglie wave leaking out of the cavity exhibits non - classical statistics .
additional possible future directions to study with atomtronics using the technology presented here include phase squeezing in matter wave beamsplitters @xcite , quantized conductance of atoms @xcite , and creating complex superfluid circuits containing devices such as atom - squids @xcite .
0 dekker n h , lee c s , lorent v , thywissen j h , smith s p , drndi m , westervelt r m and prentiss m 2000 guiding neutral atoms on a chip _ phys .
lett . _ * 84 * 1124 schneble d , hasuo m , anker t , pfau t and mlynek j 2003 integrated atom - optical circuit with continuous - wave operation _ j. opt
. soc . am .
b _ * 20 * 648 andersson e , calarco t , folman r , andersson m , hessmo b and schmiedmayer j 2002 multimode interferometer for guided matter waves _ phys . rev .
* 88 * 100401 godun r m , darcy m b , summy g s and burnett k 2001 prospects for atom interferometry _ contemp .
phys . _ * 42 * 77 fortgh j and zimmermann c 2005 toward atom chips _ science _ * 307 * 860 cronin a d , schmiedmayer j and pritchard d e 2009 optics and interferometry with atoms and molecules _ rev .
phys . _ * 81 * 1051 gustavson t l , bouyer p and kasevich m a 1997 precision rotation measurements with an atom interferometer gyroscope _ phys . rev .
lett . _ * 78 * 2046 seaman b t , kramer m , anderson d z and holland m j 2007 atomtronics : ultracold - atom analogs of electronic devices _ phys .
rev . a _ * 75 * 023615 pepino r a , cooper j , meiser d , anderson d z and holland m j 2010 open quantum systems approach to atomtronics _ phys . rev .
a _ * 82 * 013640 ruschhaupt a and muga j g 2004 atom diode : a laser device for a unidirectional transmission of ground - state atoms _ phys .
a _ * 70 * 061604 stickney j a , anderson d z and zozulya a a 2007 transistorlike behavior of a bose - einstein condensate in a triple - well potential _ phys . rev .
a _ * 75 * 013608 pepino r a , cooper j , anderson d z and holland m j 2009 atomtronic circuits of diodes and transistors _ phys .
lett . _ * 103 * 140405 micheli a , daley a j , jaksch d and zoller p 2004 single atom transistor in a 1d optical lattice _ phys .
lett . _ * 93 * 140408 vaishnav j y , ruseckas j , clark c w and juzelinas g 2008 spin field effect transistors with ultracold atoms _ phys .
lett . _ * 101 * 2653022 zozulya a a and anderson d z 2013 principles of an atomtronic battery _ phys
. rev . a _ * 88 * 043641 anderson m h , ensher j r , matthews m r , wieman c e and cornell e a 1995 observation of bose - einstein condensation in a dilute atomic vapor _ science _ * 269 * 198 davis k b , mewes m o , andrews m r , van druten n j , durfee d s , kurn d m and ketterle w 1995 bose - einstein condensation in a gas of sodium atoms _ phys . rev .
* 75 * 3969 bradley c c , sackett c a , tollett j j and hulet r g 1995 evidence of bose - einstein condensation in an atomic gas with attractive interactions _ phys
lett . _ * 75 * 1687 key m , hughes i g , rooijakkers w , sauer b e , hinds e a , richardson d j and kazansky p g 2000 propagation of cold atoms along a miniature magnetic guide _ phys .
lett . _ * 84 * 1371 fortgh j , ott h , grossmann a and zimmermann c 2000 miniaturized magnetic guide for neutral atoms _ appl . phys .
b _ * 70 * 701 mller d , anderson d z , grow r j , schwindt p d d and cornell e a 1999 guiding neutral atoms around curves with lithographically patterned current - carrying wires _ phys .
* 83 * 5194 luo x , krger p , brugger k , wildermuth s , gimpel h , klein m w , groth s , folman r , bar - joseph i and schmiedmayer j 2004 atom fiber for omnidirectional guiding of cold neutral atoms _ opt .
* 29 * 2145 muller d , cornell e a , prevedelli m , schwindt p d d , zozulya a and anderson d z 2000 waveguide atom beam splitter for laser - cooled neutral atoms _ opt .
lett . _ * 25 * 1382 cassettari d , hessmo b , folman r , maier t and schmiedmayer j 2000 beam splitter for guided atoms _ phys . rev
. lett . _ * 85 * 5483 gupta s , murch k w
, moore k l , purdy t p and stamper - kurn d m 2005 bose - einstein condensation in a circular waveguide _ phys .
_ * 95 * 143201 arnold a s , garvie c s and riis e 2006 large magnetic storage ring for bose - einstein condensates _ phys .
a _ * 73 * 041606 garcia o , deissler b , hughes k j , reeves j m and sackett c a 2006 bose - einstein - condensate interferometer with macroscopic arm separation _ phys .
a _ * 74 * 031601 folman r , krger p , cassettari d , hessmo b , maier t and schmiedmayer j 2000 controlling cold atoms using nanofabricated surfaces : atom chips _ phys .
_ * 84 * 4749 hansel w , hommelhoff p , hnsch t w and reichel j 2001 bose - einstein condensation on a microelectronic chip _ nature _ * 413 * 498 berrada t , van frank s , bcker r , schumm t , schaff j f and schmiedmayer j 2013 integrated mach - zehnder interferometer for bose
einstein condensates _ nat .
* 4 * 2077 schumm t , hofferberth s , andersson l m , wildermuth s , groth s , bar - joseph i , schmiedmayer j and kruger p 2005 matter - wave interferometry in a double well on an atom chip _ nat .
phys . _ * 1 * 57 leanhardt a e , chikkatur a p , kielpinski d , shin y , gustavson t l , ketterle w and pritchard d e 2002 propagation of bose - einstein condensates in a magnetic waveguide _ phys . rev .
lett . _ * 89 * 040401 fortgh j , kraft s , gnther a , trck c , wicke p and zimmermann c 2004 perspectives of ultracold atoms trapped in magnetic micro potentials _ opt . commun . _ * 243 * 45 wang y - j , anderson d z , bright v m , cornell e a , diot q , kishimoto t , prentiss m , saravanan r a , segal s r and wu s 2005 atom michelson interferometer on a chip using a bose - einstein condensate _ phys .
lett . _ * 94 * 090405 renn m j , montgomery d , vdovin o , anderson d z , wieman c e and cornell e a 1995 laser - guided atoms in hollow - core optical fibers _ phys . rev .
lett . _ * 75 * 3253 guerin w , riou j f , gaebler j p , josse v , bouyer p and aspect a 2006 guided quasicontinuous atom laser _ phys .
_ * 97 * 200402 mcdonald g d , keal h , altin p a , debs j e , bennetts s , kuhn c c n , hardman k s , johnsson m t , close j d and robins n p 2013 optically guided linear mach - zehnder atom interferometer _ phys .
a _ * 87 * 013632 houde o , kadio d and pruvost l 2000 cold atom beam splitter realized with two crossing dipole guides _ phys . rev .
lett . _ * 85 * 5543 gattobigio g l , couvert a , reinaudi g , georgeot b and gury - odelin d 2012 optically guided beam splitter for propagating matter waves _ phys .
lett . _ * 109 * 030403 dumke r , mther t , volk m , ertmer w and birkl g 2002 interferometer - type structures for guided atoms _ phys .
lett . _ * 89 * 220402 ryu c , andersen m f , clade p , natarajan v , helmerson k and phillips w d 2007 observation of persistent flow of a bose - einstein condensate in a toroidal trap _ phys .
* 99 * 260401 ramanathan a , wright k c , muniz s r , zelan m , hill w t , iii , lobb c j , helmerson k , phillips w d and campbell g k 2011 superflow in a toroidal bose - einstein condensate : an atom circuit with a tunable weak link _ phys . rev
. lett . _ * 106 * 130401 moulder s , beattie s , smith r p , tammuz n and hadzibabic z 2012 quantized supercurrent decay in an annular bose - einstein condensate _ phys .
a _ * 86 * 013629 ryu c , henderson k c and boshier m g 2014 creation of matter wave bessel beams and observation of quantized circulation in a bose - einstein condensate _ new j. phys . _
* 16 * 013046 wright k c , blakestad r b , lobb c j , phillips w d and campbell g k 2013 driving phase slips in a superfluid atom circuit with a rotating weak link _ phys . rev .
lett . _ * 110 * 025302 ryu c , blackburn p w , blinova a a and boshier m g 2013 experimental realization of josephson junctions for an atom squid _ phys .
* 111 * 205301 eckel s , lee j g , jendrzejewski f , murray n , clark c w , lobb c j , phillips w d , edwards m and campbell g k 2014 hysteresis in a quantized superfluid `` atomtronic '' circuit _ nature _ * 506 * 200 gaunt a l and hadzibabic z 2012 robust digital holography for ultracold atom trapping _ sci .
* 2 * 721 henderson k , ryu c , maccormick c and boshier m g 2009 experimental demonstration of painting arbitrary and dynamic potentials for bose - einstein condensates _ new j. phys .
_ * 11 * 043030 roati g , zaccanti m , derrico c , catani j , modugno m , simoni a , inguscio m and modugno g 2007 @xmath12k bose - einstein condensate with tunable interactions _ phys . rev
. lett . _ * 99 * 010403 bromley m w j and esry b d 2004 classical aspects of ultracold atom wave packet motion through microstructured waveguide bends _ phys .
rev . a _ * 69 * 053620 kitoh t , takato n , yasu m and kawachi m 1995 bending loss reduction in silica - based waveguides by using lateral offsets _ j. lightw
. technol . _
* 13 * 555 m. g. boshier and c. ryu , to be published .
lesanovsky i , schumm t , hofferberth s , andersson l m , krger p and schmiedmayer j 2006 adiabatic radio - frequency potentials for the coherent manipulation of matter waves _ phys .
a _ * 73 * 033619 shin y , sanner c , jo g b , pasquini t a , saba m , ketterle w , pritchard d e , vengalattore m and prentiss m 2005 interference of bose - einstein condensates split with an atom chip _ phys .
a _ * 72 * 021604 shin y , saba m , pasquini t a , ketterle w , pritchard d e and leanhardt a e 2004 atom interferometry with bose - einstein condensates in a double - well potential _ phys .
lett . _ * 92 * 050405 fabre c m , cheiney p , gattobigio g l , vermersch f , faure s , mathevet r , lahaye t and gury - odelin d 2011 realization of a distributed bragg reflector for propagating guided matter waves _ phys .
* 107 * 230401 wilkens m , goldstein e , taylor b and meystre p 1993 fabry - prot interferometer for atoms _ phys .
a _ * 47 * 2366 paul t , richter k and schlagheck p 2005 nonlinear resonant transport of bose - einstein condensates _ phys .
* 94 * 020404 carusotto i 2001 nonlinear atomic fabry - perot interferometer : from the mean - field theory to the atom blockade effect _ phys .
rev . a _ * 63 * 023610 jo g b , shin y , will s , pasquini t a , saba m , ketterle w , pritchard d e , vengalattore m and prentiss m 2007 long phase coherence time and number squeezing of two bose - einstein condensates on an atom chip _ phys .
lett . _ * 98 * 030407 grond j , schmiedmayer j and hohenester u 2009 optimizing number squeezing when splitting a mesoscopic condensate _ phys .
a _ * 79 * 021603 thywissen j h , westervelt r m and prentiss m 1999 quantum point contacts for neutral atoms _ phys .
lett . _ * 83 * 3762 krinner s , stadler d , husmann d , brantut j - p and esslinger t 2015 observation of quantized conductance in neutral matter _ nature _
|
an integrated coherent matter wave circuit is a single device , analogous to an integrated optical circuit , in which coherent de broglie waves are created and then launched into waveguides where they can be switched , divided , recombined , and detected as they propagate .
applications of such circuits include guided atom interferometers , atomtronic circuits , and precisely controlled delivery of atoms .
here we report experiments demonstrating integrated circuits for guided coherent matter waves .
the circuit elements are created with the painted potential technique , a form of time - averaged optical dipole potential in which a rapidly - moving , tightly - focused laser beam exerts forces on atoms through their electric polarizability .
the source of coherent matter waves is a bose - einstein condensate ( bec ) .
we launch becs into painted waveguides that guide them around bends and form switches , phase coherent beamsplitters , and closed circuits .
these are the basic elements that are needed to engineer arbitrarily complex matter wave circuitry .
|
knowledge base is always subject to change since an environment around the knowledge base is not guaranteed to be stable forever and even some error might be included at the initial stage .
therefore , study of revision of knowledge base is very important@xcite . @xcite and
@xcite consider a revision of monotonic theories and there are a lot of researches in this direction ( see @xcite for a survey ) .
@xcite and @xcite consider an update of nonmonotonic theories to derive a given goal or a given observation . @xcite and
@xcite consider a revision of nonmonotonic theories which is more related to a revision of monotonic theories studies @xcite ; they consider a revision when inconsistency arises at addition of rules in this paper , we follow the latter approach . revision of nonmonotonic theories is especially important for ai , since it is very rare that commonsense reasoning can be represented as a monotonic theory .
however , revision of nonmonotonic theories is more complicated than revision of monotonic theory . in monotonic theory ,
if some addition of knowledge or observation leads to inconsistency , then we can avoid inconsistency by deleting a part of knowledge base . on the other hand
, we might add a piece of assumptions since deletion leads to inconsistency .
consider the following program .
@xmath0 @xmath1 @xmath2 @xmath3 means `` negation as failure '' .
the first rule says that if @xmath4 is a car and @xmath4 is not known to be broken , @xmath4 should run . since there is no information about @xmath5 and @xmath6 in the current program , @xmath7 and @xmath8 are derived .
suppose , however , that we add a rule `` @xmath9 '' meaning that car @xmath10 does not run means contradiction . ] .
then , we have no stable model , that is , we are in an inconsistent situation . to fix this inconsistency , we have ( at least ) two possible ways . 1 .
we simply discard the default rule of car @xmath10 : @xmath11 2 .
we derive @xmath5 since if we assumed @xmath12 , then contradiction would occur and thus , we have a reason to assume @xmath5 . the first revision is contraction widely used in belief revision of monotonic theories @xcite but the second is special for nonmonotonic theories such as normal logic programs . in monotonic theories ,
addition of formula can not help to restore consistency , but in nonmonotonic theories addition can help .
this phenomena were firstly observed in doyle s justification tms and he introduced _
dependency - directed backtracking_. moreover , in monotonic theories , contraction or deletion of formula can not produce any inconsistency , whereas in nonmonotonic theories , deletion can cause inconsistency . therefore , we need more functions for revision in nonmonotonic theories than monotonic ones . in this paper
, we propose a top - down procedure using abduction to compute revision for a normal logic program when there exists no stable model .
our idea of using abduction for revision is as follows .
we introduce two kinds of abducibles one of which represents a deletion of each retractable rule and the other of which represents an addition of each addable rule . for a retractable rule , we add negation of a corresponding abducible in the body of the rule so that if an instance of abducible is assumed then an instance of rule corresponding with the instance of abducible is no longer applicable . for an addable rule ,
we add a corresponding abducible in the body of the rule so that if an instance of abducible is assumed then an instance of rule corresponding with the instance of abducible becomes applicable .
then , in order to compute such abducibles to specify revision , we show that we can use a modification of satoh and iwayama s query evaluation procedure on stable models @xcite which is a combination of integrity constraint checking @xcite and abductive procedure @xcite . in stead of starting with a subprocedure which show a derivation of positive literals , we start with a subprocedure for rule consistency checking to derive abducibles to specify revision
. this procedure traverses rules of a program which is related with addition or deletion of a rule and we guarantee that a minimal revision can be found by selecting a minimal set of abducibles among all the sets of abducibles computed by the rule consistency checking procedure .
firstly , we define a revision framework as follows . in this paper , we consider a function - free normal logic program .
we use domain closure axiom and unique name axiom so that constants in the language for a program are finite and denote distinct objects .
we can easily extend our results to function - free extended logic programs by translating an extended logic program into a normal logic program proposed by @xcite . a _ rule _ @xmath13 is of the form : @xmath14 where @xmath15 , @xmath16 are atoms .
we call @xmath15 the _ head _ of the rule @xmath13 denoted as @xmath17 and @xmath18 the _ body _ of the rule denoted as @xmath19 . if @xmath20 , we sometimes call the rule an _ integrity constraint_. let @xmath21 and @xmath22 be sets of rules . a _ revision framework _ @xmath23 is a pair , @xmath24 where @xmath21 can be divided into two sets of rules @xmath25 and @xmath26 .
we call @xmath25 a _ persistent part _ of @xmath23 and @xmath26 a _ temporal part _ of @xmath23 and @xmath22 a _ backup part _ of @xmath23 .
@xmath21 expresses the current logic program which consists of @xmath25 and @xmath26 .
@xmath25 is an unchanged part which should always be satisfied such as integrity constraints whereas any part of @xmath26 can be retracted and any part of @xmath22 can be added to restore consistency .
usage of @xmath22 is inspired by back - up semantics proposed by @xcite .
we use stable model semantics for the above program .
let @xmath27 be sets of rules .
we denote a set of ground rules obtained by replacing all the variables in every rule of @xmath27 by every element in the language as @xmath28 .
let @xmath29 be a set of ground atoms and @xmath30 be the following program .
@xmath31 @xmath32 @xmath33 and @xmath34 for each @xmath35 let @xmath36 be the least model of @xmath30 .
a _ stable model _ for a logic program @xmath27 is @xmath29 iff @xmath37 and @xmath38 .
we say that @xmath27 is consistent if @xmath27 has a stable model .
now , we define a revised program in a revision framework .
[ revisedprogram ] let @xmath23 be a revision framework , @xmath39 .
let @xmath40 be a rule .
then , a _ revised program _ w.r.t . @xmath23 and @xmath40 is @xmath41 such that * @xmath42 * @xmath43 * @xmath41 is consistent .
we say for such @xmath44 and @xmath45 that a pair _
@xmath46 accomplishes revision of @xmath40 to @xmath23_. a _ minimally revised program _
w.r.t . @xmath23 and @xmath40 is @xmath41 such that there is no revised program @xmath47 such that @xmath48 and @xmath49 where @xmath50 is a strict inclusion .
[ revision ] let + @xmath25 be @xmath51 and + @xmath26 be @xmath52 and + @xmath22 be @xmath53 and + @xmath40 be `` @xmath54 '' .
+ then , @xmath55 @xmath56 , + and @xmath57 @xmath58 . for the above revision
, we have the following two minimally revised programs : 1 . a program accomplished by @xmath59 where @xmath60 and @xmath61 : + @xmath62 and @xmath63 2 . a program accomplished by @xmath64 where
@xmath65 and @xmath66 + @xmath62 and @xmath26 and @xmath67 there are other non - minimally revised programs , for example , one accomplished by @xmath68 or by @xmath69 . in the above example
, we follow giordano s approach @xcite where contrapositives of default rules are in the back - up part , but we can actually assume any rules which we think are appropriate for back - up rules when inconsistency occurs . to compute a revised program , we use a translation from a specification to an abductive logic program and compute a consistent generalized stable model for the translated program which denotes deletion and addition of rules .
an _ abductive framework _ is a pair @xmath70 where @xmath71 is a set of predicate symbols , called _ abducible predicates _ and @xmath27 is a set of rules each of whose head is not in @xmath71 .
we call a ground atom for a predicate in @xmath71 _ an abducible_. the semantics of abductive framework is based on a generalized stable model @xcite .
the following is a definition of a generalized stable model which can manipulate abducibles in abductive logic programming .
let @xmath70 be an abductive framework and @xmath72 be a set of abducibles .
a _ generalized stable model _
@xmath73 is a stable model of @xmath74 .
we say that a model @xmath73 is a _ generalized stable model with a minimal set of abducibles _
@xmath72 if there is no generalized stable model @xmath75 such that @xmath76 is a proper subset of @xmath72 .
now , we define a translation of a revision framework into an abductive framework as follows .
let @xmath23 be a revision framework @xmath39 .
we firstly give a name to every rule in @xmath26 and @xmath22 such as @xmath77 where @xmath78 is a name for the rule . a _ translation for a consistency management of @xmath23 _ ( denoted as @xmath79 )
is a set of the following translation from @xmath23 to an abductive framework @xmath70 where * @xmath80 is a name of a rule in @xmath81 @xmath82 is a name of a rule in @xmath83 * we add every rule in @xmath25 into @xmath27 .
* we translate every rule in @xmath26 with a name @xmath78 @xmath84 into the following rule in @xmath27 : @xmath85 where @xmath86 is a tuple of variables in the clause .
* we translate every rule in @xmath22 with a name @xmath78 @xmath84 into the following rule in @xmath27 : @xmath87 the following shows that an revised program corresponds with a generalized stable model .
[ deltheorem ] let @xmath23 be a revision framework @xmath39 and @xmath40 be an added clause .
@xmath88 is a ( minimally , resp . )
revised program if and only if there is a generalized stable model of @xmath89 with a ( minimal , resp . )
set of abducibles @xmath72 s.t . * @xmath90 + where @xmath78 is a name of @xmath91 . * @xmath92 + @xmath93 + where @xmath78 is a name of @xmath94 @xmath95 + @xmath96 + @xmath97 where @xmath78 is a name of @xmath98 + and @xmath99 + and @xmath100 we say that _ @xmath72 ( minimally , resp . ) realizes revision of @xmath40 to @xmath23 .
_ note that in the above theorem , we delete whole rules related to inconsistency and then add modified rules with negation of conjunctions of disequality in the body of the deleted rules .
the modified rules are logically equivalent to rules in definition [ revisedprogram ] since we assume domain closure axiom and unique name axiom .
[ translation ] consider the revision framework in example [ revision ] .
let us give names to the rules in @xmath26 and @xmath22 as follows : @xmath101 @xmath102 + then @xmath103 is : * @xmath104 .
* @xmath27 becomes as follows : + @xmath105 + @xmath106 + @xmath54 .
+ @xmath107 + @xmath108 + then , we have two generalized models with minimal abducibles : 1 . @xmath109 .
then , a minimally revised program is : + @xmath62 and @xmath110 2 .
then , a minimally revised program is : + @xmath62 and @xmath26 and @xmath67
to compute a revision , it is sufficient to compute a generalized stable model of @xmath112 .
but , if we concern a minimal revision , we need to compute all the generalized stable models and then compare sets of abducibles in these models to choose minimal sets of abducibles . for this purpose
, it is desirable to restrict sets of abducibles to be compared as small as possible .
this can be done if we compute only revision related to @xmath40 .
for example , suppose that some of temporary rules are not relevant to inconsistency of addition of @xmath40 .
if we naively compute all the generalized stable models , then we have to compare all the combination of in / out of abducibles for these irrelevant rules . in order to avoid this kind of redundancy
, we modify satoh and iwayama s query evaluation procedure on stable models @xcite .
basically , we change the order of application of subprocedures so that we can use the procedure for consistency checking .
we impose rules in a revision framework must be _ range - restricted _ , that is , any variable in a rule @xmath13 must occur in @xmath113 .
however , any rule can be translated into range - restricted form by inserting a new predicate `` @xmath114 '' describing herbrand universe for every non - range - restricted variable in the rule . before showing our procedure to compute revision
, we need the following definitions .
let @xmath115 be a literal .
then , @xmath116 denotes the complement of @xmath115 .
let @xmath27 be a logic program .
a set of resolvents w.r.t . a ground literal @xmath115 and @xmath21
, @xmath117 is the following set of rules : @xmath118 @xmath119 @xmath115 is negative and @xmath120 and @xmath121 by a ground substitution @xmath122 @xmath123 @xmath120 and @xmath124 by a ground substitution @xmath125 let @xmath27 be a logic program .
a set of deleted rules w.r.t .
a ground literal @xmath115 and @xmath27 , @xmath126 , is the following set of rules : @xmath127 @xmath128 and @xmath129 by a ground substitution @xmath125 let @xmath27 be a logic program and @xmath130 be an abducible - and - negation - removed program obtained by removing all integrity constraints in @xmath27 and all the negative literals and abducibles in the body of remaining rule and @xmath131 be the least minimal model of @xmath130.we define a _
relevant ground program _
@xmath132 for @xmath27 as follows : @xmath133 @xmath134 we briefly explain our procedure .
our procedure consists of 4 subprocedures , @xmath135 , @xmath136 , @xmath137 , and @xmath138 where @xmath139 is a non - abducible atom and @xmath140 is a set of ground literals already assumed and @xmath115 is a ground literal and @xmath13 is a rule .
@xmath135 , @xmath137 , and @xmath138 return union of @xmath140 and a set of ground literals which are assumed during the execution of the subprocedures .
@xmath136 return the above union and a substitution for @xmath139 which are made during the execution of @xmath136 . in the procedure , we have a * select * operation and a * fail * operation .
the * select * operation expresses a nondeterministic choice among alternatives .
the * fail * operation expresses immediate termination of an execution with failure .
therefore , a procedure succeeds when its inner calls of subprocedures do not encounter * fail*. we say _ a subprocedure succeeds with ( a substitution @xmath141 and ) a set of assumptions @xmath140 _ when the subprocedure successfully returns ( @xmath141 and ) @xmath140 .
our procedure firstly starts from @xmath142 .
@xmath142 checks the consistency of a rule @xmath40 with a program @xmath143 .
we can show the consistency of addition of @xmath40 by showing one of the following . 1 .
a literal @xmath115 in @xmath144 can be falsified .
to do so , we invoke subprocedure @xmath145 for @xmath116 .
every positive literal @xmath139 in @xmath144 can be made true and every negative and every abducible literal @xmath115 can be consistently assumed and @xmath146 consistent .
to do so , we invoke subprocedure @xmath147 for @xmath139 and @xmath145 for @xmath115 and @xmath146 .
the informal specification of the other 3 subprocedures is as follows .
1 . [ literalcon ] @xmath137 checks the consistency of a ground literal @xmath115 with @xmath143 and @xmath140 . to show the consistency for assuming @xmath115
, we add @xmath115 to @xmath140 ; then , we check the consistency of resolvents and deleted rules w.r.t . @xmath115 and @xmath143 .
2 . @xmath136 searches a rule @xmath13 of @xmath139 in a program @xmath143 whose body can be made true with a ground substitution @xmath141 under a set of assumptions @xmath140 . to show that every literal in the body can be made true , we call @xmath147 for non - abducible positive literals in the body
. then , we check the consistency of other literals in the body with @xmath143 and @xmath140 .
note that because of the range - restrictedness , other literals in @xmath13 become ground after all the calls of @xmath147 for non - abducible positive literals .
@xmath138 checks if a deletion of @xmath13 does not cause any contradictions with @xmath143 and @xmath140 . to show the consistency of the implicit deletion of @xmath13 , it is sufficient to prove that the head of every ground instance @xmath148 in @xmath149 can be made either true or false .
@xmath135 @xmath13 : a rule ; @xmath140 : a set of literals * begin * @xmath150 , @xmath151 ground rule @xmath152 * do * case ( a ) or case ( b ) \(a ) * select * @xmath153 @xmath154 and @xmath155 succeeds with @xmath156 @xmath157 and * continue * @xmath158 and @xmath159 succeeds with @xmath160 @xmath157 and * continue * \(b ) @xmath161 , @xmath162 @xmath153 * do * @xmath163 and @xmath164 succeeds with ( @xmath165 ) @xmath166 and * continue * @xmath167 and @xmath168 succeeds with @xmath169 @xmath166 and * continue * @xmath170 succeeds with @xmath156 @xmath157 and * continue * @xmath171 * end * ( @xmath172 ) @xmath137 @xmath115 : a ground literal ; @xmath140 : a set of literals * begin * @xmath173 * then * * return * @xmath140 @xmath174 or @xmath175 * then * * fail * @xmath176 , @xmath151 @xmath177 * do * @xmath178 succeeds with @xmath156 @xmath157 and * continue * @xmath179 * do * @xmath180 succeeds with @xmath156 @xmath157 and * continue * @xmath181 * end * ( @xmath145 ) @xmath136 @xmath139 : a non - abducible atom ; @xmath140 : a set of literals * begin * @xmath139 is ground and @xmath182 * then * * return * @xmath183 @xmath139 is ground and @xmath184 * then * * fail * @xmath185 s.t . @xmath17 and @xmath139 are unifiable with an mgu @xmath141 such a rule is not found * then * * fail * @xmath186 , @xmath187 , @xmath188 , @xmath151 @xmath189 * do * take a literal @xmath115 in @xmath190 @xmath191 succeeds with ( @xmath192 ) @xmath193 , @xmath194 , @xmath157 and * continue * @xmath195 @xmath196 * do * @xmath197 succeeds with @xmath156 @xmath157 and * continue * @xmath198 succeeds with @xmath199 @xmath200 * end * ( @xmath147 ) @xmath138 @xmath13 : a rule ; @xmath140 : a set of literals * begin * @xmath173 * then * * return * @xmath140 @xmath150 , @xmath151 ground rule @xmath201 * do * case ( a ) or case ( b ) \(a ) * if * @xmath202 succeeds with @xmath160 @xmath157 and * continue * \(b ) * if * @xmath203 succeeds with @xmath156 @xmath157 and * continue * @xmath171 * end * ( @xmath204 ) now , we describe a complete specification of the subprocedures in figure [ ruleconandliteralcon ] and figure [ deriveanddeletedcon ] . in figures ,
we denote a set of non - abducible positive literals , non - abducible negative literals , and abducibles ( either negative or positive ) in a rule @xmath13 as @xmath113 , @xmath205 and @xmath206 respectively , and we denotes empty substitution as @xmath207 , and @xmath208 expresses a composition of two substitutions @xmath209 and @xmath210 . the following theorem on correctness for rule checking can be derived from correctness on query evaluation procedure of @xcite .
[ rulecheck ] let @xmath70 be a consistent abductive framework .
suppose @xmath142 succeeds for @xmath27 with @xmath140 , then there is a generalized stable model @xmath73 for @xmath211 such that @xmath72 includes positive abducibles in @xmath140 .
par the above theorem only guarantees that @xmath13 is consistent with @xmath27 and the procedure produces _ some _ abducibles included in a generalized stable model . to compute revision , however , we must have the stronger result that @xmath140 includes all the necessary @xmath212 s and @xmath213 s .
actually , we can guarantee this by the following theorem .
let @xmath23 be a revision framework @xmath214 such that @xmath215 is consistent and @xmath40 be an added rule .
suppose @xmath142 succeeds for @xmath79 with @xmath140 , then , a set of positive abducibles in @xmath140 realizes revision of @xmath40 to @xmath23 .
the following theorem means that if we can search exhaustively in selecting the rules or cases and there is a generalized stable model whose abducibles minimally realizes a revision for addition of rule @xmath40 , then we can find such a set of abducibles by our procedure .
note that this property is always guaranteed if a program is a finite propositional program or has finite constant symbols and no function symbols .
[ complete ] let @xmath23 be a revision framework @xmath214 and @xmath40 be an added rule .
suppose that every selection of rules or cases terminates for @xmath142 with either success or failure for @xmath79 .
if @xmath72 minimally realizes revision of @xmath40 to @xmath21 , then there is a selection of rules and cases such that @xmath142 succeeds with @xmath140 where a set of abducibles in @xmath140 is equivalent to @xmath72 .
note that we can not guarantee that positive abducibles of every @xmath140 always corresponds with a minimal revision .
this problem is inherited from satoh s procedure in that it is not guaranteed for the procedure to produce a minimal abducibles . however , using the procedure , we can restrict sets of abducibles related with inconsistency and , thus , considered sets of abducibles to choosing minimal sets are smaller than sets of abducibles from a naive calculation of all the generalized stable models .
consider the revision framework in example [ revision ] .
the following are two sequences of main calls of subprocedures for @xmath142 to @xmath216 shown in example [ translation ] . in the following
, @xmath217 , @xmath218 , @xmath219 and @xmath220 corresponds with @xmath172 , @xmath145 , @xmath147 and @xmath204 respectively , and indexes in the front express a nesting structure of the calls .
note that existence of `` @xmath221 . '' in @xmath25 does not influence these derivations .
* sequence 1 * ( for @xmath222 in example [ revision ] ) @xmath223 1 @xmath224 1.1 @xmath225 1.1.1 @xmath226 @xmath227 * select * @xmath228 1.1.1.2 @xmath229 1.1.1.2.1 @xmath230 1.1.1.2.1.1 @xmath231 1.1.1.2.1.1.1 @xmath232 @xmath233 1.1.1.2.1.1.1.1 @xmath234 1.1.1.2.1.1.1.1.1 @xmath235 1.1.1.2.1.1.1.1.1.1 @xmath236 @xmath237 1.1.1.2.1.1.2 @xmath238 @xmath237 1.1.1.2.1.1.2.1 @xmath239 @xmath237 1.1.1.2.1.1.2.1.1 @xmath240 @xmath241 1.1.1.2.2 @xmath242 @xmath241 1.2 @xmath243 @xmath241 1.3 @xmath244 @xmath245 * sequence 2 * ( for @xmath246 in example [ revision ] ) @xmath223 1 @xmath224 1.1 @xmath225 1.1.1 @xmath226 @xmath227 * select * @xmath228 1.1.1.2 @xmath229 1.1.1.2.1 @xmath230 1.1.1.2.1.1 @xmath247 @xmath227 * select * @xmath248 1.1.1.2.1.1.2 @xmath249 1.1.1.2.1.1.3 @xmath250 1.1.1.2.1.1.3.1 @xmath251 @xmath252 1.1.1.2.1.1.3.1.1 @xmath253 1.1.1.2.1.1.3.1.1.1 @xmath236 @xmath254 1.1.1.2.1.1.4 @xmath255 1.1.1.2.1.1.5 @xmath256 1.1.1.2.2 @xmath242 @xmath254 1.2 @xmath257 @xmath258 1.3 @xmath259
there are works on calculation method of updates @xcite .
@xcite propose a top - down procedure to compute view updates in a database for proving a given goal , but it is not applicable to updating a normal logic program in general .
@xcite give a bottom - up procedure for an update of an acyclic program to explain given observations in the perfect model of an acyclic program and therefore , can not apply to a normal logic program which has multiple stable models .
moreover , using bottom - up computation would lead to an irrelevant derivation to an added rule . in @xcite ,
translation from an update framework @xcite to an extended logic program is provided .
differences between our translation and their translation are as follows .
* they give a translation to compute an update to explain a goal whereas we consider a revision to avoid inconsistency of addition of a rule .
* they introduce a new predicate symbol in stead of abducibles .
this makes their translation rather complex .
if we translate our translated abductive logic program to a new normal logic program by a method proposed in @xcite , the new normal logic program would be the same as their program .
* they consider addition / deletion of the whole rules to derive a given observation . that is , instead of considering deletion / addition of parts of @xmath260/@xmath261 in definition [ revisedprogram ] , they propose deletion / addition of parts of @xmath26/@xmath22 .
at least , however , to handle exception of integrity constraints in software engineering @xcite , we believe that our fine - grained approach is better since we would like to keep consistent part of integrity constraints for further checking of other data when some instances cause inconsistency .
see the detailed discussion in @xcite .
there are many procedures to compute stable models , generalized stable models or abduction .
if we use a bottom - up procedure for our translated abductive logic program to compute all the generalized stable models naively , then sets of abducibles to be compared would be larger since abducibles of irrelevant temporary rules and addable rules with inconsistency will be considered .
therefore , it is better to compute abducibles related with inconsistency . to our knowledge ,
top - down procedure which can be used for this purpose is only satoh and iwayama s procedure since we need a bottom - up consistency checking of addition / deletion of literals during computing abducibles for revision .
this task is similar to integrity constraint checking in @xcite and satoh and iwayama s procedure includes this task .
in this paper , we propose an abductive top - down procedure to compute a minimal revised program which traverses only relevant parts of the program to the added rule .
it is done by translating a revision framework of a normal logic program into an abductive logic program . as a future work
, we would like to find an efficient method of computing a minimal revision directly by combining our top - down procedure and atms - like method of memorizing justifications of revisions .
this research is partly supported by grant - in - aid for scientific research on priority areas , `` principles for constructing evolutionary software '' , the ministry of education , japan .
we also thank the anonymous referees for valuable comments on this paper .
|
this paper presents a method of computing a revision of a function - free normal logic program .
if an added rule is inconsistent with a program , that is , if it leads to a situation such that no stable model exists for a new program , then deletion and addition of rules are performed to avoid inconsistency .
we specify a revision by translating a normal logic program into an abductive logic program with abducibles to represent deletion and addition of rules .
to compute such deletion and addition , we propose an adaptation of our top - down abductive proof procedure to compute a relevant abducibles to an added rule .
we compute a minimally revised program , by choosing a minimal set of abducibles among all the sets of abducibles computed by a top - down proof procedure .
|
the high precision timing solutions for pulsars in 47 tuc have been possible due to a long dedicated observation campaign for several decades with the parkes radio telescope@xcite .
we use the updated timing solutions in the extended table 1@xcite .
based on the observed number of pulsars , the total predicted number of neutron stars in the cluster may change between 200 - 1500 due to the integrated uncertainties in the luminosity distribution , beaming , flux densities , spectral indices of pulsars ; and the initial mass function , binary fraction , encounter rate , scintillation properties of the cluster .
our n - body simulations with a @xmath21 retention fraction predict @xmath221000 neutron stars in 47 tuc .
a grid of several hundred @xmath5-body simulations of star clusters , varying the initial density profile of the cluster , it s initial half - mass radius and the mass ratio of the black hole to the total cluster mass ( @xmath23 ) were used for our study@xcite .
all simulations were produced using the gpu - enabled version of the collisional @xmath5-body code nbody6 @xcite which includes two - body relaxation , tidal fields , fitting formula for the modeling of stellar and binary - stellar evolution .
the simulations were run up to t=11.75 gyr , the age of 47 tuc@xcite , and then scaled to have the same half - mass radius as 47 [email protected] best - fitting models to the surface density and velocity dispersion profiles of 47 tuc were determined with a @xmath24 test .
all simulations were run with 100,000 particles except the simulations with 0.5% imbhs which used 200,000 particles@xcite .
the initial conditions are king ( 1962 ) models with initial concentrations of @xmath25=0.2 , 0.5 , 1.0 , 1.5 , 2.0 and 2.5 .
these values cover the observed concentrations of galactic globular clusters@xcite .
initial relaxation times of the n - body models range from 300 myr to 12 gyr with a half - mass radius @xmath26=2 pc to @xmath26=35 pc .
it is unlikely that 47 tuc has started with a half - mass relaxation time outside this range since this would require that the cluster started with either a mass or half - mass radius very different from that of observed clusters .
also we find that the best - fitting n - body models are well inside the trial grid away from the edges in the parameter space .
all models were isolated .
imbhs influence mainly the core of a star cluster and the dynamics in the core are hardly influenced by tidal effects .
therefore , the use of isolated models is justified .
the black hole was treated as a massive star in the simulations .
so the imbh was not fixed in these simulations and was allowed to wander stochastically around in the core .
all models became mass segregated , the amount of mass segregation depending on the initial half - mass relaxation time and initial concentration of the model . during the simulations , stellar mass black holes form from stellar evolution ,
concentrate near the imbh due to mass segregation , and then kick each other out due to close encounters .
stellar mass black holes have kicked each other out by t=11.75 gyr in all simulations , except those with the largest relaxation times where only a handful remain .
models with black hole masses of 1% of the cluster mass deviate significantly from the observed density profile of 47 tuc .
the discrepancy is strongest for radii around 10@xmath27 where the observed density profile has a near constant density core .
a near constant density profile at this distance is in good agreement with theoretical predictions@xcite .
the 1% imbh model also has a strong central rise in the velocity dispersion profile which is not seen in the observations .
we expect that both problems would become even more pronounced for more massive black holes .
hence the presence of black holes with mass equal to or larger than @xmath221% of the cluster mass ( @xmath228000 @xmath28 ) can already be excluded .
the 0.5% imbh model leads to a better fit of the observational data . at the core of the cluster within 5@xmath27 ,
the surface density of the 0.5% imbh model is compatible with the observed density profile . only at distances beyond
r@xmath2210@xmath27 the surface density of the model cluster is @xmath2920% lower than the observed profile .
such differences could be attributed to the uncertainties in the mass function and binary fractions in 47 tuc@xcite .
primordial binaries are expected to play an important role in the early evolution of the cluster core@xcite . for 47 tuc
, it has been estimated that primordial binaries may have pushed the core radii up to an additional 20% during the early stages of the evolution@xcite . in order to comparatively study the possible effects of primordial binaries onto the current mass segregation , we ran additional simulations for three different model clusters .
the mass function , neutron star and black hole retention fractions were kept consistent with other runs .
these simulations base code was the gpu enabled nbody6 with a hermite integration scheme for variable time steps@xcite .
this integration scheme applies chain regularization@xcite in order to have sufficient time resolution to follow the tight orbits of binaries and their close encounters with other stars over the cluster lifetime .
our first cluster was evolved with 10% primordial binaries .
we took a snapshot of that simulation at t=11.75 gyr by which time almost half of all stars left the cluster and the global binary fraction went down to 7% .
the fraction of binaries within 50% of the projected half - light radius increased to 15% at this time , fully consistent with observations of 47 tuc@xcite .
we compare this with snapshots of the other two simulations , one without primordial binaries and the other with a 0.5% imbh model .
all snapshots are taken at the same time of their evolution and scaled by their projected half - light radii to compare their stellar distributions .
extended data figure 1 shows clearly that the presence of primordial binaries does not play a significant role in shaping the current mass segregation profile of massive stars . once the imbh forms
, it becomes the dominant driver of the cluster core dynamics .
the kullback - leibler ( kl ) divergence ( @xmath12 ) provides a mathematically robust approach for a comprehensive treatment of the acceleration distributions@xcite .
@xmath12 quantifies the entropy between two distributions .
specifically , @xmath12 is proportional to @xmath30[log@xmath31 , where @xmath32 is the expectation value and @xmath13 is the likelihood .
this likelihood value has a trivial numeric relation to the measured ( @xmath33 ) and simulated ( @xmath34 ) pulsar accelerations , which then can be calculated given that the information entropy is @xmath35
the measured spin - down rates ( @xmath36 ) of galactic millisecond pulsars can be used as a proxy to disentangle the overall potential contribution of the intrinsic spin - down ( @xmath37 ) to the measured apparent accelerations . for pulsars in globular clusters @xmath38 because @xmath36 is modified by the cluster potential .
the acceleration due to the cluster potential is @xmath39 the variation in the pulsar magnetospheric geometry , formation channels and internal structure ( i.e. , moment of inertia ) , which determine @xmath37 , are expected to be similar for pulsars in clusters and the galactic disk .
therefore , the @xmath37 distribution of pulsars in clusters should be comparable to the @xmath36 distribution of galactic pulsars .
hence , we use the two - dimensional observed period - spin down ( @xmath40 ) distribution of galactic pulsars to infer the possible range for the intrinsic spin - down contribution to pulsars accelerations in 47 tuc@xcite .
the level of uncertainty due to the unknown contribution of the intrinsic @xmath41 to the measured pulsar accelerations are represented with different shades in extended data figure 2 .
because the intrinsic @xmath41 of pulsars are not known , we treat the observed range of @xmath42 and @xmath41 of galactic millisecond pulsars as a distribution of values to offset the intrinsic contribution to the measured acceleration . instead of assuming a single acceleration term
, we use the 68% , 95% , and a 99% probable intervals for the range of acceleration values corresponding to each pulsar in 47 tuc .
we use the kinematic centre ( @xmath43 , @xmath44 ) = ( @xmath45,@xmath46 ) of 47 tuc@xcite for our calculations .
the uncertainty in the centre is about @xmath47 in each coordinate@xcite . in order to study the potential contribution of an asymmetry to our calculations , we randomly shift the cluster centre within the error box and re - calculate the black hole mass with randomly selected pulsars for many bootstrap cycles .
we find that the effect of uncertainty in the cluster centre to the inferred black hole mass is less than 5% .
the simulations@xcite were produced with the gpu - enabled version of the collisional @xmath5-body code nbody6@xcite publicly available online at this http://www.ast.cam.ac.uk/~sverre/web/pages/nbody.htm[url ] .
the authors are happy to provide further details for reproducing the simulations .
00247205aa & 00:24:07.2779(9 ) & 72:05:19.520(4 ) & [email protected]@xmath27 & 1.8453805304199(2 ) & 4.5918(2 ) + 00247204ab & 00:24:08.1657(4 ) & 72:04:47.616(2 ) & [email protected]@xmath27 & 3.7046395539391(2 ) & + 0.9844(3 ) + 00237204c & 00:23:50.35448(4 ) & 72:04:31.5047(1 ) & [email protected]@xmath27 & 5.75677999551637(5 ) & 4.98500(7 ) + 00247204d & 00:24:13.88092(3 ) & 72:04:43.8524(1 ) & [email protected]@xmath27 & 5.35757328486572(5 ) & 0.34220(9 ) + 00247205e & 00:24:11.10309(7 ) & 72:05:20.1453(2 ) & [email protected]@xmath27 & 3.53632914765559(8 ) & + 9.85126(8 ) + 00247204f & 00:24:03.85546(5 ) & 72:04:42.8183(2 ) & [email protected]@xmath27 & 2.62357935251262(3 ) & + 6.45031(6 ) + 00247204 g & 00:24:07.9603(1 ) & 72:04:39.7029(4 ) & [email protected]@xmath27 & 4.0403791435651(1 ) & 4.2159(2 ) + 00247204h & 00:24:06.7031(1 ) & 72:04:06.8069(5 ) & [email protected]@xmath27 & 3.21034070935036(8 ) & 0.1827(1 ) + 00247204i & 00:24:07.9348(1 ) & 72:04:39.6809(4 ) & [email protected]@xmath27 & 3.4849920616629(1 ) & 4.5874(2 ) + 00237203j & 00:23:59.40763(1 ) & 72:03:58.79077(4 ) & [email protected]@xmath27 & 2.100633545352538(6 ) & 0.97932(1 ) + 00247204l & 00:24:03.7719(3 ) & 72:04:56.9240(9 ) & [email protected]@xmath27 & 4.3461679994613(2 ) & 12.20406(9 ) + 00237205 m & 00:23:54.4901(3 ) & 72:05:30.757(1 ) & [email protected]@xmath27 & 3.6766432176010(2 ) & 3.8433(1 ) + 00247204n & 00:24:09.1879(1 ) & 72:04:28.8911(7 ) & [email protected]@xmath27 & 3.0539543462609(1 ) & 2.1857(2 ) + 00247204o & 00:24:04.65257(5 ) & 72:04:53.7667(2 ) & [email protected]@xmath27 & 2.64334329724350(3 ) & + 3.03510(5 ) + 00247204q & 00:24:16.4909(1 ) & 72:04:25.1643(6 ) & [email protected]@xmath27 & 4.03318118457303(7 ) & + 3.40077(4 ) + 00247204r & 00:24:07.5852(1 ) & 72:04:50.3952(7 ) & [email protected]@xmath27 & 3.4804627074933(2 ) & + 14.8352(3 ) + 00247204s & 00:24:03.9795(1 ) & 72:04:42.3529(4 ) & [email protected]@xmath27 & 2.8304059578794(1 ) & 12.0541(1 ) + 00247204 t & 00:24:08.5488(5 ) & 72:04:38.932(2 ) & [email protected]@xmath27 & 7.5884798073657(5 ) & + 29.3830(3 ) + 00247203u & 00:24:09.83655(7 ) & 72:03:59.6882(3 ) & [email protected]@xmath27 & 4.34282669639236(8 ) & + 9.5228(1 ) + 00247204w & 00:24:06.058(1 ) & 72:04:49.088(3 ) & [email protected]@xmath27 & 2.3523445319370(3 ) & 8.6553(1 ) + 00247201x & 00:24:22.38567(9 ) & 72:01:17.4416(7 ) & [email protected]@xmath27 & 4.77152291069354(5 ) & + 1.83610(7 ) + 00247204y & 00:24:01.4025(1 ) & 72:04:41.8365(4 ) & [email protected]@xmath27 & 2.19665714352121(5 ) & 3.51716(7 ) + 00247205z & 00:24:06.039(1 ) & 72:05:01.466(6 ) & [email protected]@xmath27 & 4.554447383946(1 ) & 0.455(1 ) + numbers in parentheses correspond to uncertainties in the last digit .
the period ( p ) and spin - down ( dp / dt ) of millisecond pulsars are very precisely measured .
the errors quoted in the parentheses are the 1-@xmath50 errors of the weighted least squares fit to the timing residuals .
[ table:47tuc ]
|
intermediate mass black holes play a critical role in understanding the evolutionary connection between stellar mass and super - massive black holes@xcite .
however , to date the existence of these species of black holes remains ambiguous and their formation process is therefore unknown@xcite .
it has been long suspected that black holes with masses @xmath0 should form and reside in dense stellar systems@xcite .
therefore , dedicated observational campaigns have targeted globular cluster for many decades searching for signatures of these elusive objects .
all candidates found in these targeted searches appear radio dim and do not have the x - ray to radio flux ratio predicted by the fundamental plane for accreting black holes@xcite .
based on the lack of an electromagnetic counterpart upper limits of @xmath1 and @xmath2 have been placed on the mass of a putative black hole in 47 tucanae ( ngc 104 ) from radio and x - ray observations respectively@xcite .
here we show there is evidence for a central black hole in 47 tuc with a mass of m@xmath3@xmath4 when the dynamical state of the globular cluster is probed with pulsars .
the existence of an intermediate mass black hole in the centre of one of the densest clusters with no detectable electromagnetic counterpart suggests that the black hole is not accreting at a sufficient rate and therefore contrary to expectations is gas starved .
this intermediate mass black hole might be a member of electromagnetically invisible population of black holes that are the elusive seeds leading to the formation of supermassive black holes in galaxies .
harvard - smithsonian center for astrophysics , 60 garden st . , cambridge , ma 02138 usa school of mathematics and physics , university of queensland , brisbane , qld 4072
, australia an intermediate mass black hole ( imbh ) strongly affects the spatial distribution of stars in globular clusters ( gcs ) .
massive stars sink into the centre more efficiently during relaxation in order to achieve energy equipartition . as a consequence , as stars sink closer to the centre they are scattered by the black hole which heats up the core .
this process quenches mass segregation ( figure [ fig : segregation ] ) . over the lifetime of a cluster , dynamical processes such as energy equipartition and two - body relaxation
therefore contribute to the outward propagation of an integrated dynamical effect beyond the black hole s radius of direct influence .
we find that this dynamical signature is efficiently propagated outward into the cluster . in relation to this effect ,
the distributions of pulsar accelerations for any projected distance from the centre show distinct features also sensitive to an imbh .
there are currently 25 millisecond pulsars detected within the cluster 47 tuc .
23 of these have phase resolved timing solutions@xcite(see extended data table 1 ) which we use to infer spatial accelerations caused by the gravitational potential of the cluster@xcite .
pulsar acceleration measurements together with @xmath5-body simulations provide stringent constraints on the mass of the central black hole in one of the most massive clusters ( m@xmath6@xmath7 ) with a compact core@xcite .
imbhs have distinct imprints on how massive stars dynamically settle three - dimensionally after the cluster has relaxed .
the upper limit of m@xmath8 placed on the black hole mass in 47 tuc in earlier kinematic studies@xcite appears inconclusive because new kinematic data@xcite and @xmath5-body models@xcite imply that no clear distinction between globular cluster models can be made based on available velocity dispersion measurements alone ( figure [ fig : velocity ] ) . in order to constrain the dynamical effects of a black hole in a cluster ,
we take a fundamentally different approach .
in addition to pulsar accelerations , we jointly use this spatial imprint that carries information about the black hole beyond the radius of influence .
we quantify how likely it is that a range of observed pulsar accelerations are related to specific model distributions .
the dynamical @xmath5-body simulations of isolated star clusters evolve under the influence of stellar evolution and two - body relaxation . a grid of several hundred star clusters starting with different initial half - mass radii , density profiles and masses of their central black hole are run up to an age of t=11.75 gyr to match the age of 47 tuc@xcite .
we select those clusters that best match the surface density and velocity dispersion profile of 47 tuc .
the presence of primordial binaries does not play a significant role in the final segregation profile of heavy stars if clusters with the same surface density profile are compared ( see extended data figure 1 and methods ) .
no other a priori assumptions are made limiting the dynamics of the cluster .
the best - fitting models for the no - imbh case and for imbhs with 0.5% and 1% of the cluster mass are selected as a subset of viable replicas of 47 tuc .
models with black hole masses larger than 1% of the cluster mass lead to fits which significantly deviate from the observed density and velocity dispersion profile of 47 tuc .
hence such massive black holes are ruled out while models with smaller black hole masses or without a black hole lead to comparable fits .
+ in each simulated replica of 47 tuc we have full spatial and kinematic information of all stars , including neutron stars .
the majority of neutron stars ( @xmath9 ) become unbound due to high kick velocities following the supernovae explosions during formation@xcite .
we extract accelerations of all simulated neutron stars with projected distances that match the observed pulsars in 47 tuc .
integrated accelerations of these matched neutron stars form distinctly different distributions ( see extended data figure 2 ) .
n - body models with an imbh produce pulsar accelerations that are overall more consistent with observations compared to models without an imbh .
the relative likelihoods for models in figure [ fig : models ] show that a central black hole is required to produce the observed pulsar accelerations and distributions in 47 tuc .
the inferred black hole mass shows a strong peaks at @xmath10@xmath4 ( figure [ fig : imbhmass ] ) .
the total cluster mass is another free parameter in our calculation that serves as a cross check . with this method
, we independently measure the total cluster mass of 47 tuc as @xmath11 ( figure [ fig : imbhcontour ] ) consistent with kinematic measurements@xcite .
+ we employ bootstrapping in order to cross - check whether the black hole detection signature comes from pulsar observations and we can rule out systematic effects . any _ statistical distance
_ approach can be employed to calculate differences for two distributions .
the kullback - leibler ( kl ) divergence ( @xmath12 ) is particularly well suited and mathematically robust for cases where the information entropy is quantified as a likelihood ( @xmath13 ) ( see section [ sec : stat ] ) .
kl divergences are calculated for randomly selected pulsars to quantify relative likelihoods for all models .
we trace how the inference for the black hole mass changes in figure [ fig : imbhmass ] .
we find that the inference flattens with decreasing number of randomly selected pulsars ( extended data figure 3 ) ; this indicates that the black hole detection signal is produced by pulsar data .
such behavior is expected only in systems where there is _ statistical learning _ , i.e. the information driving the inference comes from data , and not from random errors or outliers . for the case of 47 tuc , the dynamical information extracted from @xmath1410 pulsars appears insufficient to conclusively infer an black hole mass while the inference is sufficiently informative for @xmath15 pulsars .
stars are expected to form a density cusp in power - law form @xmath16 around the black hole because of short relaxation times in the centres of globular clusters@xcite .
however , such a sharp cusp is predicted only for idealized model clusters with single mass .
the actual shape of the cusp is difficult to predict because of the unknowns in stellar evolution such as the mass function and binary fraction .
while the projected density profile is expected to be relatively sharp , this can not be used as a direct indication of a similar cusp in luminosity because the mass - to - light ratio ( @xmath17 ) is not constant .
more realistic multi - mass cluster models find that giant stars should follow a shallower slope @xmath18 in a typical cluster in projection@xcite . a segregated neutron star and massive white dwarf population lead to a sharp rise of the dark mass making the luminosity cusp shallower .
moreover , the stochastic motion of an imbh relative to the cluster centre will enhance the flattening of a centrally rising cusp even further .
therefore , it is not surprising that an imbh in 47 tuc would produce an optically unresolved shallow cusp .
an inescapable conclusion for having an electromagnetically undetectable black hole in the cluster centre is that the core is devoid of gas within its radius of influence .
winds driven by main sequence stars@xcite , novae@xcite , and m - dwarfs@xcite contribute to clearing the gas .
additionally , the kinetic energy injected by the pulsars high energy emission into the intra - cluster environment alone may be sufficient to clear the ionized plasma from the central core@xcite .
a spherically symmetric distribution of pulsars can provide the integrated ram pressure enhancement in the centre to create a gas free cavity .
the observed signature in the dispersion measures of pulsars in 47 tuc@xcite is likely due to the bound ionized plasma remaining present beyond the radius of influence of the black hole .
[ fig : imbhmass ] [ fig : imbhcontour ] it is possible that such black holes sitting in the gas deficient central cavities of primordial globular clusters constitute a subpopulation of progenitor seeds that formed the supermassive black holes in galaxy centres . for this
, the black hole must form rapidly during the early evolutionary phases of the cluster and continue to grow exponentially after the cluster sinks to the core of a galaxy .
although massive and compact clusters are preferred environments where imbhs can be found , mounting evidence suggests that the formation and retention of such black holes are stochastic@xcite . for cases in which seed black holes are retained
, it has been shown that they can grow rapidly through accretion at super - eddington rates@xcite .
stellar winds of massive stars can provide a gas reservoir with @xmath19 yr@xmath20 to partially feed the black hole@xcite .
the gas present in the core can then be accreted as the imbh stochastically moves in the core due to the encounters .
it is likely that runaway collisions@xcite play at least a partial role in the growth of black holes as well as the mergers of compact binaries@xcite .
in fact , binaries with a black hole primaries are expected to form in the early phases of the cluster evolution and may be retained@xcite . some of these massive binaries can eventually merge and drift toward the centre feeding the black hole making these sources ideal for tidal disruption and gravitational wave detection .
the rapid growth of a black hole in a primordial globular cluster that merges with the galactic bulge as it passes close to the centre of a galaxy may continue exponentially at super eddington rates@xcite producing the supermassive black holes in quasars@xcite .
* references * 100 url # 1`#1`urlprefix[2]#2 [ 2][]#2 .
_ _ * * , ( ) . , & _ _ * * , ( ) . & .
_ _ * * , ( ) .
_ et al . _ . _
_ * * , ( ) . &
_ _ * * , ( ) . , , &
_ _ * * , ( ) .
_ et al . _ . _
_ * * , ( ) . , & .
_ _ * * , ( ) . , , & .
_ _ * * , ( ) .
_ et al . _ . _
_ * * , ( ) .
_ et al . _ . _
_ * * , ( ) .
_ et al . _ . _
_ * * , ( ) . .
_ _ * * , ( ) . .
_ _ * * , ( ) .
_ et al . _ . _
_ * * , ( ) .
, , & .
_ _ * * , ( ) . , & .
_ _ * * , ( ) . , , , & .
_ _ * * , ( ) . & .
_ _ * * , ( ) . .
_ _ * * , ( ) . & .
_ _ * * , ( ) . & .
_ _ * * , ( ) . .
_ _ * * , ( ) .
_ et al . _ . _
_ * * , ( ) . , , , & . _ _ * * , ( ) .
_ et al . _ . _
_ * * , ( ) .
, , , & .
_ _ * * , ( ) . , , , & .
_ _ * * , ( ) . & .
_ _ * * , ( ) .
_ et al . _
_ _ * * , ( ) .
this work was supported in part by the black hole initiative at harvard university , through a grant from the john templeton foundation .
b. k. initiated the project , led the collaboration , and analyzed the data .
b. k. and h. b. wrote the manuscript .
h. b. calculated the n - body models .
a. l. made contributions to the conceptual definition of the project .
all authors contributed to the interpretation of the analysis .
the authors declare that they have no competing financial interests .
correspondence and requests should be addressed to blent kiziltan ( email : [email protected] ) .
|
dwarf novae belong to a subclass of cataclysmic variable stars that consist of a white dwarf ( primary ) and a late - type star ( secondary ) .
the secondary star fills its roche lobe and transfers mass to the primary via inner lagragian point ( l1 ) and the transferred matter forms an accretion disk ( for a review , see @xcite ; @xcite ; @xcite ) . among dwarf novae ,
there exist three subtypes based on their light curves .
su uma - type dwarf novae , whose orbital period are shorter than 0.1 days in the most cases , are one of the subtypes , characteristic of exhibiting two types of outbursts .
one is normal outburst , continuing for a few days .
the other is superoutburst , lasting about two weeks , during which modulations called superhumps are shown .
the period of the superhumps are a few percent longer than that of the orbital period of the system .
this is well explained by a phase - dependent dissipation of a tidally deformed precessing accretion disk .
the most acceptable model for su uma stars is the thermal - tidal instability model developed by @xcite , well reproducing the majority of observations .
v844 her was discovered by @xcite as a variable star near @xmath0 her .
@xcite classified the variable , originally named var 43 her , as a dwarf nova based on the detection of a long outburst .
the light curve in @xcite is reminiscent of a superoutburst of su uma - type dwarf novae .
time - resolved ccd photometry was performed by @xcite during the 1999 september outburst of v844 her . detecting superhumps with a period of 0.05592(2 ) days , they firstly confirmed the su uma nature of v844 her . by radial velocity studies @xcite determined 0.054643(7 ) days ( 78.69 min ) as the orbital period of the system .
these results indicate that v844 her is one of the shortest periods among dwarf novae ever known . in order to thoroughly investigate the short period system ,
the vsnet @xcite has placed v844 her as one of the highest priorities since the confirmation of the su uma nature of the system . in 2002 may , 2002 october , and 2003 october ,
v844 her underwent an outburst and the vsnet collaboration team detected superhumps during these superoutbursts . on 2006 april 24 , pavol a. dubovsky reported a brightening of v844 her ( 12.4 mag ) to the vsnet ( [ vsnet - alert 8914 ] ) . thanks to this prompt report , as well as the seasonal condition of v844 her ( the precise coordinate is ra : @xmath1 , dec : @xmath2 , @xcite ) , we firstly succeeded in observing almost the whole superoutburst of v844 her .
the object is identical with usno a2.0 1275 - 8931436 ( @xmath3 = 16.9 , @xmath4 = 16.2 ) .
the infrared counterpart of the binary is 2mass j16250181 + 3909258 ( @xcite ; @xcite ) .
v844 her is also catalogued as a bright x - ray source by rosat , 1rxs j162501.2 + 390924 .
time - resolved ccd photometry during outburst was performed from 2002 may 20 to 2006 may 20 using 10 - 100 cm telescopes at 8 sites .
the log of these observations is summarized in table 1 .
details of observers are listed in table 2 . in total , we observed v844 her for 31 nights , during which 4 outbursts including one normal outburst were detected .
the exposure time was 10 - 40 seconds .
the read - out time was typically a few seconds .
the resultant cadence is much shorter than the time scale of variations that we focus on .
photometric data were obtained through the @xmath5 filter at the kolonica saddle site .
the other sites used no filter , which makes the effective wavelength close to that of @xmath6-system .
the total data points amounted to 14674 , which is the largest data ever obtained for v844 her .
after subtracting a dark - current image from the original ccd frames , flat fielding was performed as the usual manner .
the images obtained by mhh and ku were processed by the task apphot in iraf .
kyoto , ous , riken , and okayama data were analyzed by aperture photometry using a java - based software developed by one of the authors ( tk ) .
data of mie were analyzed using fitsphot4.1 .
the maxim dl and the c@xmath7munipack were used for data obtained at kolonica saddle . after correcting systematic differences between sites ,
the magnitude was adjusted to that of kyoto system except for 2006 observations , for which the magnitude was adjusted to that of saitama system . as comparison stars , usno a2.0 1275 - 8932542 ( ra : @xmath8 , dec : @xmath9 , @xmath5=12.836 , @xmath10=0.994 ) and usno - a2.0 1292 - 0262723 ( ra : @xmath11 , dec : @xmath12 , @xmath5=12.334 , @xmath10=0.662 )
were used for kyoto and saitama system , respectively @xcite .
, whose constancy was checked by the local stars in the same images .
heliocentric correction was applied to the observation times before the following analyses .
the light curve during the 2002 may outburst is presented in figure [ fig:0205lc ] .
the magnitude declined linearly at the rate of 0.13(1 ) mag d@xmath13 until hjd 2452420 , after which the system kept almost constant magnitude until the end of our run .
such a halt is sometimes observed in other su uma - type dwarf novae ( e.g. , v1028 cyg , @xcite . for a comprehensive review ,
see @xcite ) . on hjd 2452425 , 12 days after the detection of the outburst , v844 her likely entered the rapid decline stage , when the visual magnitude was fainter than 15 .
figure [ fig:0205mhp ] shows the representative light curves during the plateau phase after removing daily decline trends for each run .
rapid rises and slow declines , characteristic of superhumps , are visible . in order to estimate the superhump period , we applied the phase dispersion minimization ( pdm , @xcite ) method to the prewhitened light curves during the plateau stage .
we determined 0.05584(64)d as being the best estimated period of the superhump .
the error of the resulting period was estimated using the lafter@xmath7kinman class of methods as applied by @xcite .
the obtained superhump period was in good agreement with the previous studies ( @xcite ; @xcite ) .
we extracted the maximum times of superhumps calibrated mainly by eye .
tbl:02o - c ] shows the timings of the superhump maxima . a linear regression to the observed times yields the ephemeris in the following equation : @xmath14 where @xmath15 is the cycle count of the maximum timings of superhumps , and the values within the parentheses in the right side of the equation denote 1-@xmath16 error , respectively .
using the above equation , we derived the @xmath17 diagram illustrated in figure [ fig:0205o - c ] , in which the dashed line means the best fitting quadratic equation as follows : @xmath18 the quadratic term yields @xmath19 = @xmath20/@xmath21 = 4.4(1.2)@xmath2210@xmath23 , indicating that superhump period increases through the superoutburst . because of the apparent sparse data listed in table [ tbl:02o - c ] , it is likely that the obtained value might include large uncertainty
nevertheless , we can conclude the conspicuous increase in the superhump period during the plateau stage .
figure [ fig:0210lc ] shows the light curve of our run during the 2002 october outburst .
the outburst was caught on hjd 2452571 , when the visual magnitude of v844 her was about 12.5 .
our observations started one night after the detection , when the mean magnitude was about 13.3 . using the data
obtained on the first two nights , we determined the mean decline rate to be 1.20(1 ) mag d@xmath13 .
the value is large for the plateau stage of a superoutburst in su uma - type dwarf novae . on hjd 2452575 , v844
her faded to 17 mag , which is almost the same as that of the quiescent magnitude .
based on the negative observation on hjd 2452570 , we can estimate that the duration of the outburst was at most 5 days .
figure [ fig:0210e ] represents the de - trended , enlarged light curves taken on the first two nights .
interestingly , one can see hump - like profiles on 2002 october 25 ( hjd 2452573 ) with the amplitude as large as 0.4 mag .
they are , however , definitely not superhumps , since three peaks are detectable with an amplitude of @xmath24 0.4 mag during the 0.06 days run .
these results indicate this outburst was the normal outburst .
this is the first recorded @xmath25 @xmath26 of v844 her .
figure [ fig:0310lc ] displays the overall light curves during the 2003 outburst of v844 her .
the duration of the outburst appeared to be about 2 weeks .
the decline rate of 0.12(1 ) mag d@xmath13 until hjd 2452945 , which is a typical value for su uma - type dwarf novae . due to the absence of observations between hjd 2452945 and hjd 2452951
, we can not specify whether there was a phase of a constant magnitude , as was observed in the 2002 may superoutburst .
after the plateau stage , v844 her entered the rapid decline phase around hjd 2452956 , and the magnitude returned to its quiescent level on hjd 2452958 .
no rebrightenings were observed during our run . due to the lack of our observations , we were unable to trace superhump period change and whether a precursor was present .
the whole light curve of the 2006 april - may outburst is shown in figure [ fig:0604lc ] .
the duration of the plateau phase was about 2 weeks .
the magnitude declined almost constantly at the rate of 0.15(1 ) mag d@xmath13 from hjd 2453851 to hjd 2453860 , after which the magnitude kept almost constant at the end of the plateau stage . on hjd 2453867 ,
v844 her became faint with the magnitude of 16.5 .
there provided no evidence of a rebrightening during our run .
figure [ fig:0604e ] shows the enlarged light curves on the first two days of our observations .
there is no signal of superhumps on hjd 2453851 , while there were hump - like modulations on hjd 2453853 .
however , their profile suggested that superhumps had not yet fully grown .
then we suppose the superhumps were detected form hjd 2453854 .
we performed a period analysis using the 7473 points between hjd 2453854 and hjd 2453864 , after subtracting the linear declining trend .
the theta diagram of the pdm analysis provides the best estimated period of 0.055883(3)d .
this value is in good accordance with that obtained during the 2002 may superoutburst .
figuer [ fig:0604dlc ] indicates the daily averaged light curves during the plateau phase folded by 0.055883(3)d .
a rapid rise and slow decline are a typical feature of superhumps .
the data obtained on april 28(hjd 2453854 ) showed the superhumps with an amplitude of 0.2 mag , then the superhump amplitude decreased gradually .
a hint of the regrowth of the superhump can be seen on may 5(hjd 2453861 ) .
the superhump maximum timings measured by eye are listed in table 4 . a linear regression yields the following equation on the superhump maximum timings ; @xmath27 the obtained @xmath17 diagram is exhibited in figure [ fig:0604o - c ] . for @xmath71
< @xmath15 < 128 , the best fitting quadratic equation is given by : @xmath28 this equation yields @xmath19 = @xmath20/@xmath21=10.9(1.0 ) @xmath2210@xmath23 , meaning that the superhump period increases through the superoutburst .
it is well known that an accurate estimation of the distance to the dwarf novae is not easy .
nevertheless , we can roughly estimate it using an empirical relation derived by @xcite as follows : @xmath29 where @xmath30 is the absolute magnitude at the maximum during a normal outburst , and @xmath21 is the orbital period of the system in the unit of hours .
the above equation could be applied for the systems which have a low inclination and do not reach the period minimum .
based on the previous investigations by @xcite , both conditions can be satisfied for v844 her . substituting @xmath21=1.3115 into equation ( [ eqn : mv ] ) and with a little algebra ( here we assume that the maximum @xmath5 magnitude of v844 her is 12.6 )
, we roughly derived d=290(30 ) pc as an estimated distance .
using the obtained distance , we can also estimate the x - ray luminosity of v844 her following the same manner as @xcite , who showed that the @xmath31 pspc countrate in channel 52 - 201 corresponds to a flux in the 0.5 - 2.5 kev bandpass given by @xmath32 where cr@xmath33 denotes the countrate in channel 52 - 201 . with a little algebra ,
we can obtain the x - ray luminosity between 0.5 - 2.5 kev is 10@xmath34 erg s@xmath13 .
although the observed x - ray luminosity will be affected by some effects including the inclination of the system , the derived value is relatively large compared to other su uma - type dwarf novae given by @xcite .
historically , the superhump period had been known to decrease during the course of the superoutburst before the tidal instability was discovered (;
the decrease of superhump period was ascribed to shrinkage of the disk radius , or simply a natural consequence of mass depletion from the accretion disk @xcite .
recently , particularly over the past decade , the picture has been altered since numerous systems showed an increase of the superhump period .
such systems are mainly wz sge - type dwarf novae , and su uma - type dwarf novae with short orbital periods ( @xcite ; @xcite ; @xcite ; @xcite ; @xcite ; @xcite ; @xcite ; @xcite ; @xcite ) .
observationally , there appears to be a borderline around @xmath35=0.063 days below which the superhump period tends to increase @xcite .
figure [ fig : p - d ] illustrates the superhump period derivative against the mean superhump period of su uma - type dwarf novae .
the value for v844 her is pointed with the filled circles . in figure
[ fig : p - d ] , v844 her is likely to lie in the general trend .
hence , we firstly confirmed that v844 her showed the positive @xmath19 derivative and became the shortest period su uma - type dwarf nova that was confirmed to exhibit an increase of the superhump period .
additionally , we should briefly note on figure [ fig:0604o - c ] , where one can see data points departured from the quadratic equation ( [ eqn:06o - c ] ) , corresponding to @xmath15 @xmath24 @xmath720 .
recent ccd photometry indicates that this feature is observed not only for v844 her , but also for other short period su uma - type stars exhibiting the positive @xmath19 .
the systems include asas 102522 - 1542.4 ( maehara et al . in preparation ) , fl tra ( imada et al .
( 2006 ) ) , asas 160048 - 4846.2 ( @xcite ; imada et al . in preparation ) .
theoretical models suggest a dramatic variation in temperature or pressure in the accretion disk is a possible cause of the superhump period change in this stage ( @xcite ; @xcite ; @xcite ) .
we require further samples in order to discuss the nature of the superhump period change at this early stage .
it is well known that su uma - type dwarf novae show two types of superoutbursts : superoutbursts with a precursor , and superoutbursts without a precursor , though precursor - main superoutburst is hardly observed . for superoutbursts without a precursor , a handful of systems , especially archetype toads wx cet and sw uma , show three types of superoutburst : a short superoutburst with a duration as short as 10 days , an intermediate superoutburst continuing for 2 weeks , and a long superoutburst lasting longer than 20 days @xcite . in the case of v844 her ,
all the three superoutbursts reported here lasted about 2 weeks , suggesting that these superoutbursts belong to the intermediate category , or v844 her simply lies in the majority of su uma - type dwarf novae . in order to examine whether v844 her shows other types of superoutbursts , we extracted the observations reported to aavso and vsnet since the 1996 discovery .
table [ tbl : pre_outbursts ] summarizes the recorded outbursts of v844 her , from which we can properly give a constraint on the durations of outbursts .
as can be noticed in table [ tbl : pre_outbursts ] , no superoutburst provides evidence for a duration longer than 20 days , and the durations of the superoutbursts appears to converge to 2 weeks . from the archives
, we newly discovered two facts .
one is that v844 her shows no precursor .
of course we have overlooked the onset of superoutburst in a few cases , for which we can not specify the type of superoutburst .
however , the absence of a precursor has been confirmed in the most cases of the superoutbursts by virtue of the amateur observers .
the other is that no rebrightenings have been observed in v844 her despite careful monitoring of the system since the 1996 discovery @xcite .
recent extensive studies over the past decades suggest that rebrightenings tend to occur among su uma - type dwarf novae with short orbital periods ( @xcite ; @xcite ) .
the most interesting fact is , although it has been mentioned for a long time , that v844 her shows almost no normal outbursts ( @xcite ; @xcite ) . from the viewpoint of the original thermal - tidal instability model ,
normal outbursts occur more frequently as the mass transfer rate from the secondary increases ( @xcite ; @xcite ) .
further , the optical spectrum of v844 her suggests a relatively high mass transfer rate @xcite , which accelerates a circulation on the limit cycle in the @xmath36-@xmath37 diagram . from table
[ tbl : pre_outbursts ] , the supercycle of v844 her is estimated to be about 300 days , which is in agreement with the previous work @xcite .
when compared to the other systems having similar supercycles , e.g. , z cha , the peculiarity of v844 her becomes much clear with respect to the absence of normal outbursts .
although we can not draw a firm conclusion against the infrequent normal outbursts , one possibility is that evaporation in the accretion disk is working well so that a hole is created in the inner region of the disk and avoids an outburst ( @xcite ; @xcite ; @xcite ; @xcite ; @xcite ) .
the model also suggests large x - ray luminosity @xcite and expansion of the accretion disk during quiescence @xcite .
as for the former , the relatively large x - ray luminosity of v844 her may be suggestive of the evaporation .
the latter should be investigated by peak - separation studies during quiescence .
in this paper , we reported on time resolved ccd photometry during 2002 may , 2002 october , 2003 october , and 2006 april outbursts , of which three were superoutburst .
we estimated the mean superhump periods of 0.05584(64)d for 2002 may and 0.055883(3)d for 2006 april superoutbursts , respectively .
we successfully examined superhump period change during the 2002 may and 2006 april superoutburst .
the resultant period derivatives showed an period increase of superhumps during the plateau stages in both superoutbursts , which we confirmed in v844 her for the first time .
we also derived the distance to v844 her to be 290(30 ) pc . using the value ,
the x - ray luminosity between 0.5 and 2.5 kev is estimated 10@xmath34 erg s@xmath13 . to appreciate the long - term behavior of v844 her
, we investigated the archival light curves since the discovery of the variable . using the extensive data
, we estimated a possible supercycle to be @xmath24 300 days , which is in good agreement with @xcite . from the archives , it turned out that v844 her shows neither precursors nor rebrightenings , although we can not rule out the possibility that we missed a few events .
we also confirmed that v844 her shows almost no normal outbursts in spite of the intermediate supercycle among su uma - type dwarf novae .
a possible explanation on the absence of a normal outburst is that the evaporation mechanism may play a role , which is consistent with the relatively large x - ray luminosity of v844 her . in the future
, the evolution of the disk radius during quiescence should be investigated in order to test our suggestion .
we are grateful to many observers who have reported vital observations .
we acknowledge with thanks the variable star observations from the aavso and vsnet international database contributed by observers worldwide and used in this research . this work is supported by a grant - in - aid for the 21st century coe `` center for diversity and universality in physics '' from the ministry of education , culture , sports , science and technology ( mext ) .
this work is partly supported by a grant - in aid from the ministry of education , culture , sports , science and technology ( no .
17340055 , 16340057 , 17740105 ) .
part of this work is supported by a research fellowship of the japan society for the promotion of science for young scientists ( ri ) .
rccccc date & @xmath38 hjd start & @xmath38 hjd end & @xmath39exp & @xmath40 n & @xmath41 i d + 2002 may 20 & 52415.0264 & 52415.2657 & 10 & 1051 & km + 21 & 52416.2469 & 52416.2942 & 30 & 140 & kyoto + 24 & 52419.0356 & 52419.3031 & 10 & 1244 & km + & 52419.2251 & 52419.2973 & 30 & 182 & kyoto + 25 & 52420.2194 & 52420.2997 & 30 & 215 & kyoto + 27 & 52422.2133 & 52422.2934 & 30 & 195 & kyoto + 2002 october 24 & 52571.8704 & 52571.9989 & 30 & 122 & kyoto + & 52571.9074 & 52571.9792 & 10 & 290 & ous + 25 & 52572.8721 & 52573.0185 & 30 & 261 & kyoto + & 52572.9032 & 52572.9724 & 10 & 257 & ous + 27 & 52574.9060 & 52574.9557 & 40 & 89 & riken + 28 & 52575.8676 & 52575.9966 & 40 & 222 & riken + & 52575.8758 & 52575.9884 & 30 & 188 & kyoto + 29 & 52576.8715 & 52576.9917 & 30 & 122 & kyoto + & 52576.9266 & 52576.9620 & 25 - 30 & 77 & ous + 30 & 52577.8781 & 52577.9332 & 30 & 91 & kyoto + 2003 october 27 & 52939.8714 & 52939.9845 & 30 & 185 & kyoto + 29 & 52941.8702 & 52941.9478 & 30 & 148 & kyoto + 2003
november 1 & 52944.8663 & 52944.9188 & 30 & 65 & kyoto + 7 & 52950.8707 & 52950.9481 & 30 & 116 & kyoto + 12 & 52955.8781 & 52955.9428 & 30 & 93 & kyoto + 14 & 52957.8631 & 52957.8721 & 30 & 18 & kyoto + 2006 april 25 & 53851.0347 & 53851.1642 & 30 & 235 & njh + & 53851.1684 & 53851.2948 & 30 & 278 & kyoto + & 53851.2312 & 53851.3075 & 30 & 124 & mhh + 27 & 53853.0808 & 53853.3080 & 30 & 533 & kyoto + 28 & 53854.0403 & 53854.1606 & 30 & 221 & njh + & 53854.1103 & 53854.2534 & 30 & 408 & mhh + & 53854.1183 & 53854.3170 & 30 & 373 & kyoto + 29 & 53855.4640 & 53855.5378 & 30 & 107 & pd + 30 & 53856.1070 & 53856.2445 & 30 & 506 & mhh + & 53856.3505 & 53856.5867 & 30 & 343 & pd + 2006 may 1 & 53857.1459 & 53857.1463 & 30 & 200 & mhh + 2 & 53858.0236 & 53858.1938 & 30 & 307 & njh + & 53858.1229 & 53858.3097 & 30 & 383 & kyoto + 3 & 53859.0606 & 53859.1608 & 30 & 182 & njh + & 53859.0655 & 53859.2031 & 30 & 363 & mhh + & 53859.1268 & 53859.2947 & 30 & 192 & kyoto + & 53859.1568 & 53859.3309 & 30 & 439 & ku + 4 & 53860.0693 & 53860.2805 & 30 & 534 & mhh + & 53860.0842 & 53860.2142 & 30 & 237 & njh + & 53860.1026 & 53860.3091 & 30 & 498 & kyoto + & 53860.1559 & 53860.3223 & 30 & 369 & ku + 5 & 53861.0016 & 53861.2309 & 30 & 406 & njh + & 53861.0656 & 53861.3024 & 30 & 538 & kyoto + & 53861.0755 & 53861.3121 & 30 & 452 & ku + & 53861.1565 & 53861.2755 & 30 & 312 & mhh + 8 & 53864.1508 & 53864.2961 & 30 & 184 & kyoto + 11 & 53867.0282 & 53867.2747 & 30 & 469 & kyoto + 14 & 53870.0300 & 53870.1547 & 30 & 230 & kyoto + 20 & 53876.1544 & 53876.3008 & 30 & 243 & kyoto
+ + + + + cccc i d & observer & site & telescopes + km & k. morikawa & okayama , japan & 25 cm + ku & s. oizumi @xmath42 @xmath38 & kagoshima , japan & 100 cm + kyoto & a. imada @xmath42 @xmath39 & kyoto , japan & 40 cm + mhh & h. maehara & saitama , japan & 25 cm + njh & k. nakajima & mie , japan & 25 cm + ous & k. tanabe & okayama , japan & 10 cm + pd & p. a. dubovsky & kolonica saddle , slovakia & 28 cm @xmath43 30 cm + riken & k. torii & saitama , japan & 25 cm + + + ccccc e@xmath38 & hjd@xmath44 & @xmath45 & error@xmath46 & i d + 0 & 2415.0431 & 0.000700 & 0.001 & km + 1 & 2415.1019 & 0.003641 & 0.001 & km + 3 & 2415.2110 & 0.001023 & 0.001 & km + 22 & 2416.2681 & -0.003195 & 0.003 & kyoto + 72 & 2419.0634 & -0.000839 & 0.001 & km + 73 & 2419.1189 & -0.001198 & 0.002 & km + 74 & 2419.1754 & -0.000557 & 0.002 & km + 75 & 2419.2281 & -0.003716 & 0.002 & km + 129 & 2422.2523 & 0.004105 & 0.003 & kyoto + + + + + e@xmath38 & hjd@xmath44 & @xmath45 & error@xmath46 & i d + -18 & 3853.1201 & -0.002163 & 0.006 & kyoto + -17 & 3853.1723 & -0.005855 & 0.007 & kyoto + -16 & 3853.2292 & -0.004846 & 0.005 & kyoto + -15 & 3853.2893 & -0.000638 & 0.005 & kyoto + -1 & 3854.0782 & 0.005781 & 0.004 & njh + 0 & 3854.1351 & 0.006789 & 0.001 & mhh + & 3854.1354 & 0.007089 & 0.002 &
njh + & 3854.1356 & 0.007289 & 0.002 & kyoto + 1 & 3854.1904 & 0.006298 & 0.002 & kyoto + & 3854.1905 & 0.006298 & 0.002 & mhh + 2 & 3854.2495 & 0.009506 & 0.005 & kyoto + 3 & 3854.3027 & 0.006815 & 0.003 & kyoto + 36 & 3856.1368 & -0.003606 & 0.002 & mhh + 37 & 3856.1901 & -0.006098 & 0.002 & mhh + 41 & 3856.4149 & -0.004964 & 0.003 & pd + 42 & 3856.4693 & -0.006456 & 0.009 & pd + 43 & 3856.5204 & -0.011247 & 0.006 & pd + 70 & 3858.0322 & -0.008419 & 0.019 & njh + 71 & 3858.0950 & -0.001610 & 0.009 & njh + 72 & 3858.1442 & -0.008202 & 0.004 & njh + & 3858.1470 & -0.005502 & 0.004 & kyoto + 73 & 3858.2022 & -0.006193 & 0.004 & kyoto + 74 & 3858.2577 & -0.006485 & 0.005 & kyoto + 89 & 3859.0978 & -0.004858 & 0.003 & mhh + & 3859.0984 & -0.004158 & 0.005 & njh + 90 & 3859.1539 & -0.004550 & 0.003 & mhh + & 3859.1570 & -0.001550 & 0.005 & kyoto + 91 & 3859.2111 & -0.002837 & 0.003 & ku + 92 & 3859.2699 & -0.000078 & 0.002 & ku + 93 & 3859.3239 & 0.001807 & 0.005 & ku + 108 & 3860.1624 & -0.002097 & 0.008 & kyoto + & 3860.1670 & 0.003014 & 0.002 & ku + 109 & 3860.2242 & 0.003811 & 0.010 & kyoto + & 3860.2243 & 0.004429 & 0.002 & ku + 110 & 3860.2806 & 0.004319 & 0.005 & kyoto + 125 & 3861.1191 & 0.004446 & 0.007 & kyoto + & 3861.1212 & 0.007165 & 0.003 & ku + 126 & 3861.1706 & 0.000055 & 0.012 & kyoto + & 3861.1714 & 0.000755 & 0.003 & mhh + 127 & 3861.2305 & 0.004063 & 0.004 & mhh + 128 & 3861.2889 & 0.007209 & 0.003 & ku + + + + lccccc date & & duration & type & precursor + 1996 & october & 7@xmath4720 & s & @xmath48 + 1997 & may & @xmath4915 & s & @xmath48 + 1998 & december & 8@xmath4719 & s & ?
+ 1999 & september & 14@xmath4716 & s & @xmath50 + 2000 & july & 16@xmath4717 & s & @xmath48 + 2001 & august & 16@xmath4718 & s & @xmath48 + 2002 & may & 12@xmath4715 & s & @xmath48 + 2002 & october & 5@xmath476 & n & + 2002 & december & 10@xmath4718 & s & ? + 2003 & may & 3@xmath475 & n & + 2003 & october & 12@xmath4716 & s & ?
+ 2005 & january & 14@xmath4716 & s & @xmath50 + 2006 & april & 15@xmath4718 & s & @xmath48 + + + + + +
|
we report on time - resolved ccd photometry of four outbursts of a short - period su uma - type dwarf nova , v844 herculis .
we successfully determined the mean superhump periods to be 0.05584(64 ) days , and 0.055883(3 ) for the 2002 may superoutburst , and the 2006 april - may superoutburst , respectively . during the 2002 october observations
, we confirmed that the outburst is a normal outburst , which is the first recorded normal outburst in v844 her .
we also examined superhump period changes during 2002 may and 2006 april - may superoutbursts , both of which showed increasing superhump period over the course of the plateau stage . in order to examine the long - term behavior of v844 her
, we analyzed archival data over the past ten years since the discovery of this binary .
although photometry is not satisfactory in some superoutbursts , we found that v844 her showed no precursors and rebrightenings .
based on the long - term light curve , we further confirmed v844 her has shown almost no normal outbursts despite the fact that the supercycle of the system is estimated to be about 300 days . in order to explain the long - term light curves of v844 her , evaporation in the accretion disk may play a role in the avoidance of several normal outbursts , which does not contradict with the relatively large x - ray luminosity of v844 her .
|
particles interacting by means of bounded and purely repulsive interaction potentials @xmath1 can form cluster crystals at sufficiently high densities @xcite . though the existence of clustering in the full absence of attractions might seem counterintuitive at first , its existence rests on solid mathematical and physical grounds and has also been amply demonstrated by means of detailed computer simulations @xcite .
a necessary and sufficient condition for the occurrence of clustering is that the fourier transform of the interaction potential , @xmath2 , has negative parts . in this case , the properties of the system , both in the liquid and in the crystal phases , are largely determined by the position of the wavevector , @xmath3 , at which @xmath2 attains its most negative value and by the negative amplitude @xmath4 of the fourier spectrum of the potential there @xcite .
the physical realizability of such potentials as _ effective interactions _ between suitably tailored macromolecules has been demonstrated for the case of amphiphilic dendrimers as well as ring polymers @xcite . at moderate to high temperatures , the thermodynamics of the system
is very accurately described by a mean - field density functional theory , which predicts , among other properties , that the lattice constant of the crystal is density - independent @xcite .
this is brought about by the mechanism of occupying the crystal sites by a multiple number of particles @xmath5 which scales proportionally to the density of particles , @xmath6 , in the crystal . within this crystal ,
the equilibrium dynamics of the particles is characterized by two kinds of processes : the first one , operating at short time scales , is determined by lattice oscillations . at the same time , there is incessant particle hopping between sites , which brings about an overall average occupancy @xmath5 per site .
the latter process is activated , diffusive dynamics , which brings about a self - diffusion coefficient of the particles that scales as @xmath7 , where @xmath8 is the absolute temperature and @xmath9 is a coefficient that depends on the precise form of the interaction potential @xcite . whereas the diffusive , long - time dynamics of the clustered crystals is by
now well - understood due to a number of different studies , the phonon dynamics , which is the dominant one at very low temperatures , has not been analyzed in detail .
similarly , whereas the phase behavior of the system at moderate to high temperatures is very thoroughly examined , little is known regarding the phase diagram of the system at very low temperatures . the analysis of the elementary excitations ( phonons ) in the cluster crystals opens a route to low - temperature thermodynamics as well , since it allows for the calculation of the free energy of the system on the basis of a harmonic lattice theory .
the purpose of this paper is precisely to calculate and analyze the phonon spectra of cluster crystals and to draw , on this basis , conclusions on the stability , dynamics and thermodynamics of these systems at low temperatures .
we find that cluster crystals have very unusual , yet simple , phonon spectra , in which all optical modes are degenerate and the optical frequencies are independent of the wavenumber , rendering thereby these solids into perfect einstein crystals .
further , we discover a cascade of mechanical instabilities in the system , initiated at the long - wavelength limit of the low - lying acoustical modes , which trigger an avalanche of isostructural phase transitions from crystals of occupancy @xmath5 to crystals of occupancy @xmath10 , whereby @xmath5 is an integer number .
finally , we predict that the ( infinitely many ) density gaps associated with the isostructural transitions at @xmath0 gradually close up as @xmath8 grows and they terminate at critical points at low temperatures . the rest of the manuscript is organized as follows : in sec .
[ zerot : sec ] we calculate the zero - temperature phase diagram , to establish the ground states on which the phonon spectra have to be determined .
the calculation of the phonon curves follows then in sec .
[ phonon : sec ] and the ensuing phonon - based phase diagram of the system in sec .
[ phdg : sec ] . finally , in sec .
[ concl : sec ] we summarize and draw our conclusions .
to set up the stage and gather the key results for the system , we commence with the definition of the relevant quantities and a collection of hitherto known results .
we consider a system of @xmath11 particles in a volume @xmath12 , which interact via a bounded and positive interaction potential @xmath1 : @xmath13 the system is characterized by its density @xmath14 as well as the absolute temperature @xmath8 . without loss of generality , we can introduce an energy scale @xmath15 and a length scale @xmath16 that set the strength and range of the potential @xmath1 , respectively , writing the latter in the form : @xmath17 with some dimensionless function @xmath18 of a dimensionless argument @xmath19 .
accordingly , one can define the scaled density @xmath20 as well as the scaled temperature @xmath21 , where @xmath22 is boltzmann s constant .
another crucial quantity is the fourier transform @xmath23 of @xmath1 , which contains negative parts , i.e. , @xmath1 belongs to what has been termed the @xmath24 class of interactions @xcite . in particular
, we consider its most negative value , which is assumed at the wavenumber @xmath3 : @xmath25 at temperatures @xmath26 , analytical considerations and comparisons with extensive computer simulations have shown that the system forms clustered crystals with the following properties @xcite .
first , the freezing line @xmath27 from a fluid ( at low densities ) to the cluster crystal ( at high densities ) takes , approximately , the form : @xmath28 second , the clustered crystals feature an average lattice site occupancy @xmath5 that scales as @xmath29 resulting into a density - independent lattice constant @xmath30 of the crystal . finally , the long - time , diffusive dynamics of the crystals is determined by an activated - hopping mechanism , resulting into a diffusivity @xmath31 that is given by @xcite : @xmath32\exp(-\psi\rho^*/t^ * ) , \label{diffusivity : eq}\ ] ] with a potential - dependent coefficient @xmath9 . as is clear from eq .
( [ nc : eq ] ) above , the average occupancy @xmath5 of the lattice is a general , real number . at any instant , of course
, every crystal site is occupied by some integer number of particles . at finite temperatures , the activated hopping mechanism that brings about the long - time diffusivity @xmath31 of eq .
( [ diffusivity : eq ] ) , constantly redistributes particles between instantaneous ` donor ' and ` acceptor ' sites , bringing about the aforementioned average occupancy @xmath5 . at @xmath0 , however , this mechanism is absent , therefore each site has an occupancy that is frozen - in . in addition , the zero - temperature phase has to fulfill the strict requirement of _ mechanical equilibrium_. ] on each and every lattice site , i.e. , the _ force _ exercised on every particle on the lattice site , due to the rest of the crystal must vanish .
this immediately excludes the possibility of a periodic crystal with arbitrarily varying occupancies between the lattice sites : inversion symmetry of the decorated crystal structure must be guaranteed , so that mechanical equilibrium can result . and @xmath33 .
the vertices of the elementary unit cell , denoted by the parallelogram , are decorated by clusters of different occupancy , marked by the four different colors .
afterwards , the decorated cell is flipped along the sides generated by the vectors @xmath34 and @xmath33 .
the procedure is repeated for every new decorated unit cell until the whole lattice is filled.,width=453 ] the most general way of achieving this is depicted in fig .
[ decorate : fig ] .
one takes the elementary unit cell of the lattice and decorates each vertex with ( in general different ) integer occupancies of the lattice sites .
then , the cell is ` flipped ' along the axes defined by the elementary vectors and the process is repeated until the whole lattice is filled .
for the two - dimensional case shown for purposes of clarity in fig .
[ decorate : fig ] , the resulting structure can also be regarded as a crystal with a unit cell spanned by the vectors @xmath35 and @xmath36 , accompanied by a four - membered basis at the vectors @xmath37 , @xmath38 , @xmath39 , and @xmath40 .
the generalization to three dimensions is straightforward , where one obtains up to eight sublattice occupancies @xmath41 , @xmath42 and the corresponding overall occupancy of the lattice is a _ rational _ number : @xmath43 evidently , even this combinatorial freedom does not restore the general possibility of real ( i.e. , in general irrational ) @xmath5 numbers present for @xmath44 . for the concrete example of the interaction potential examined in the rest of the paper
, we checked a few possible combinations of @xmath41 s at selected densities and we found that the most stable lattice decoration at @xmath0 is the one in which _ all _ @xmath41 s coincide . on these grounds , and on the basis of parsimony and clarity , we will consider in the rest of the paper precisely only the case for which every lattice site is occupied by the same , integer number of particles .
thus , we will be discussing crystals of single , double , triple , etc .
occupancy .
though a host of results to be derived below are quite general for all potentials in the @xmath24 class , we are working with the generalized exponential model with exponent 4 , gem4 , described by the interaction potential : @xmath45 .
\label{gem4:eq}\ ] ] at @xmath0 , the stable crystal lattice for this potential is fcc , as confirmed earlier by applying a very efficient search among all bravais lattices via a genetic algorithm approach @xcite .
the ground - state phase diagram is determined by the configurations that minimize the internal energy @xmath46 ( the lattice sum ) .
in particular , one has to minimize the internal energy density @xmath47 and the latter must be a convex function of the particle density @xmath6 of the system . of gem4 fcc crystals with successively high occupancies per lattice site ,
@xmath5 , as a function of the density @xmath6 .
the occupancies are indicated in the legend .
( b ) a zoom on the lower density part of ( a ) , showing also the convex reconstruction of the curves and the resulting two - phase coexistence regions between crystals with successive occupancy numbers .
the coexistence regions correspond to the straight - line segments of the @xmath47 vs. @xmath6-curves and are also marked by the vertical broken lines .
the inscriptions in these regions denote the occupancies of the coexisting crystals.,title="fig:",width=298 ] of gem4 fcc crystals with successively high occupancies per lattice site , @xmath5 , as a function of the density @xmath6 .
the occupancies are indicated in the legend .
( b ) a zoom on the lower density part of ( a ) , showing also the convex reconstruction of the curves and the resulting two - phase coexistence regions between crystals with successive occupancy numbers .
the coexistence regions correspond to the straight - line segments of the @xmath47 vs. @xmath6-curves and are also marked by the vertical broken lines .
the inscriptions in these regions denote the occupancies of the coexisting crystals.,title="fig:",width=302 ] results are shown in fig .
[ groundstate : fig ] . in fig .
[ groundstate : fig](a ) it can be seen that with increasing density , crystals with occupancy @xmath5 become thermodynamically less favorable and they give their place to crystals with occupancy @xmath48 . at the same time
, regions of non - convexity appear around the points on which the @xmath47 vs. @xmath6-curves of the various occupancies cross .
these have to be made convex by means of the common tangent construction , as shown in fig .
[ groundstate : fig](b ) , giving rise to regions of coexistence between the crystal of occupancy @xmath5 and the crystal of occupancy @xmath48 . in
what follows , we denote with @xmath49 and @xmath50 the lower and upper density boundaries of the coexistence region between the fcc solids of occupancy @xmath5 and @xmath48 . for consistency in the notation , let us also define @xmath51 .
the phase behavior of the system at @xmath0 can be summarized as follows . for @xmath52
a single crystal of occupancy @xmath5 is stable , whereas for densities @xmath53 in the range @xmath54 , a macrophase separation between the crystals of successive occupancies @xmath5 and @xmath48 takes place . of gem4 fcc crystals on their density .
the thick horizontal lines lie in single - phase density regions , separated by phase - coexistence density gaps , denoted by the thin broken lines.,width=453 ] in fig .
[ occupancy_steps : fig ] we show the dependence of the @xmath0 site occupancy @xmath5 on density .
the characteristic step - like structure that emerges is a signature of the ground state and a precursor of the @xmath55 dependence , eq .
( [ nc : eq ] ) , encountered for higher temperatures .
indeed , the ` smoothing - out ' of the steplike form of the occupancy into a continuous , straight line is brought about by the hopping mechanism mentioned above .
the ranges of stability of the @xmath5-occupied crystals and the associated regions of coexistence for the fcc gem4-crystals at zero temperature are summarized in table [ phdg : tab ] for the first few values of @xmath5 .
site occupancy @xmath5&density range @xmath53 + 1 & 0 - 0.5985 + 1 + 2 & 0.5985 - 0.8961 + 2 & 0.8961 - 1.1100 + 2 + 3 & 1.1100 - 1.4094 + 3 & 1.4094 - 1.6182 + 3 + 4 & 1.6182 - 1.9179 + 4 & 1.9179 - 2.1267 + 4 + 5 & 2.1267 - 2.4267 + 5 & 2.4267 - 2.6336 + 5 + 6 & 2.6336 - 2.9336 + of gem4 fcc crystals on their density .
the thick solid lines have the dependence @xmath56 and they are artificially extended beyond the range of stability of single phases , within the coexistence regions , for purposes of demonstration .
the coexistence regions are denoted by thin vertical lines and the inscriptions in those denote the occupancies of the two coexisting crystals .
the lower- and upper densities of the coexistence gaps are marked with arrows .
for the overall , reconstructed dependence of @xmath30 on density , taking phase coexistences properly into account , see the text.,width=453 ] in fig .
[ lattice_constant : fig ] , we show the resulting dependence of the lattice constant @xmath30 of the fcc crystals on their density . within the regions @xmath52 , in which a pure crystal with occupancy @xmath5 is stable , the lattice constant decreases with density as @xmath56 , with some geometric prefactor @xmath57 . in the regions @xmath54
, we have the coexistence between two crystals , one with occupancy @xmath5 and lattice constant @xmath58 , and one with occupancy @xmath48 and lattice constant @xmath59 . whereas for a crystal of single occupancy the lattice constant would monotonically decrease with density as @xmath60 , here we see already at @xmath0 the propensity of the cluster crystals to suppress the variation of @xmath30 with @xmath6 .
this comes about through the growth of @xmath5 with @xmath6 in two complementary fashions : on the one hand , for the single phase regions , the relative difference between @xmath61 and @xmath62 shrinks as @xmath5 grows . on the other hand , the intervening two - phase coexistence regions have the effect that , although the lattice constant decreases with @xmath6 within a single - phase region , it gets ` kicked up ' again when the transition to the next single - phase region takes place . indeed , as can be seen in fig .
[ lattice_constant : fig ] , the range of variation of @xmath30 with density becomes increasingly narrower as @xmath6 , and thus also @xmath5 grow . once again , this is the @xmath0 precursor of the ensuing density - independence of @xmath30 on @xmath6 that has been established for higher temperatures .
the way in which the discontinuous change of @xmath5 and/or @xmath30 with density at zero temperature evolve into the smooth curves at higher temperature will be discussed in sec .
[ phdg : sec ] .
the calculation of the phonons spectra , once the ground states have been determined , follows along standard ways @xcite .
the particles are subjected to small deviations around their equilibrium lattice positions @xmath63 , denoted by the displacement field @xmath64 , and the total interaction energy is split into the lattice sum @xmath46 and a harmonic potential @xmath65 , which is a quadratic form of @xmath64 : @xmath66 where the dynamical matrix @xmath67 has the elements @xmath68 @xmath69 denoting cartesian coordinates .
determination of the phonon spectrum @xmath70 results from diagonalization of the matrix @xmath71 .
\label{dofk : eq}\ ] ] in particular , the eigenvalues of the matrix @xmath72 are equal to @xmath73 , @xmath74 being the particle mass . the phonon spectra have been calculated along the @xmath75-path in the first brillouin zone of the fcc - lattice , shown in fig .
22.13(b ) of ref .
we note , in particular , that the @xmath76 section of this path corresponds to wavevectors @xmath77 along the @xmath78 $ ] direction of the cubic crystal , whereas the last part of the path , @xmath79 , to wavevectors along the @xmath80 $ ] direction , which is also the nearest - neighbor vector in the fcc crystal .
the points @xmath81 and @xmath82 lie at the edges , whereas the point @xmath83 at the center of the brillouin zone .
.,width=453 ] the phonon spectrum of the single occupied fcc gem4-crystal at density @xmath84 is shown in fig .
[ phonons_3d_single_stable : fig ] . along the high - symmetry , @xmath76 ,
@xmath78 $ ] direction , we find , as expected , two degenerate transverse acoustical branches , as well as a separate longitudinal branch .
the degeneracy of the two former ones is lifted along the other directions in the brillouin zone . upon increasing the density of the crystal , starting at vanishingly small values
, the slope of the lowest branch in the @xmath85 direction has a nonmonotonic behavior .
it first increases and then starts decreasing again , a development that is at odds with that for usual , monatomic crystals .
the decrease of the slope results , at a density @xmath86 at the appearance of a _ negative eigenvalue _ , which corresponds to an imaginary frequency and thus signals a _
mechanical instability _ of the crystal .
we have numerically found for @xmath86 the value @xmath87 .
the cross - hatched region on the horizontal axis denotes the values of wavenumbers for which the matrix @xmath72 has negative eigenvalues.,width=453 ] since the instability first occurs at the point @xmath83 , i.e. , for @xmath88 , it signals a propensity of the single occupied crystal to undergo a long - wavelength transformation ( compression / expansion ) , fully consistent with the thermodynamic instability towards the formation of a doubly occupied crystal with a different lattice constant . referring to table [ phdg : tab ]
, we see that the point of mechanical instability lies above the point of thermodynamic instability , @xmath89 , of the single occupied crystal . in other words ,
the incipient mechanical instability of the system , at which the slightest perturbation to the crystal will result into its collapse , is _ preempted _ by a first - order transition to a doubly occupied crystal .
the scenario is reminiscent of the first - order freezing transition of the cluster fluid into the cluster crystal preempting the instability line at @xmath90 , encountered for the very same systems @xcite , the difference being that the latter occurs at finite values of the wavenumber and signals thus an ordering with finite wavelength . at any rate , the fact that the thermodynamic phase transition precedes the mechanical instability is the only consistent scenario , since the opposite would lead to an obvious contradiction . upon further increasing the density beyond the value @xmath86 , an increasingly long segment of modes along
the @xmath85 branch goes unstable , as shown in fig .
[ phonons_3d_single_unstable : fig ] .
a multiply occupied crystal is a particular case of a bravais lattice with a @xmath5-component basis , for which all basis vectors @xmath91 , @xmath92 are equal to zero .
the displacement field @xmath64 has to be augmented to a ` supervector ' @xmath93 , @xmath94 , where the superscript characterizes the particle species ( all species and their interactions being identical in our case ) .
the dynamical matrix in @xmath95 spatial dimensions now becomes @xmath96 and we obtain , accordingly , @xmath95 acoustical and @xmath97 optical branches in the phonon spectrum . for purposes of simplicity and transparency ,
we consider first , the case @xmath98 , i.e. , a one - dimensional crystal of site occupancy @xmath5 , formed by @xmath24-particles . due to the combination of the facts that all interactions between the species are identical _ and _ that all basis vectors vanish , it is straightforward to show that the @xmath99 dynamical matrix @xmath100 takes , in this case , a special form of a toeplitz matrix : all its diagonal elements are equal to @xmath101 and all non - diagonal ones to @xmath102 , namely : @xmath103 where @xmath104 \label{a1d : eq } \\
b(k ) & = & -\sum_{r}v''(r)\cos ( kr ) , \label{b1d : eq}\end{aligned}\ ] ] and the summations in eqs . ( [ a1d : eq ] ) and
( [ b1d : eq ] ) are carried over all bravais lattice vectors @xmath105 . in the appendix
, we show that matrices of the form given by eq .
( [ matrix1d : eq ] ) above have a highly degenerate eigenvalue spectrum and , in particular , one eigenvalue @xmath106 and @xmath107 degenerate eigenvalues all equal to @xmath108 . introducing the specific expressions for @xmath101 and @xmath109 from eqs .
( [ a1d : eq ] ) and ( [ b1d : eq ] ) , we obtain the phonon frequencies of the one - dimensional cluster crystal as : @xmath110 , \label{ac1d : eq } \\
\omega_i^2(k ) & = & \frac{n_c}{m}\sum_{r}v''(r),\qquad\qquad i=2,3,\ldots , n_c .
\label{opt1d : eq}\end{aligned}\ ] ] the first branch , eq .
( [ ac1d : eq ] ) , is the acoustical one , whereas the remaining , @xmath107 branches are not only all degenerate but they are also @xmath111-independent as well .
this is a remarkable result that shows that cluster crystals are physical realizations of the einstein phonon model .
in addition , we observe that the squared frequencies scale linearly with the occupancy , @xmath5 .
the dependence on the density @xmath6 of the crystal comes through the fact that the lattice vectors @xmath105 , measured in units of the interaction range @xmath16 , change with concentration .
we can thus write down the scaling relations for the acoustical and optical phonon frequencies , @xmath112 and @xmath113 respectively , their dependence on the occupation number @xmath5 and the density @xmath6 of the crystal : @xmath114 as shown in sec .
[ zerot : sec ] , the dependence of the lattice constant on density is rather weak and becomes increasingly suppressed as the density grows , as long as we stay within the limits of thermodynamic stability of the cluster crystals .
concomitantly , to an approximation that becomes more and more accurate as the density grows , we can state that having calculated the phonon spectra for some @xmath115 is sufficient to accurately predict the same for all higher values of @xmath5 .
essentially , a single phonon spectrum for some lattice constant in the middle of the stability domain of the highly occupied crystal is , to zeroth - order approximation , the same for all densities in that regime and in going from the @xmath5- to the @xmath48-occupied crystal , one can simply take the frequencies and multiply them by the factor @xmath116 to obtain the new ones .
all these results carry over to arbitrary dimensions .
the dynamical matrix of eq .
( [ matrix1d : eq ] ) takes in @xmath95-dimensions a similar form , however the entries @xmath101 and @xmath109 become themselves @xmath117 matrices @xmath118 and @xmath119 with entries : @xmath120 , \label{aalphabeta : eq } \\ b_{\alpha\beta}({\bf k } ) & = & -\sum_{\bf r } \frac{\partial^2 v({\bf r})}{\partial r_{\alpha}\partial r_{\beta } } \cos ( { \bf k}\cdot{\bf r } ) , \label{balphabeta : eq}\end{aligned}\ ] ] where @xmath69 are cartesian coordinates and the summations run over all bravais lattice vectors @xmath63 .
the eigenvalue spectrum of this matrix is , as in the one - dimensional case , highly degenerate : a number @xmath121 of them , corresponding to the optical branches , are given by the difference of the diagonal elements of the block submatrices @xmath118 and @xmath119 . for crystals of cubic symmetry , as is our case , we make use of the identity @xmath122 for each cartesian coordinate @xmath123 , to find that the frequencies of the optical branches , @xmath124 , are given by the simple expression : @xmath125 where , as in the one - dimensional case , the frequencies are @xmath77-independent and we obtain , once more , einstein crystals .
in fact , all optical modes are internal , `` breathing modes '' of the individual clusters on the lattice sites . in the acoustical oscillation modes , on the other side , all the particles within a given site move together as a composite object , i.e.
, the instantaneous displacements of all particles in a cluster are identical .
the result of eq .
( [ einstein : eq ] ) above , has been anticipated in ref .
@xcite at the level of an approximation . there ,
the limit @xmath115 has been considered , and the lattice oscillations have been modeled by deviations of a single particle from each of the lattice sites , while the remaining @xmath126 ones remain approximately immobile , due to the large mass discrepancy between the two entities . that simplified approach results into an expression almost identical to eq .
( [ einstein : eq ] ) , with @xmath5 being replaced by @xmath126 , the two quantities becoming arbitrarily similar in the region @xmath115 . in the same limit , it has been found that density functional theory predicts precisely the same dependence of the localization parameter of the one - particle density around the lattice sites .
the phonon calculation performed here shows that all these features stem from the peculiar , einstein - solid phonon spectrum of cluster crystals . in fig .
[ phonons_2d_double_stable : fig ] , we show as a concrete example the phonon spectrum of a doubly occupied gem4 fcc crystal , featuring the acoustical and the triply degenerate optical branches . .
the thick vertical line is the triply degenerate optical branch.,width=453 ] we return now to the acoustical modes , to establish the existence of a cascade of mechanical instabilities in the system . due to the inversion symmetry of the bravais lattice , the acoustical modes of the crystal have low-@xmath111 expansion that involves only even powers of the wavenumber : @xmath127 with coefficients @xmath128 that depend on the density @xmath6 as well as the propagation direction @xmath129 and the branch index @xmath130 .
is the squared speed of sound of the @xmath130-branch for propagation along the @xmath129 direction . for completeness
, we also note that an expansion of the form of eq .
( [ expand : eq ] ) holds true for the optical modes as well , with the addition of a term @xmath131 .
] stability against long - wavelength modulations means that the coefficient of the quadratic term , @xmath132 , must be positive for all directions and branches at a given density .
accordingly , a mechanical instability occurs at the point in which @xmath133 changes sign , the direction @xmath129 and the branch @xmath130 for which this first occurs , as density grows , being the most unstable ones . .
note the dependence @xmath134 of the lowest - lying acoustical branch along the @xmath135 direction .
( b ) the same at a density beyond the threshold value , @xmath136 , where a whole region , marked by the cross - hatched symbols , has become unstable.,title="fig:",width=298 ] .
note the dependence @xmath134 of the lowest - lying acoustical branch along the @xmath135 direction .
( b ) the same at a density beyond the threshold value , @xmath136 , where a whole region , marked by the cross - hatched symbols , has become unstable.,title="fig:",width=298 ] in sec .
[ acoustical : sec ] we established the existence of a threshold density @xmath137 for the single - occupied crystal , at which a mechanical instability occurs along the @xmath138 direction of the fcc gem4-crystal ; evidently , it holds @xmath139 for the lowest branch in the @xmath138-direction @xmath129 there .
this fact , combined with the scaling relation , eq .
( [ scaleac : eq ] ) , implies that there exist infinitely many mechanical instabilities , one for each occupancy @xmath5 , and occurring at the corresponding threshold densities @xmath140 given by : @xmath141 we have numerically confirmed this prediction for gem4-crystals of a few low occupancies and we show , as a representative result , in fig .
[ doubleinstability : fig ] the phonon spectra of double occupied gem4-crystals exactly at the threshold density @xmath142 , and at a slightly higher one where a number of modes feature imaginary frequencies .
-occupied gem4-crystals at @xmath0 , as indicated in the legend , are shown together with the coexistence regions ( marked gray ) and the threshold densities @xmath143 of mechanical stability marked by the arrows .
the latter are color - coded in the same way as the thermodynamic stability regions of the @xmath5-occupied crystals.,width=453 ] in fig .
[ stability_diagram : fig ] we graphically summarize the zero - temperature thermodynamic and stability phase diagrams for the gem4-model , showing the sequence of the first few occupancies for the fcc crystals .
it can be seen that the instability densities @xmath140 of the @xmath5 occupied crystals occur well beyond the limits of their thermodynamic stability . the existence of the former can be understood in terms of the particular form of the interaction potential .
suppose we restrict ourselves to single occupied crystals and we increase the density .
this compresses the system , causing the lattice constant to shrink .
since the interaction potential contains a flat part at small separations , and in view of the fact that the restoring forces in the crystal come mainly from the nearest neighbors , at some threshold value of the shrinking lattice constant the restoring force will vanish , at least along one crystallographic direction ( in our case , along the nearest - neighbor vector ) .
this signals the mechanical instability .
now , _ the same _ argument carries over to a crystal of occupancy @xmath5 at @xmath5 times the threshold density of the single occupied crystal , since the two are copies of each other as far as the lattice constant is concerned and only the forces in the latter are @xmath5 times bigger than the ones in the former .
if the restoring force in some direction of the single occupied crystal vanishes , it will also do so at the same direction for the @xmath5 occupied crystal .
we see , therefore , that the whole cascade of isostructural transitions is a straightforward consequence of the transition from the single to the double occupancy , which can be seen as the strategy of the system to avoid the mechanical collapse .
indeed , multiple occupancy allows the system to readjust its lattice constant and to arrange the particles in such a way that all restoring forces are nonvanishing , offering the solid stability against thermal fluctuations . the fact that the thermodynamic phase transitions preempt the mechanical ones is reassuring and in line with a number of other examples , in which a system undergoes a first - order transition that has an instability line deep in the region of the two - phase coexistence ( or even beyond it , as in our case ) .
the existence of the instability , however , has implications on the nucleation rates of the multiply occupied crystals , since , as one approaches it , it becomes exceedingly difficult to maintain thermodynamically metastable crystals of the wrong occupancy , a feature that should drastically suppress the corresponding nucleation barriers .
up to now our considerations were limited to @xmath0 and to mechanical equations of motion of the particles around their equilibrium positions . here
we will reintroduce finite temperatures and calculate free energies and the ensuing phase diagrams that follow on the basis of phonon modes _
the normal modes render the harmonic part of the system s hamiltonian diagonal , i.e. , a decoupling among the modes results . in a crystal with @xmath11 particles ,
occupancy per site @xmath5 and @xmath144 lattice sites , we have @xmath145 .
the original coordinates are the @xmath146 cartesian displacements of the particles from their equilibrium sites . in the normal mode description ,
these are replaced by @xmath147 branches , each one containing @xmath144 values of the wavevector @xmath77 compatible with the born - von karman boundary conditions , giving rise , consistently , to @xmath148 coordinates , as in the original description .
the helmholtz free energy @xmath149 of the harmonic solid takes , after the introduction of the normal modes , and on the basis of the decoupling among them , the simple form : @xmath150 , \label{fren : eq}\ ] ] which includes the contributions from the momentum degrees of freedom as well as the zero - point energy @xmath46 ( the lattice sum ) of the crystal . in eq .
( [ fren : eq ] ) above , the integral over @xmath111 extends within the first brillouin zone of the crystal , a fact indicated by the prime , whereas the sum runs over the @xmath147 branches of the phonon spectrum , characterized by the dispersion curves @xmath151 .
let us now focus on the limit @xmath115 . here , the @xmath152 optical branches dominate over the three acoustical ones , so we ignore the latter in what follows , and we further make the approximation @xmath153 .
since the optical frequencies @xmath124 are @xmath77-independent , we can pull the whole integrand out of the integral in eq .
( [ fren : eq ] ) above . the remaining integral @xmath154 yields the density of _ lattice sites _ , @xmath155 ,
whilst the summation over @xmath130 gives the number @xmath147 of phonon branches ; the two combine into @xmath156 , to give , within the stated approximations : @xmath157 .
\label{fren1:eq}\ ] ] apart from the manifest @xmath6-dependence of the second term on the right - hand - side of eq .
( [ fren1:eq ] ) , there is an implicit one through the optical frequency @xmath124 ; an explicit expression for the latter is given in eq .
( [ einstein : eq ] ) .
the density enters there through the dependence of @xmath5 and the lattice vectors @xmath63 on @xmath6 . as discussed in length in sec .
[ zerot : sec ] , at the limit @xmath115 considered here , the lattice constant can be taken as density independent , due to the approximate property @xmath55 .
accordingly , we define the @xmath6-independent frequency @xmath158 , for which the set of lattice vectors @xmath63 is determined by the interparticle interaction _ alone _ and , in particular , by the property that the length of the shortest reciprocal lattice vector of the crystal coincides with @xmath3 , the wavenumber at which @xmath2 has its most negative amplitude @xcite .
then , we have @xmath159 and its @xmath6-dependence comes exclusively through @xmath5 , for which we know that @xmath55 . gathering all the results together , introducing them into eq .
( [ fren1:eq ] ) , and absorbing all terms linear in @xmath6 into a separate term with some proportionality constant @xmath160 , we arrive at the result : @xmath161 where we have introduced an arbitrary length scale @xmath16 in the logarithm to render its argument dimensionless .
the last term , @xmath162 , has no effect on the determination of phase boundaries ; it acts simply as a constant shift to the chemical potential , leaving the pressure invariant , thus it will be dropped in what follows .
we arrive , thus , at a remarkably simple ( albeit approximate ) result , eq .
( [ fren2:eq ] ) , which states that the phonon free energy of the cluster crystals can be expressed as the sum of the ground state energy @xmath46 plus an ideal gas contribution , the latter corresponding to a ` fictitious temperature ' equal to three - halves the real one .
spatial dimensions this would be @xmath95-halves the real temperature . ] and @xmath10 are indicated by the inscriptions , whereas unlabeled domains correspond to single - phase regions .
the coexistence density gaps found at @xmath0 are being narrowed as @xmath8 grows , due to phonon contributions , but they never close . in the true phase diagram ,
hopping processes eliminate regions of stability of crystals with different site occupancies at sufficiently high temperatures.,width=453 ] we can now , on the basis of the free energy of eq .
( [ fren2:eq ] ) above draw a _ phononic _ phase diagram of the crystal states of the system .
the qualification pertains to the fact that in writing down eq .
( [ fren2:eq ] ) , the contributions from the phonons _ only _ have been taken into account .
both the disturbances of the phonon spectra arising from imperfect crystals with nonuniform site occupancies and the entropy associated with hopping are _ not _ included in the free energy of eq .
( [ fren2:eq ] ) .
the first therm there simply gives the @xmath0 phase diagram with the phase coexistence regions discussed in sec .
[ zerot : sec ] . the second term , seen as a function of @xmath6 , has curvature @xmath163 , thus it will act to narrow the width of the density gaps as temperature grows .
the phase diagram can be drawn by performing the double tangent construction on the @xmath164 vs. @xmath6-curves and it is shown in fig .
[ isostructural : fig ] for a region of moderate values of the density and thus site occupation numbers . as anticipated above
, the phonon contributions lead to a narrowing of the density gaps as temperature increases .
nevertheless , the crystals of different occupancy remain distinct at all temperatures , as is natural for an approach that does not allow hopping and smooth adjustments of the lattice constant .
the phonon - based phase diagram of fig .
[ isostructural : fig ] must now be juxtaposed to the ` exact ' one , arising from density functional theory @xcite and computer simulations @xcite .
the full phase diagram is free of the coexistence regions between crystals of different occupancy , for temperatures at least as high as @xmath165 .
evidently , all the contributions from hopping , inhomogeneous lattice site occupancy and anharmonicities , bring about the effect of ` washing out ' the step - like behavior of @xmath5 and thus eliminating the first - order phase transitions between crystals of different site occupancies at sufficiently high temperatures . on the other hand , at @xmath0 the density gaps and the transitions between such crystals are definitely present . whichever contribution hopping and anharmonicities have on the free energy , it can not simply eliminate the gaps present at @xmath0 for arbitrarily small temperatures .
it follows that the only possibility to reconcile the low- and high - temperature properties of the model is to put forward the conjecture that the coexistence gaps between crystals of occupancy @xmath5 and crystals of occupancy @xmath48 must narrow down as @xmath8 grows and terminate at critical points .
as there are infinitely many gaps at @xmath0 , it follows that the system must feature infinitely many , low-@xmath8 critical points at which the isostructural fcc - fcc transitions end .
preliminary simulations seem to confirm this hypothesis @xcite , whereby the critical temperature @xmath166 has been estimated in the order of magnitude @xmath167 . the isostructural fcc - to - fcc transitions and their termination at a critical point bear some similarities to the ones discovered , and extensively investigated , in the 1990s for hard - sphere systems with short - range attractive or repulsive interactions @xcite .
also in that case , two fcc - lattices with different values of the lattice constant were found to coexist at sufficiently low temperatures , the difference between the two gradually disappearing as the temperature grows .
there are , however , a number of important differences .
first , in the previous cases , the coexistence was not a ground - state feature , since it disappeared below the triple , fcc - fcc - gas temperature , to give its place to a usual , fcc - gas coexistence .
second , the two crystals had the same occupancy there , unity , whereas here they are characterized by different values of @xmath5 .
finally , the systems at hand present _ infinitely many _ such regions of coexistence and corresponding critical temperatures , as opposed to a single one in previous models .
we have investigated the ground states and the phonon spectra of cluster crystals formed by @xmath24-particles , finding a cascade of isostructural transitions and the emergence of these systems as realizations of einstein crystals .
we have put forward the hypothesis that the systems display infinitely many , low - temperature critical points in which the fcc - fcc transitions terminate . the predictions made here
hold quite generally for all @xmath24 potentials , _ provided _ there exist mutual , nonvanishing forces between the particles , so that the arguments presented in sec .
[ zerot : sec ] hold .
an important exception is the so - called penetrable sphere model ( psm ) @xcite , which is formally a gem with infinite exponent , and for which no forces act between the particles ; consequently , neither the ground state requires a vanishing force on each particle due to its neighbors nor is a harmonic expansion of the potential possible .
for the psm , the low - temperature phase diagram features _ gradual _
crossovers from low- to high - occupancy fcc - crystals instead @xcite . finally , we comment on the possible limitations of our simplifying assumption that in any given solid at @xmath0 all sites have the same , integer occupancy @xmath5 . as discussed in sec .
[ zerot : sec ] , the occupancy can be at most a rational number . therefore ,
even in full generality , isostructural transitions between rationally occupied crystals would appear the rational numbers forming a discrete set , as opposed to the general , real occupancy found by density functional theory and simulation at moderate and high temperatures .
it follows that the conclusions on the necessity to terminate the isostructural transitions at critical points remain unaffected .
the authors acknowledge useful discussions with ronald blaak , bianca mladek , daniele coslovich , lukas strauss , gerhard kahl , and daan frenkel .
consider a @xmath168 toeplitz matrix @xmath169 with the property that all its diagonal elements are equal to @xmath170 and the nondiagonal ones equal to @xmath171 , i.e. : @xmath172 replacing the first diagonal element of @xmath169 with @xmath171 we obtain the matrix @xmath173 with the form : @xmath174 the following statements hold true for the determinants @xmath175 and @xmath176 of the two matrices : @xmath177 . now
assume that they hold for @xmath178 .
developing the determinant of the matrices @xmath179 and @xmath180 around the entries of their first rows , we obtain : @xmath181 introducing into eqs .
( [ mnupone : eq ] ) and ( [ mbnupone : eq ] ) the expressions of eqs .
( [ mnu : eq ] ) and ( [ mbnu : eq ] ) , assumed valid for @xmath178 , results into the relations , now proven valid for @xmath182 : @xmath183 of the matrix @xmath169 is , due to the special form of the latter , obtained through the formal substitution @xmath184 in the expression for @xmath175 .
accordingly , the eigenvalue equation @xmath185 reads , for this matrix , as : @xmath186 = 0,\ ] ] from which the eigenvalue spectrum of the matrix @xmath169 follows as : @xmath187 the second equation manifests the degeneracy of @xmath188 eigenvalues of the matrix .
|
we analyze the ground states and the elementary collective excitations ( phonons ) of a class of systems , which form cluster crystals in the absence of attractions . whereas the regime of moderate - to - high - temperatures in the phase diagram has been analyzed in detail by means of density functional considerations ( likos c n , mladek b m , gottwald d and kahl g 2007 _ j. chem . phys .
_ * 126 * 224502 ) , the present approach focuses on the complementary regime of low temperatures .
we establish the existence of an infinite cascade of isostructural transitions between crystals with different lattice site occupancy at @xmath0 and we quantitatively demonstrate that the thermodynamic instabilities are bracketed by mechanical instabilities arising from long - wavelength acoustical phonons .
we further show that all optical modes are degenerate and flat , giving rise to perfect realizations of einstein crystals .
we calculate analytically the complete phonon spectrum for the whole class of models as well as the helmholtz free energy of the systems . on the basis of the latter ,
we demonstrate that the aforementioned isostructural phase transitions must terminate at an infinity of critical points at low temperatures , brought about by the anharmonic contributions in the hamiltonian and the hopping events in the crystals .
pacs numbers : 64.70.dv , 61.20.ja , 82.30.nr , 82.70.dd
|
in the past several decades , the covariant density functional theories have achieved great successes in exploring the finite nuclei and nuclear matter .
one of the most outstanding schemes is the relativistic mean field ( rmf ) theory with a limited number of free parameters @xcite . because of its covariant formulation of strong scalar and vector fields , the rmf theory is able to self - consistently describe the nuclear spin - orbit effect .
however , important degrees of freedom associated with the @xmath0 and tensor-@xmath1 fields are missing in the limit of hartree approach .
in fact , the dominant part of one - pion exchange process is the nuclear tensor force component @xcite that plays significant roles in nuclear structure @xcite , excitation and decay modes @xcite , and symmetry energy @xcite . as an important ingredient of nuclear force
, the tensor force , together with the spin - orbit coupling , characterizes the spin dependent feature @xcite .
it was firstly recognized by the discovery of electric quadrupole moment of the deuteron @xcite . from the viewpoint of the meson exchange picture of the nucleon - nucleon interaction @xcite , the nuclear tensor force
was thought to be originated from the exchange processes of the @xmath0 and tensor-@xmath1 fields , corresponding to the long and short range parts , respectively @xcite . in general ,
the nuclear tensor force is identified by the following form , @xmath2 where @xmath3 is a rank-2 irreducible tensor well defined in the non - relativistic quantum mechanics , with the momentum transfer @xmath4 . in recent years
, the nuclear tensor force was shown to play an essential role in determining the shell evolution from the stable to exotic nuclear systems , either by the nonrelativistic or relativistic calculations @xcite , although some suspicious remains due to the fact that the particle - vibration couplings were not included @xcite .
furthermore , the inclusion of the nuclear tensor force also brought substantial impact on understanding the nature of the nuclear excitations and decay modes @xcite .
for the density - dependent behavior of nuclear symmetry energy the key quantity to understand the nuclear equation of state ( eos ) and relevant astrophysical processes @xcite , the tensor effects have also been revealed to be among the physics responsible for the uncertainty of the symmetry energy at supranuclear densities @xcite .
although the nuclear tensor force has been well identified with the form , researchers encounter some difficulties due to the evident model dependence in determining its coupling strength based on the well - developed energy functionals such as the skyrme forces @xcite . within the covariant density functional theory founded on the meson exchange picture of the nuclear force , people attempt to investigate the tensor effects by including the lorentz tensor couplings , e.g. , in terms of @xmath5-tensor couplings @xcite . however , such so - called `` tensor '' is just pure central - type contributions in the limit of hartree approach .
the solution is to introduce explicitly the fock diagrams of the meson - nucleon couplings , so that the degrees of freedom associated with the @xmath0 and tensor-@xmath1 fields can be efficiently taken into account , for instance by the density - dependent relativistic hartree - fock ( ddrhf ) theory @xcite . within ddrhf
, substantial improvements due to the tensor effects have been revealed in reproducing the shell evolution without additional adjusted parameters @xcite .
what is more , the relativistic representation of the nuclear tensor force was proposed very recently , with the new origin associated with the fock diagrams of the isoscalar scalar @xmath6- and vector @xmath5- couplings @xcite .
it has been confirmed that the spin - dependent feature the nature of the nuclear tensor force can be extracted and quantified almost completely by the proposed relativistic formalism @xcite . in this work , we will study the effects of the nuclear tensor force components which hide in the fock diagrams of the meson - nucleon couplings , particularly the isoscalar scalar @xmath6- and vector @xmath5-couplings , on the properties of nuclear matter and neutron stars .
section [ sec : theory ] briefly introduces the relativistic formalism of the nuclear tensor forces for nuclear matter . in sec .
[ sec : result ] are presented the calculated results and discussions , including the tensor effects on the bulk properties of symmetric nuclear matter and the eos in sec .
[ sec : result - eos ] , on the density - dependent behavior of the symmetry energy in sec .
[ sec : result - esym ] , and on the neutron star structure in sec .
[ sec : result - ns ] .
finally , a summary is given in sec .
[ sec : sum ] .
relativistically , the nucleon - nucleon ( @xmath7 ) interactions can be established on the picture of the meson exchanges , including the isoscalar and isovector ones . consistent with this criterion , the lagrangian density , i.e. , the starting point of the relativistic hartree - fock ( rhf ) theory , can be constructed by enclosing the degrees of freedom of nucleon ( @xmath8 ) , two isoscalar mesons ( scalar @xmath6 and vector @xmath9 ) , two isovector ones ( pseudo - scalar @xmath10 and vector @xmath11 ) , and photon ( @xmath12 ) fields @xcite . namely the @xmath6- and @xmath5-meson fields are introduced to simulate the strong mid - range attraction and short - range repulsion , respectively , and the isovector part is evaluated by the @xmath0- and @xmath1-meson fields , and the photons take the coulomb effects into account . in general ,
the lagrangian density @xmath13 is composed of two parts , the free lagrangian @xmath14 and the one @xmath15 describing the interactions between the nucleons and mesons ( photons ) , @xmath16\psi , \label{li}\end{aligned}\ ] ] where @xmath17 , @xmath18 , and @xmath19 . in the lagrangians ( [ lfree]-[li ] ) ,
@xmath20 and @xmath21 ( @xmath22 or @xmath23 ) denote the masses ( coupling constants ) of ( between ) nucleon and mesons . in the above expressions and the following context ,
the arrows are used to denote isovector quantities , and the bold types for the vectors in coordinate space . following the standard variational procedure ,
the hamiltonian @xmath24 can be effectively derived from the lagrangian density @xmath13 as , @xmath25 where @xmath26 denotes the meson - nucleon coupling channels , namely the lorentz scalar ( @xmath6-s ) , vectors ( @xmath5-v , @xmath1-v and @xmath27-v ) , vector - tensor ( @xmath1-vt ) , tensor ( @xmath1-t ) and pseudo - vector ( @xmath0-pv ) couplings . in the hamiltonian ( [ hamitonian ] ) ,
the interacting vertices @xmath28 read as [ vertex ] @xmath29}_1 { \big[g_\omega\gamma^\mu\big]}_2 , \\ \gamma_{{\rho
\text{-v } } } ( 1,2 ) \equiv & + { \big[g_\rho\gamma_\mu{\vec}\tau \big]}_1 \cdot { \big [ g_\rho\gamma^\mu{\vec}\tau\big]}_2 , \\
\gamma_{{\rho \text{-t } } } ( 1,2 ) \equiv & + { \big[\frac{f_\rho}{2 m } \sigma_{\mu\nu}{\vec}\tau \partial^\nu\big]}_1\cdot { \big[\frac{f_\rho}{2 m } \sigma_{\mu\lambda}{\vec}\tau \partial^\lambda\big]}_2 , \label{rhot}\\ \gamma_{{\rho \text{-vt}}}(1,2 ) \equiv & + { \big[\frac{f_\rho}{2m}\sigma_{\mu\nu}{\vec}\tau\partial^\mu\big]}_1\cdot{\big[g_\rho \gamma^\nu{\vec}\tau\big]}_2{\nonumber\\}&+{\big[g_\rho \gamma^\nu{\vec}\tau\big]}_1\cdot{\big[\frac{f_\rho}{2m}\sigma_{\mu\nu}{\vec}\tau\partial^\mu\big]}_2 ,
\label{rhvt}\\ \gamma_{{\pi \text{-pv}}}(1,2 ) \equiv & -{\big[\frac{f_\pi}{m_\pi}{\vec}\tau\gamma_5\gamma_\mu\partial^\mu \big]}_1 \cdot { \big[\frac{f_\pi}{m_\pi}{\vec}\tau\gamma_5\gamma_\nu\partial^\nu \big]}_2 , \label{pi - sv}\\ \gamma _ { { a \text{-v } } } \equiv & + { \big[e\gamma_\mu\frac{1-\tau_3}2\big]}_1 { \big[e\gamma^\mu\frac{1-\tau_3}2\big]}_2,\end{aligned}\ ] ] and @xmath30 are the propagators of meson and photon fields with the following yukawa form : @xmath31 it should be noticed that in deriving the hamiltonian ( [ hamitonian ] ) we have introduced the simplifying assumption of neglecting the time component of the four - momentum carried by the mesons , which means that the meson fields are time independent .
this assumption has no consequence , in the static case , on the direct ( hartree ) terms while it amounts to neglecting the retardation effects for the exchange ( fock ) terms @xcite . to provide an accurate quantitative description of nuclear systems , one also has to treat the nuclear in - medium effects of the nucleon - nucleon interactions properly , either by introducing the non - linear self - couplings of the meson fields @xcite or the density dependence of meson - nucleon couplings @xcite . in the current framework ,
i.e. , the density - dependent relativistic hartree - fock ( ddrhf ) theory @xcite , the meson - nucleon coupling constants are assumed to be a function of baryon density @xmath32 . for the isoscalar @xmath6- and @xmath5-mesons , the density dependences of the coupling constants @xmath22 ( @xmath33 , @xmath5 ) are chosen as , @xmath34 where @xmath35 , and @xmath36 denotes the saturation density of nuclear matter .
in addition , five constraint conditions @xmath37 , @xmath38 , and @xmath39 are introduced to reduce the number of free parameters . for the ones in the isovector channels ,
i.e. , @xmath40 , @xmath41 , and @xmath42 , an exponential density dependence is utilized , @xmath43 at the mean field level , the contributions from the dirac sea are neglected , i.e. , the widely used no - sea approximation .
consequently the hf ground state can be determined as , @xmath44 where @xmath45 is the creative operator of the particle , @xmath46 is the vacuum state , and the index @xmath47 only runs over the positive energy states . with respect to the ground state @xmath48 , the rhf energy functional
can be obtained from the expectation of the hamiltonian @xmath24 as , @xmath49 where @xmath50 and @xmath51 denote the kinetic and potential energy parts , respectively , and the later contains two types of contributions : the direct ( hartree ) @xmath52 and exchange ( fock ) terms @xmath53 @xcite . in ref
@xcite , the relativistic formalism to identify the nuclear tensor force components hidden in the fock terms of the meson - nucleon couplings are proposed respectively for @xmath0-pv , @xmath6-scalar ( s ) , @xmath5-vector ( v ) and @xmath1-tensor ( t ) couplings , and they read as , @xmath54}_1\cdot{\big[\frac{f_\pi}{m_\pi } \bar\psi\gamma_0 \sigma_\nu { \vec}\tau \psi\big]}_2 d_{{\pi \text{-pv}}}^{t,\ \mu\nu}(1,2),\label{wigner - tensor - pi - r}\\ { { \mathscr h}}_{{\sigma\text{-s } } } ^t = & -{\frac{1}{4}}{\big[\frac{g_\sigma}{m_\sigma}\bar\psi \gamma_0\sigma_\mu\psi \big]}_1{\big[\frac{g_\sigma}{m_\sigma}\bar\psi \gamma_0\sigma_\nu \psi \big]}_2 d_{{\sigma\text{-s } } } ^{t,\ \mu\nu}(1,2),\label{wigner - tensor - sig - r}\\ { { \mathscr h}}_{{\omega\text{-v } } } ^t = & + { \frac{1}{4}}{\big[\frac{g_\omega}{m_\omega}\bar\psi \gamma_\lambda\gamma_0\sigma_\mu\psi \big]}_1 { \big[\frac{g_\omega}{m_\omega}\bar\psi \gamma_\delta \gamma_0\sigma_\nu \psi \big]}_2 d_{{\omega\text{-v } } } ^{t,\ \mu\nu\lambda\delta}(1,2),\label{wigner - tensor - ome - r}\\ { { \mathscr h}}_{{\rho \text{-t } } } ^t = & + { \frac{1}{2 } } { \big[\frac{f_\rho}{2m}\psi \sigma_{\lambda\mu}{\vec}\tau\psi \big]}_1\cdot { \big[\frac{f_\rho}{2m}\psi \sigma_{\delta\nu}{\vec}\tau\psi \big]}_2 d_{{\rho \text{-t } } } ^{t,\ \mu\nu\lambda\delta } ( 1,2),\label{wigner - tensor - rhot - r}\end{aligned}\ ] ] where @xmath55 , and the propagator terms @xmath56 read as , @xmath57 } d_\phi ( 1,2){\nonumber\\}&\hspace{4.8em } + { \frac{1}{3 } } g^{\mu\nu } \delta(x_1-x_2 ) , \label{propagator - s}\\ d_{\phi'}^{t,\ \mu\nu\lambda\delta}(1,2 ) = & \partial^\mu(1)\partial^\nu(2 ) g^{\lambda\delta } d_{\phi'}(1,2 ) { \nonumber\\}&- { \frac{1}{3 } } { \big(g^{\mu\nu } g^{\lambda\delta } - { \frac{1}{3 } } g^{\mu\lambda } g^{\nu\delta } \big ) } m_{\phi'}^2 d_{\phi'}(1,2){\nonumber\\}&+ { \frac{1}{3 } } { \big(g^{\mu\nu } g^{\lambda\delta } - { \frac{1}{3 } } g^{\mu\lambda } g^{\nu\delta } \big ) } \delta(x_1-x_2).\label{propagator - v}\end{aligned}\ ] ] in above expressions ( [ propagator - s]-[propagator - v ] ) , @xmath26 stands for the @xmath6-s and @xmath0-pv couplings , and @xmath58 represents the @xmath5-v and @xmath1-t channels . for the @xmath1-v coupling , corresponding formalism @xmath59 can be obtained simply by replacing @xmath60 ( @xmath61 ) in eqs .
( [ wigner - tensor - ome - r ] ) and ( [ propagator - v ] ) by @xmath62 ( @xmath40 ) and inserting the isospin operator @xmath63 in the interacting index . in consistence with the theory itself , the @xmath64 components of the propagator terms will be omitted in practice , which amounts to neglecting the retardation effects . transferring to the momentum space , the interaction index together with the propagator term in @xmath65 ( @xmath66 and @xmath67 )
can be expressed as , @xmath68 and the numerator term in the right - hand side is exactly a rank-2 irreducible tensor operator similar as @xmath3 [ see eq .
( [ tensor_wigner ] ) ] . for @xmath69 , @xmath70 and @xmath71 , one
may obtain the irreducible tensor operators with higher ranks .
the @xmath72 term in the numerator of eq .
( [ vphit ] ) , together with the denominator @xmath73 , contributes two types of the interactions , @xmath74 which are respectively the delta and @xmath75 terms in the propagator term ( [ propagator - s ] ) if transferring back to the coordinate space .
for the uniform nuclear matter , relevant contributions to the energy density functional ( edf ) from the nuclear tensor force components , namely the expectations of the proposed hamiltonians ( [ wigner - tensor - pi - r]-[wigner - tensor - rhot - r ] ) , can be derived as , @xmath76 } , \label{eq : tsig}\\ e_\omega^t = & + { \frac{1}{(2\pi)^4 } } \frac{g_\omega^2}{m_\omega^2 } \sum_{\tau_1 , \tau_2 } \delta_{\tau_1 , \tau_2 } \int p_1dp_1 p_2dp_2 { \nonumber\\}&\times \big\{{\big [ { \big(p_1 ^ 2 + p_2 ^ 2 + { \frac{1}{6 } } m_\omega^2\big)}\phi_\omega - p_1p_2\theta_\omega \big ] } \hat p_1
\hat p_2 { \nonumber\\ } & \hspace{2 em } + { \big({\frac{1}{4 } } m_\omega^2\theta_\omega - p_1p_2\big ) } { \big(\hat m_1\hat m_2 -1\big ) } \big\ } , \label{eq : tome}\\ e_\pi^t = & + { \frac{1}{(2\pi)^4 } } \frac{f_\pi^2}{m_\pi^2 } \sum_{\tau_1,\tau_2}{\big(2- \delta_{\tau_1 , \tau_2 } \big ) } \int p_1dp_1 p_2dp_2 { \nonumber\\}&\times \hat p_1\hat p_2{\big[{\big(p_1 ^ 2 + p_2 ^ 2-{\frac{1}{3}}m_\pi^2\big)}\phi_\pi - p_1p_2\theta_\pi \big ] } , \label{eq : tpio}\\ e_{{{\rho \text{-t } } } } ^t = & + { \frac{1}{2 } } { \frac{1}{(2\pi)^4 } } \frac{f_\rho^2}{m^2 } \sum_{\tau_1,\tau_2}{\big(2- \delta_{\tau_1 , \tau_2 } \big ) } \int p_1dp_1 p_2dp_2 { \nonumber\\}&\times \hat p_1\hat p_2{\big[{\big(p_1 ^ 2 + p_2 ^ 2-{\frac{1}{3}}m_\pi^2\big)}\phi_\pi - p_1p_2\theta_\pi\big ] } , \label{eq : trtn}\end{aligned}\ ] ] where @xmath77 and @xmath78 denote the third components of the isospin of nucleons . for the quantities @xmath79 , @xmath80 , and the hatted ones @xmath81 and @xmath82 , they read as , @xmath83 with @xmath84 , @xmath85 and @xmath86 @xcite , where @xmath87 is the scalar self - energy , and @xmath88 and @xmath89 are the time and space components of the vector one , respectively . for @xmath1-v coupling
, its expression can be obtained by replacing @xmath60(@xmath61 ) and isospin factor @xmath90 in eq .
( [ eq : tome ] ) with @xmath62(@xmath40 ) and @xmath91 , respectively . with the edfs ( [ eq : tsig]-[eq : trtn ] ) of the nuclear tensor forces , the corresponding contributions to the self - energies
can also be obtained .
notice that the extraction of the tensor force contributions does not introduce any additional free parameters , which is exactly the advantage of the method to treat the tensor effects self - consistently .
since a nuclear tensor force emerges simultaneously with the presence of fock diagrams of meson - nucleon couplings , it is worthwhile to study its effects with the proposed relativistic representation [ see eqs .
( [ wigner - tensor - pi - r]-[wigner - tensor - rhot - r ] ) ] . in this study
, we focus on the role played by the naturally involved tensor force components in the saturation mechanism , the eos and the symmetry energy of nuclear matter , and the bulk properties of neutron star , using the ddrhf functionals pka1 @xcite , pko1 @xcite , pko2 and pko3 @xcite . among these functionals , pka1 has the complete rhf scheme of meson - nucleon couplings as list in eq .
( [ vertex ] ) , whereas in pko series the @xmath70 and @xmath92 couplings are missing , and the @xmath67 one is not included in pko2 , either . in order to reveal the self - consistent tensor effects in describing the nuclear matter and neutron star properties , we performed the comparison between two self - consistent calculations : one with the full edf and the other with an edf that drops the tensor force components . with these two self - consistent procedures , the tensor force contributions to the edf can be completely included or excluded , respectively .
8.5cm@cccc & @xmath36(w / o ) & @xmath93(w / o ) &
@xmath94(w / o ) + pka1 & 0.160 ( 0.148 ) & @xmath9515.83 ( @xmath9514.18 ) & 229.96 ( 203.56 ) + pko1 & 0.152 ( 0.140 ) & @xmath9516.00 ( @xmath9514.21 ) & 250.24 ( 221.96 ) + pko2 & 0.151 ( 0.139 ) & @xmath9516.03 ( @xmath9514.31 ) & 249.60 ( 222.65 ) + pko3 &
0.153 ( 0.140 ) & @xmath9516.04 ( @xmath9514.22 ) & 262.47 ( 229.82 ) + table [ tab : sat - bulk ] shows the bulk properties of symmetric nuclear matter at saturation point , namely the saturation density @xmath36 ( @xmath96 ) , the binding energy per nucleon @xmath93 ( mev ) and the incompressibility @xmath94 ( mev ) . to reveal the tensor effects in determining the saturation mechanism , the values in the brackets are the results extracted from the calculations which drop the tensor contributions . with the full ddrhf functionals which have the nuclear tensor force components involved in the fock diagrams automatically ,
the saturation points have been well established as the saturation density @xmath36 @xmath97 @xmath96 and the binding energy @xmath93 @xmath98 mev , and both are in a good agreement with the empirical values .
as generally expected , the nuclear tensor force presents tiny contributions to the energy functional indeed . while the saturation mechanism is disturbed essentially , if removing the tensor force contributions from the ddrhf functionals .
as shown in table [ tab : sat - bulk ] , the saturation densities become 0.012 @xmath96 smaller due to the dropping of the nuclear tensor force components and such reduction almost accounts for 8% of the saturation density .
consistently the changes of 1.7 mev are found on the binding energy per nucleon @xmath93 .
for the incompressibility @xmath94 that has wide ranges in theoretical predictions @xcite , the tensor effects enhance distinctly the @xmath94 values by 26 @xmath99 33 mev .
it should be noticed that the candidates of nuclear tensor - related observables , such as the nuclear spin - isospin resonances @xcite and single - particle shell evolution @xcite , were not utilized in parameterizing the ddrhf functionals pka1 @xcite and pko series @xcite .
even though , as seen from table [ tab : sat - bulk ] , the nuclear saturation mechanism is influenced fairly distinctly by the natural tensor force components in the ddrhf functionals . from eqs .
( [ eq : tsig][eq : trtn ] ) , these tensor contributions to the energy functional depend on the momentum @xmath100 , and gradual enhancements on the tensor edfs are therefore predictable at high - density region , as well demonstrated in fig .
[ fig : eos ] . the black lines in the fig .
[ fig : eos ] are the eoss of symmetric nuclear matter calculated with the full ddrhf functionals pka1 , pko1 and pko3 , and the red lines correspond to the relevant calculations which drop the tensor force components . comparing the calculations with the full ddrhf functional and those dropping the tensor terms
, it seems that in the low - density region ( @xmath101 ) , the nuclear tensor force does not change much the eoss .
if concentrating on the density region @xmath102 with smaller scale , the deviations between two types of the calculations are still remarkable as seen from the subset in fig .
[ fig : eos ] .
qualitatively it can be easily justified that the presence of the tensor terms in the full ddrhf functionals increase the curvatures of the eoss at @xmath103 , i.e. , the values of the incompressibility @xmath94 are enhanced by the tensor effects . in the supranuclear density region ,
the nuclear tensor force presents much more distinct effects , which contributes about 40 mev to the energy functional and makes the eoss softer . the symmetry energy and its density - dependent behavior play a crucial role in understanding the properties of neutron - rich nuclei , isospin asymmetric nuclear matter , and neutron stars . despite much efforts
were devoted by the experimental and theoretical researchers , the density behavior of symmetry energy at supranuclear density region is still not well constrained
. theoretically very different high - density behaviors of symmetry energy were predicted by various models , varying from extreme soft to very stiff ones @xcite .
recently , some studies were performed to reveal the tensor effects on the density dependence of the symmetry energy @xcite .
8.5cm@cccc & @xmath104(w / o ) & @xmath105(w / o ) & @xmath106(w / o ) + pka1 & 36.02 ( 35.95 ) & 103.50 ( 115.49 ) & 212.90 ( 317.31 ) + pko1 & 34.37 ( 33.50 ) & 97.70 ( 101.66 ) & 105.85 ( 158.87 ) + pko2 & 32.49 ( 31.73 ) & 75.93 ( 81.12 ) & 77.51 ( 128.77 ) + pko3 & 32.98 ( 32.26 ) & 83.00 ( 88.91 ) & 116.43 ( 176.39 ) + table [ tab : sat - esym ] shows the symmetry energy @xmath104 with its slope @xmath105 and curvature @xmath107 , and those extracted from the calculations dropping the tensor contributions are given in the brackets for comparison .
for the symmetry energy @xmath104 at saturation density , the subtractions of the tensor contributions bring very tiny changes whereas both the slope @xmath105 and curvature @xmath107 increase fairly distinctly .
similar as the results shown in table [ tab : sat - bulk ] , the tensor effects on the symmetry energy @xmath104 with its slope @xmath105 and curvature @xmath107 are not so notable . such results are closely connected with the nature of relativistic edfs of the tensor force components ( [ eq : tsig ] [ eq : trtn ] ) which essentially depend on the momentum carried by the nucleons . in the low - density region
associated with the low momentum , the nuclear tensor force shows little impact on the nuclear matter properties , and with the density increasing that is equivalent to increasing the momentum @xmath100 the tensor effects may become remarkable . in fig .
[ fig : esym ] the symmetry energies calculated with the ddrhf functionals pka1 , pko1 and pko3 are shown as a function of baryon density @xmath32 .
to reveal the tensor effects , figure [ fig : esym](a ) presents the comparison between the calculations with the full functionals and those dropping the tensor terms , and using the ddrhf functional pka1 , figure [ fig : esym](b ) shows the tensor contributions to the symmetry energy .
consistent with the results in table [ tab : sat - esym ] , it is found from fig .
[ fig : esym ] that the withdraw of the tensor force contributions does not bring distinct changes on the symmetry energy at subsaturation density region .
however , with the density increasing , the tensor effects on the symmetry energy become notable due to the fact that the relativistic edfs [ see eqs .
( [ eq : tsig]-[eq : trtn ] ) ] of the tensor forces depend on the momentum @xmath100 essentially .
compared to the calculations which drop the tensor terms [ red lines in fig .
[ fig : esym ] ( a ) ] , the symmetry energies at supranuclear density region are fairly softened by the tensor effects .
this is well demonstrated by the tensor contributions to the symmetry energy in fig .
[ fig : esym ] ( b ) which are negative and counteract about 20% of the contributions from the other channels at high density .
it is worthwhile to mention that the tensor force components are naturally introduced with the presence of the fock diagrams in the ddrhf functionals .
hence the current results provide a self - consistent explanation for the tensor effects on the density dependence of the symmetry energy .
furthermore with the relativistic edfs ( [ eq : tsig ] [ eq : trtn ] ) , the tensor contributions to the symmetry energy from different channels can be extracted , namely the @xmath6-s , @xmath5-v , @xmath1-v , @xmath1-t and @xmath0-pv couplings as shown in fig .
[ fig : o3-esym - cctm ] ( a ) . using the ddrhf functional pka1 ,
it is clear that the tensor component in the @xmath108 coupling channel dominates the tensor contributions to the symmetry energy , followed by the @xmath109-couplings , while those in @xmath0- and @xmath1- exchanges are close to zero .
this result can be well understood from the tensor coupling constants shown in fig .
[ fig : o3-esym - cctm](b ) , namely @xmath110 , @xmath111 , @xmath112 , @xmath113 and @xmath114 in the relativistic formalism ( [ wigner - tensor - pi - r]-[wigner - tensor - rhot - r ] ) .
it is seen that the tensor coupling constants in @xmath108 and @xmath109 channels tend to certain values at high density , whereas due to the exponential density - dependent behavior of @xmath40 , @xmath41 and @xmath42 , those from the isovector @xmath1-v , @xmath1-t and @xmath0-pv channels vanish at the supranuclear density region where the tensor effects become notable . in understanding the cooling mechanism of neutron stars , the proton fraction @xmath115 is a key quantity which carries significant information of the eos of asymmetric nuclear matter .
by emitting thermal neutrinos through the direct urca ( du ) processes @xmath116 and @xmath117 , the stars would cool rapidly .
if the proton fraction goes beyond a threshold value @xmath118 , the du process works .
following the triangle inequality for momentum conservation and charge neutrality condition @xcite , it is easy to obtain the threshold of the proton fraction @xmath118 as @xmath119 . within the density range of static and @xmath120-equilibrium neutron star matter
, the proton fractions @xmath121 are shown as functions of baryon density @xmath32 in fig .
[ fig : o3-fractp ] and the results are extracted from the calculations with the ddrhf functionals pka1 , pko1 and pko3 , as compared to those dropping the tensor force components .
it is seen from fig .
[ fig : o3-fractp ] that the density - dependent behaviors of the proton fraction @xmath121 are also softened with the presence of the nuclear tensor force components in the ddrhf functionals , consistent with the systematics of the symmetry energy in fig .
[ fig : esym](a ) . for a given @xmath122
, it corresponds to a threshold density @xmath123 of the du process occurring that relies on the symmetry energy .
once when the central density @xmath124 of a neutron star exceeds the threshold density @xmath123 , the star will cool rapidly via the du processes .
one can see that the threshold density @xmath123 , determined by the ddrhf calculations with the full edfs which contain the tensor force components in the fock diagrams , are higher than those dropping the tensor terms .
such result indicates that the nuclear tensor force is unfavourable for the occurrence of the du process .
considering the well - known fact that the occurrence of the du process is not supported by the modern observational soft x - ray data in the cooling curve , it seems that the predictions with nuclear tensor force are in better agreement with the observations .
7.5cm@c|rr|rr|rr & & & + & & & & & & + pka1 & 1.40 & 1.40 & 14.06 & 14.84 & 0.31 & 0.27 + pko1 & 1.40 & 1.40 & 14.17 & 14.84 & 0.31 & 0.27 + pko2 & 1.40 & 1.40 & 13.79 & 14.44 & 0.32 & 0.28 + pko3 & 1.40 & 1.40 & 13.96 & 14.64 & 0.31 & 0.27 + & & & + & & & & & & + pka1 & 2.42 & 2.54 & 12.35 & 12.93 & 0.81 & 0.74 + pko1 & 2.45 & 2.56 & 12.42 & 12.94 & 0.80 & 0.74 + pko2 & 2.46 & 2.56 & 12.30 & 12.82 & 0.81 & 0.74 + pko3 & 2.50 & 2.61 & 12.49 & 13.01 & 0.78 & 0.72 + fig .
[ fig : tov ] shows the mass - radius relation of neutron stars calculated with the ddrhf functionals pka1 , pko1 and pko3 , and those dropping the tensor terms ( in red lines ) are also shown for comparison .
it is found that the curves of mass - radius relation of neutron star are collectively shifted rightward about 0.8 km with the dropping of the tensor terms . in general , larger neutron star radius corresponds to a stiffer density - dependent symmetry energy .
combined with the results in fig .
[ fig : esym ] ( a ) , it can be concluded that the tensor effects on the mass - radius relation of neutron star and on the density dependence of symmetry energy are congruous with each other . due to the fact that the tensor edfs depend on the momentum @xmath100
essentially [ see eqs .
( [ eq : tsig]-[eq : trtn ] ) ] , fairly distinct tensor effects are therefore observed on the mass - radius relation of neutron star .
table [ tab : mr ] lists the radii @xmath125 ( km ) and central densities @xmath124 ( @xmath96 ) of the canonical neutron stars with 1.4@xmath126 ( upper panel ) and the ones with the maximum mass limits @xmath127 ( lower panel ) .
the results are calculated with the ddrhf functionals ( w / t ) and those dropping the tensor terms ( w / o ) . for the canonical neutron stars ,
the radii @xmath125 are reduced about @xmath128 km and the central densities @xmath124 become larger , as compared to the calculations dropping the tensor terms .
that is , the presence of the nuclear tensor force leads a neutron star to be more compact . from the lower panel of table [ tab : mr ] , one can also find similar systematical changes due to the nuclear tensor force .
with the relativistic representation of the nuclear tensor forces that originate from the fock diagrams of the meson - nucleon coupling , we studied the self - consistent tensor effects on the saturation mechanism , the equation of state , the density - dependent behavior of the symmetry energy and the neutron star properties . within the density - dependent relativistic hartree - fock ( ddrhf ) theory , two types of the calculations were performed to reveal the tensor effects , i.e. , the ones with the full ddrhf functional and those dropping the tensor terms .
it is found that if removing the tensor force components in the ddrhf functionals the saturation mechanism of nuclear matter is notably influenced . due to the fact that the tensor edfs depend on the momentum essentially , the tensor effects become more and more distinct with the density increasing . due to the naturally involved tensor force components in the fock diagrams , the density - dependent behavior of the symmetry energy
is fairly softened and consequently it leads neutron stars to be more compact . moreover , for the direct urca ( du ) process that cools the neutron star rapidly , the threshold density is raised by the nuclear tensor force .
finally we would like to emphasize that different from other nuclear functionals such as the skyrme+tensor methods , the nuclear tensor force is included automatically with the presence of the fock diagrams in ddrhf , and therefore the current scheme paves a self - consistent way to explore the tensor effects on the nuclear matter system .
this work is partly supported by the national natural science foundation of china under grant nos .
11375076 and 11405223 , the specialized research fund for the doctoral program of higher education under grant no .
20130211110005 , and the youth innovation promotion association of chinese academy of sciences .
|
with the relativistic representation of the nuclear tensor force that is included automatically by the fock diagrams , we explored the self - consistent tensor effects on the properties of nuclear matter system .
the analysis were performed within the density - dependent relativistic hartree - fock ( ddrhf ) theory .
the tensor force is found to notably influence the saturation mechanism , the equation of state and the symmetry energy of nuclear matter , as well as the neutron star properties . without introducing any additional free parameters , the ddrhf approach paves a natural way to reveal the tensor effects on the nuclear matter system .
|
the problem of computing local marginals of an arbitrary probability measure is computationally hard but essential , for example , in the study of inverse problems and in solving for solutions to random constraint satisfaction problems .
the loopy belief propagation ( bp ) algorithm is an efficient approximate algorithm that has proven very helpful in the study of random optimization problems @xcite .
the bp algorithm , relying on the bethe approximation , is exact for systems living on tree interaction graphs .
it is also expected to be asymptotically exact for locally tree - like graphs as long as the variables are not strongly correlated . in general , the accuracy and convergence of the loopy bp algorithm are not guaranteed , especially in the presence of short loopy interactions . therefore , characterizing the algorithm performance in the presence of loopy interactions @xcite , and its generalizations @xcite , have been the subject of many studies in recent years . in fact , the loopy bp marginals are not globally consistent in the presence of loopy interactions .
this means that the algorithm performance can be improved by demanding more consistency for the bp marginals , e.g. , by ensuring that the local marginals satisfy the fluctuation - response relations @xcite .
there are also many efforts to take into account more accurately the effect of loopy interactions by an expansion around the bp solution in a loopy graph @xcite . in this study
, we consider the ising model on a loopy interaction graph , e.g. , the two - dimensional square lattice . given a spanning tree of the interaction graph
, we can unambiguously identify the set of loopy interactions , the number of links between the endpoints of a loopy interaction on the spanning tree ( length of loop ) , and the minimum number of links between the endpoints of two loopy interactions ( distance of two loops ) ( see fig . [ f1 ] ) .
we write a high - temperature expansion for the loopy interactions that is given by a sum over all the possible frustration - free loop configurations .
a subset of loopy interactions is called frustration - free if there exists at least one spin configuration that satisfies all the interactions in that subset .
we consider the loop expansion as a partition function for a system of globally interacting loop variables .
the nonlocal loop interactions can be expressed in terms of the correlation functions of the spin variables in the tree interaction graph .
we then resort to approximations of different complexity to study such a system of globally interacting loop variables .
see also ref .
@xcite , where a high - temperature expansion of loopy interactions was proposed , and the low orders of loop corrections to the gibbs free energy were computed . and @xmath0 of lengths @xmath1 and @xmath2 , respectively .
the two loops have distance @xmath3 defined as the minimum separation of their endpoints along the tree.,width=151 ] the following arguments rely heavily on the clustering property of the pure gibbs state defined by the probability measure of the spin variables on the tree interaction graph : the fact that spin correlations decay exponentially with distance along the tree .
therefore , as long as the loops are well separated , we can approximate the system of loopy interactions by an effective system of independent loops .
this independent - loop ( il ) approximation works reasonably well in the high - temperature phase down to the critical region of the ferromagnetic ising model in two spatial dimensions . here
, the approximation performance is determined by the length and distance distribution of the loopy interactions ; in one hand , short loops have more contribution in the loop expansion than the longer ones , and in the other hand , the more distant loops are better described within the il approximation . to go beyond the il approximation
, we make use of the bethe approximation to write the global loop interactions in terms of local correlation functions involving the neighboring loopy interactions .
these local correlations in the tree graph can be computed exactly , and then the loop corrections can be computed by the standard belief propagation ( bp ) algorithm . in this way , and in the low orders of the approximation , we obtain results that are comparable to those of the loopy bp , with algorithms that exhibit better convergence properties than the loopy bp algorithm . this is because these approximations rely on the structure of the loopy interactions instead of the whole interaction graph .
indeed for the ising model with random couplings on the two - dimensional square lattice , already the il approximation works better than the loopy bp in estimating the spin correlations on loopy interactions .
the paper is organized as follows .
section [ s1 ] gives the basic definitions and equations we use in this paper . in sec .
[ s2 ] , we write the main equations for the high - temperature loop expansions . in sec .
[ s3 ] , we present approximate algorithms based on the loop expansions , starting from the independent - loop approximation .
finally , sec .
[ s4 ] gives the concluding remarks .
consider the ising model on an arbitrary graph @xmath4 with hamiltonian : @xmath5=-\sum_{i=1}^n h_i \sigma_i-\sum_{(ij ) \in g } j_{ij } \sigma_i\sigma_j,\end{aligned}\ ] ] where @xmath6 with @xmath7 .
the probability of finding the system in spin configuration @xmath8 at inverse temperature @xmath9 is given by the gibbs probability measure , @xmath10}.\end{aligned}\ ] ] the partition function @xmath11 and the free energy @xmath12 are given by @xmath13}=e^{-\beta f}.\end{aligned}\ ] ] we choose a spanning tree @xmath14 of the interaction graph @xmath4 .
this defines the set of loopy interactions @xmath15 .
the length @xmath16 of a loopy interaction @xmath17 is defined by the number of links connecting the endpoints @xmath18 and @xmath19 on the spanning tree .
the distance @xmath20 of two loopy interactions is defined by the minimum number of links connecting the endpoints @xmath21 to @xmath22 on the spanning tree , i.e. , @xmath23 .
figure [ f2 ] shows the length and distance distributions of loops in two- and three - dimensional cubic lattices .
we use @xmath24 and @xmath25 for the neighborhood set of @xmath18 in @xmath26 and @xmath27 , respectively . of the loopy interactions in the two - dimensional ( @xmath28 ) and three - dimensional ( @xmath29 ) square lattices with respect to a random spanning tree .
here @xmath30 is the linear size of the lattices with periodic boundary conditions.,width=604 ] let us rewrite the hamiltonian as @xmath31=h_0[\boldsymbol\sigma]+h_{\epsilon}[\boldsymbol\sigma]$ ] , separating the energy contribution of interactions on the spanning tree and the loopy interactions , @xmath32 & = -\sum_{i=1}^n h_i \sigma_i-\sum_{(ij ) \in \mathsf{t } } j_{ij } \sigma_i\sigma_j,\\ h_{\epsilon}[\boldsymbol\sigma ] & = -\sum_{(ij ) \in \mathcal{l } } j_{ij } \sigma_i\sigma_j,\end{aligned}\ ] ] and define @xmath33}=e^{-\beta f_0}.\end{aligned}\ ] ] for the model on the spanning tree @xmath14 , we can exactly compute the local marginals by the bethe equations for the cavity marginals @xmath34 . here
, @xmath34 is the probability of finding spin @xmath18 in state @xmath35 , in the absence of interaction with spin @xmath19 , that is , @xmath36 these are the belief propagation ( bp ) equations @xcite . to shorten the notation
, we defined @xmath37 and @xmath38 . there is only one solution to the above equations which can be obtained by iteration starting from arbitrary cavity marginals .
then , the local one- and two - spin marginals are given by @xmath39 at the end , the free energy is obtained by @xmath40 , where the local free energy changes read as @xmath41 the reader can refer to ref .
@xcite for derivation of the above equations .
the above marginals are enough to obtain the local expectation values @xmath42 .
we can use the bethe formalism to compute the higher order correlations @xmath43 for @xmath44 , @xmath45}=e^{\sum_i \delta_i-\sum_{(ij)\in \mathsf{t}}\delta_{ij}},\end{aligned}\ ] ] where @xmath46 the @xmath34 are the bp cavity marginals given before , and the new cavity marginals satisfy the following bp equations : @xmath47 along the same lines , one can compute the probability of observing a subset of spin variables in a given configuration , @xmath48 .
the extension of the above equations to loopy graphs gives the approximate loopy bp equations ; this is simply done by replacing @xmath49 with @xmath50 in the above equations .
let us expand the partition function @xmath11 for small couplings in @xmath51 to get @xmath52 where @xmath53 identifies a loop configuration with @xmath54 showing the absence or presence of an interaction . here
the averages @xmath56 } \prod_{(ij ) \in \mathcal{l}}(\sigma_i\sigma_j)^{s_{ij}},\end{aligned}\ ] ] can be computed efficiently , as described in the previous section .
note that these averages are trivial when all the nodes in the subgraph @xmath57 , induced by the present loopy interactions in @xmath58 , have an even degree .
we rewrite the right hand side of the above expansion in another form as @xmath59 with @xmath60 independent of the value of @xmath61 .
the leading term @xmath62 is obtained when all the @xmath63 are zero . in this way
, we obtain a high - temperature expansion for the loopy interactions : @xmath64 where now @xmath65 let us define @xmath66 as the number of ferromagnetic loopy interactions in loop configuration @xmath58 .
the step function @xmath67 for ferromagnetic loopy interactions ( @xmath68 ) , otherwise @xmath69 .
this allows us to separate the positive and negative contributions to the expansion , @xmath70 we know that the right hand side is positive , and if we are interested in the free energy densities @xmath71 in the thermodynamic limit , then @xmath72 in other words , it is @xmath73 that is responsible for any nonanalytic behavior of the free energy density .
note that instead of working with the ferromagnetic loopy interactions , we could work with the anti - ferromagnetic ones to get @xmath74 where @xmath75 . here
the positive contribution to the expansion comes from loop configurations with an even number of anti - ferromagnetic loopy interactions @xmath76 .
thus , in the thermodynamic limit , we have @xmath77 the two different loop expansions provide the same free energy , therefore , @xmath78 these are indeed relationships between the free energies of two interacting systems of loop variables , with effective energy functions @xmath79 $ ] that depend on the couplings of the original system defined by @xmath80 $ ] . and
, in the presence of a phase transition , the two systems display the same critical behavior .
finally , we can use both the ferromagnetic and anti - ferromagnetic representations to write the loop expansion as @xmath81 now @xmath82 , @xmath83 , and we defined @xmath84 . note that here every loop configuration @xmath58 has a nonnegative contribution to the expansion .
the average @xmath85 is the probability of having @xmath86 for all the loopy interactions .
we denote this probability by @xmath87 .
the complexity of computing this probability grows exponentially with the number of connected clusters in the graph @xmath57 induced by the set of present loopy interactions in @xmath58 ; the neighboring spins in each cluster should satisfy @xmath88 , and this is possible only if each cluster of loopy interactions is free of frustration , otherwise @xmath89 .
in other words , only frustration - free loop configurations enter in the above expansion , and each configuration has a nonnegative contribution .
so far , we have made no approximation in deriving the high - temperature loop expansions presented in the previous section .
but , to obtain the loop corrections , still we have to compute the nonlocal loop interactions that appear in these expansions . in this section
, we will focus on different approximations to simplify the loop interactions . for tree interaction graphs , we are sure that the averages @xmath90 are computed within a pure state , where clustering holds and with a good approximation for distant loopy interactions @xmath91 this independent - loop ( il ) approximation results in @xmath92 .
\end{aligned}\ ] ] note that in a ferromagnetic system this approximation underestimates the correlations and thus provides an upper bound for the free energy . in a homogeneous system @xmath93 , and in the absence of external fields
@xmath94 , we have @xmath95 , thus @xmath96 where @xmath97 is the number of loopy interactions of length @xmath98 . in general , we obtain the magnetization by @xmath99}{\cosh k_{kl}+\sinh k_{kl}\left\langle \sigma_k\sigma_l \right\rangle_0 } , \end{aligned}\ ] ] and the two - spin correlation @xmath100 by @xmath101}{\cosh k_{kl}+\sinh k_{kl}\left\langle \sigma_k\sigma_l \right\rangle_0 } , \hskip0.5 cm ( ij)\notin \mathcal{l}. \end{aligned}\ ] ] in fig .
[ f3 ] we show the free energy obtained by the il approximation for the ferromagnetic ising model on the two - dimensional square lattice .
we also compare the il approximation with a mean - field approximation of the loopy interactions @xcite , where a loopy interaction @xmath102 is replaced with two effective fields @xmath103 and @xmath104 acting on spins @xmath18 and @xmath19 . here
the local magnetizations @xmath105 are determined self - consistently .
one can indeed do better than this naive approximation by replacing the @xmath106 in the effective fields with the cavity magnetizations @xmath107 computed in the absence of the loopy interaction . ) on the @xmath28 square lattice is compared with the independent - loop approximation based on a random and regular ( chain ) spanning tree .
( b ) the il approximation compared with a mean - field approximation of loopy interactions ( mfl ) , where a loopy interaction @xmath17 is replaced with effective fields @xmath103 and @xmath104 determined self - consistently .
the arrows show the position of the ferromagnetic phase transition.,width=604 ] in the independent - loop ( il ) approximation we do not consider the overlap of the neighboring loopy interactions . an approximate way to do this is the following : @xmath108 leading to @xmath109 where @xmath55 , and @xmath110 this independent - site ( is ) approximation , which again relies on the clustering property , is appropriate for systems in external fields where the local magnetizations @xmath111 are nonzero .
now , the sum over the loop configurations can be computed by the bethe approximation , which is exact when the loopy interactions have a tree structure . here
we need the cavity messages @xmath112 for loop variables @xmath63 in the absence of @xmath113 and @xmath114 .
these cavity messages are governed by the following bethe equations : @xmath115 we solve the above equations by iteration , and compute the free energy from @xmath116 , where @xmath117 in fig .
[ f4 ] we compare the above free energy with the exact and loopy bp free energies for random ising models on the two - dimensional square lattice .
we see that in the presence of external fields , the is free energy is very close to the one obtained by the loopy bp , and at the same time , the is algorithm converges down to very low temperatures , where the loopy bp does not converge . ) on the @xmath28 square lattice .
the positive and negative values for the fields and the couplings are chosen with equal probability .
the loopy bp algorithm does not converge for @xmath118 indicated by the arrows.,width=377 ] here the local magnetizations are given by @xmath119 which can easily be computed given the @xmath112 and derivatives @xmath120 note that the bethe free energy @xmath121 is stationary with respect to changes in the messages @xmath112 , computed at the fixed point of the bp equations
. thus we can safely take @xmath122 in appendix [ app - mc ] we obtain the two - spin correlations @xmath123 .
figure [ f5 ] displays the error we make in estimating the local magnetizations by the il and is approximations .
the figure also shows the results obtained by the loopy bp algorithm , which is still doing better than the above approximations . in the following ,
we propose a systematic scheme to obtain more accurate approximations .
, on the @xmath28 square lattice .
both the il and is approximations are based on a maximum - weight spanning tree , with weight @xmath124 .
the @xmath125 are computed by an exact algorithm ( for the smaller system in panel(a ) ) or by the monte carlo algorithm ( for the larger system in panel ( b ) ) .
the errorbars are about the size of the symbols .
the rms error is given by @xmath126.,width=604 ] more accurate expressions for the nonlocal loop interactions can be obtained by the bethe approximation , for example , @xmath127 here , @xmath128 and @xmath129 are , respectively , the probability of having @xmath130 and @xmath131 , for a given loop configuration @xmath58 .
more precisely , @xmath132 by @xmath129 we are indeed taking into account interactions between the neighboring loopy interactions @xmath133 with distance @xmath134 . the above approximation for the nonlocal loop interactions can be used in any of the high - temperature loop expansions given in sec . [ s2 ] . in appendix
[ app - even ] we show how the even or odd parts of expansions [ l2 ] and [ l4 ] can be computed by an approximate message - passing algorithm . in the following
, instead we focus on the expression given by eq .
[ l5 ] , where all the expansion terms are nonnegative .
then we have @xmath135 where now @xmath136 the sum over the loop configurations can be computed within the bethe approximation . here the cavity marginal @xmath112 in the absence of @xmath137 , is recursively obtained by the following bp equations @xcite : @xmath138 the solutions to these equations are found by iteration ; starting from arbitrary cavity messages , we update the messages according to the above equations for a sufficiently large number of iterations to reach a fixed point of the equations . then , the free energy reads as @xmath139 , where @xmath140 and @xmath126 , obtained by different approximations : independent - loop ( il ) , correlated - loops ( cl ) , loopy belief propagation ( lbp ) , and generalized belief propagation ( gbp ) for random ising models with random couplings uniformly distributed in @xmath141 on the @xmath28 square lattice .
the fields are zero in panels ( a ) and ( b ) , but they are nonzero and uniformly distributed in @xmath141 in panels ( c ) and ( d ) . both the il and cl approximations are based on a maximum - weight spanning tree , with weight @xmath124 .
the @xmath142 and @xmath125 are computed by an exact algorithm for the smaller system in panel ( a ) , and by the monte carlo algorithm for the larger systems in panels ( b)-(d ) .
the errorbars are about the size of the symbols.,width=604 ] in appendix [ app - mc ] we give the expressions for @xmath143 and @xmath144 obtained by taking derivatives of the above free energy .
in particular , we can easily obtain an estimation of spin correlations on loopy interactions , @xmath145 note that in computing the above derivatives , we need only to consider the explicit dependence of the free energy on @xmath146 , because the bethe free energy is stationary with respect to changes in the cavity messages at the fixed point .
figure [ f6 ] displays the error in spin correlations on loopy interactions obtained by the above correlated - loops ( cl ) approximation for some random ising models on the two - dimensional square lattice . for comparison
, the figure also shows the error obtained by the il approximation , loopy bp algorithm , and generalized bp ( shown only for the small system in panel ( a ) ) . the latter works with larger regions of variables ( here plaquettes ) to extend the consistency range of the local marginals @xcite .
we see that already the il approximation is working better than the loopy bp in the absence of the external fields . and , considering the interactions between the neighboring loop variables in the cl approximation results in spin correlations that are comparable to those of the generalized bp .
nevertheless , one could still obtain better results with the loopy bp for small temperatures , especially in the presence of strong fields .
the same is true for homogeneous systems with uniform couplings , where the absence of a dominant spanning tree means we have to take into account the longer - ranged interactions between the loop variables .
we can still improve on the above approximation by increasing the interaction range of the loop variables .
for instance , to get better approximations , we may add interactions between the more distant loop variables , @xmath147 where @xmath148 the interactions @xmath149 can be limited to close loopy interactions with distances @xmath20 smaller than a given distance @xmath150 .
again , from the clustering property , we expect to obtain a good approximation already for small @xmath150 .
we leave further studies in this direction for future works .
we started from a high - temperature loop expansion for the ising model on a loopy interaction graph regarded as a globally interacting system of loop variables .
we then suggested different approximations for the nonlocal loop interactions , writing the problem of computing loop corrections in a form that is amenable to local message - passing algorithms .
the quality of these approximations depends on the structure of the loopy interactions ; when the loopy interactions have a tree structure , we can use the belief propagation algorithm to compute `` exactly '' the loop corrections within a local approximation of the nonlocal loop interactions .
this provides approximate message - passing algorithms that rely on the structure of the loopy interactions , and thus exhibiting better convergence properties than the loopy bp algorithm . in this way
, we could obtain more accurate estimations for spin correlations on loopy interactions from these approximations than from the loopy bp algorithm for the ising model on the @xmath28 square lattice with random couplings .
it would be interesting to find out the typical loop configurations close to a phase transition point .
the relevant loop configurations with no frustration could behave differently as we approach ferromagnetic or spin - glass phase transition points .
close to a continuous phase transition the loop variables are strongly correlated , and the global nature of loop interactions becomes more important .
therefore , it would be important to have better approximations for the nonlocal loop interactions , and work with the higher - order interactions between the loop variables .
finally , instead of starting from a spanning tree in the loop expansions , we could start from any subgraph that allows for accurate computation of the local expectations needed in the expansions .
for instance , the starting subgraph can be obtained by adding an appropriate subset of loopy interactions to the spanning tree @xcite .
we would like to thank a. braunstein , e. daryaei , and j. realpe - gomez for reading the manuscript and helpful discussions .
given the free energy @xmath12 , we obtain the local expectation values from @xmath151 and @xmath100 . in the independent - site (
is ) approximation the free energy is given by @xmath152 , where @xmath153 is the free energy of the system living on the spanning tree .
the other terms give the loop corrections computed by the belief propagation ( bp ) algorithm in the is approximation .
the precise definitions along with the computation of the local magnetizations can be found in the main text , sec .
for the two - spin correlations , we have @xmath154 to compute the above quantities we need to know the bp messages @xmath112 satisfying eq . [ bp - is ] , and the following derivatives : @xmath155 here the loop variables @xmath156 show the absence or presence of loopy interactions , and the averages @xmath157 are taken in the tree interaction graph . in the correlated - loops ( cl )
approximation , the free energy reads @xmath158 .
the precise definitions are given in the main text , sec .
[ s33 ] . here
we obtain a local magnetization by @xmath159 where @xmath160 to compute the loop corrections , we need the cavity messages @xmath112 governed by eq . [ bp - cl ] , and the following derivatives : @xmath161 the two - spin correlations in the cl approximation are given by @xmath162 where @xmath163 in the above equations , we defined @xmath164
in this section we explain an approximate message - passing algorithm to compute the contribution of loop configurations @xmath166 with an even number of ferromagnetic loopy interactions , @xmath167 here @xmath66 , @xmath168 is the heaviside function , and @xmath169 .
the aim is to find an estimation of @xmath165 by the bethe approximation , given an approximate factorization of the nonlocal loop interactions @xmath170 here @xmath171 , and @xmath128 is the probability of having @xmath130 in the tree interaction graph .
similarly , @xmath129 is the probability of having @xmath131 .
then we have @xmath172 with @xmath173 let us for simplicity assume that the graph @xmath174 induced by the loopy interactions is a tree , otherwise we choose one of its spanning trees @xcite .
then , we introduce auxiliary variables @xmath175 on the directed links of @xmath174 , where @xmath176 gives the parity of the ferromagnetic loopy interactions in the cavity tree @xmath177 ; a branch of the tree that includes @xmath18 and cavity trees @xmath178 . more precisely , @xmath179 now we have to work with extended marginals @xmath180 governed by the following bp equations @xmath181 the solution to these equations can be found by iteration .
then , for an even number of ferromagnetic loopy interactions , the free energy shifts are given by @xmath182 where @xmath183 and the local @xmath184 is computed like the cavity one @xmath185 but considering the contribution from all the neighbors in @xmath25 .
finally , the even free energy reads as @xmath186 .
|
we consider an interacting system of spin variables on a loopy interaction graph , identified by a tree graph and a set of loopy interactions .
we start from a high - temperature expansion for loopy interactions represented by a sum of nonnegative contributions from all the possible frustration - free loop configurations .
we then compute the loop corrections using different approximations for the nonlocal loop interactions induced by the spin correlations in the tree graph .
for distant loopy interactions , we can exploit the exponential decay of correlations in the tree interaction graph to compute loop corrections within an independent - loop approximation .
higher orders of the approximation are obtained by considering the correlations between the nearby loopy interactions involving larger number of spin variables . in particular
, the sum over the loop configurations can be computed `` exactly '' by the belief propagation algorithm in the low orders of the approximation as long as the loopy interactions have a tree structure .
these results might be useful in developing more accurate and convergent message - passing algorithms exploiting the structure of loopy interactions .
|
a pulsar wind nebula ( pwn ) is a bubble of relativistic plasma energized by a rapidly rotating , magnetized neutron star .
the prototypical pwn is the crab nebula , which is decisively the best studied celestial object beyond our solar system , having served for decades as a testbed for theories of astrophysical outflows and their radiative processes .
pwne are also of broad interest in astroparticle physics , as potential sources of galactic positrons @xcite and ultra - high energy cosmic rays @xcite .
more generally , pwne have ultra - energetic ( albeit hypothetical ) counterparts in the winds of so - called `` millisecond magnetars . ''
such exotic objects may be formed in the coalescence of binary neutron star systems , in which case they could yield the first electromagnetic counterparts to gravitational wave detections @xcite . or , if formed during the core - collapse of a massive star , a millisecond magnetar could re - energize the ejecta shell , helping to explain the light - curves of certain hydrogen - poor superluminous supernovae @xcite .
morphological and radiative characteristics of pulsar winds are shaped by the rate with which they dissipate magnetic energy .
this can be seen from basic considerations of the crab . while energy streams away from the pulsar in the form of an ultra - magnetized plasma wind ( e.g. * ? ? ?
* ) , throughout the nebula it is shared equitably with particles @xcite . to be sure ,
low magnetization levels in the range of @xmath2 are required to explain the nebula s expansion speed @xcite , synchrotron spectrum @xcite , and mildly prolate appearance @xcite .
such low magnetization can not be attained in the absence of dissipative effects @xcite . at the time of the early models , little
was known about how and where magnetic dissipation operates in a pwn , so the issue was dealt with pragmatically .
@xcite assigned a small , nominal magnetization to the plasma emerging from the inner nebula , and modeled its flow henceforth adiabatically . by construction , this procedure leaves any dissipative processes , and thus the electron heating profile unspecified ; all the nebula s relativistic electrons appear in this description to be sourced from the immediate vicinity of the wind termination shock . today
the details of magnetic energy dissipation in the crab are much better understood .
magnetic field supplied by the pulsar has distinct ac and dc components , which dissipate in different ways .
the ac field ( also referred to as striped wind
* ; * ? ? ?
* ; * ? ? ?
* ) consists of magnetic reversals , which dissipate by mixing and annihilating one another .
the striped wind fills the equatorial wedge of the freely expanding pulsar wind , whose opening angle is determined by the pulsar s magnetic obliquity ( e.g. * ? ? ?
the pulsar wind s dc power on the other hand , is carried outward by a regular toroidal magnetic field . after being decelerated at the so - called wind termination shock ( or a series of shocks , * ? ? ?
* ) , the toroidal field s hoop stress induces flow toward the polar axis @xcite , yielding an unstable @xmath3-pinch configuration @xcite seen in x - ray observations ( e.g. * ? ? ?
* ) as the crab s polar jet feature .
dynamical instabilities act continuously to convert the energy of freshly supplied magnetic hoops into turbulence and eventually heat @xcite .
our aim in this paper is to develop a model for the flow of magnetized plasma in a pwn , which captures these dissipative processes as they influence the object s morphological and radiative characteristics .
most of the new ideas presented here go toward crafting a physically motivated prescription for the dissipation of the magnetic field by _
our key assumption is that relaxation proceeds dynamically , in the sense that magnetic energy density may be assigned a local half - life @xmath4 , determined at each point by a comoving alfvn speed and `` eddy '' scale .
this assumption is motivated by recent numerical work , which reveals a tendency for magnetic field configurations to relax and dissipate by exciting turbulent motions , provided that states of lower magnetic energy are topologically accessible in the sense of @xcite .
this principle was illustrated by @xcite and @xcite using high resolution numerical simulations of prototypical force - free equilibria ( so - called abc fields , * ? ? ?
more specific to the crab , both sources of magnetic free energy supplied by the pulsar the striped wind and the large - scale magnetic hoops are seen to dissipate by turbulent relaxation .
simulations reported in @xcite indicate the alternating magnetic field pattern survives only of order its proper alfvn time before most of its energy has been dissipated into turbulence , while simulations by @xcite and also @xcite make analogous predictions regarding relaxation of the large - scale toroidal magnetic field by kink instabilities .
the general formalism we adopt for solving the wind equations , and a precise explication of our dissipation model , is presented in section [ sec : equations ] .
much of the present paper focuses on validating this model for magnetic dissipation in pwne by matching features of the crab nebula .
we explore the ac and dc dissipation modes in turn , treating dissipation of striped wind in section [ sec : striped - wind ] , and relaxation of the large - scale toroidal field in section [ sec : mean - field ] .
the purely ac case corresponds to a hypothetical nebula powered by an orthogonally rotating pulsar , while the purely dc case corresponds to an aligned rotator . while the crab pulsar s
obliquity is likely somewhere in between ( e.g. * ? ? ?
* ) , we show in section [ sec : mean - field ] that the nebula s appearance can actually be described in terms of a dc - only wind , in which relaxation of magnetic structures having the scale of the termination shock yields a volume - average magnetization around @xmath5 . by comparison , in a purely ac wind
, dissipation runs too quickly and the nebular magnetization falls to @xmath6 .
our model is based on steady - state non - ideal mhd flow in spherical geometry .
flow solutions are obtained using robust numerical methods for ordinary differential equations . in this way we are able to simulate realistic wind lorentz factors and magnetization levels ( each reaching @xmath7 ) .
such ultra - relativistic conditions can not be accessed by typical godunov - type mhd codes .
our formalism also includes a self - consistent treatment of optically thin synchrotron radiation , which we describe in sections [ sec : conservation - laws ] and [ sec : synchrotron - radiation ] . energy and momentum
losses to radiation only marginally influence the dynamics of the crab , as its radiative efficiency is in the vicinity @xmath8 @xcite .
however , we point out in section [ sec : gamma - ray - revshock ] that radiative losses will become dominant in certain scenarios of interest .
in particular , forced magnetic reconnection in a pulsar wind that is shocked at relatively close range due to confinement by a dense medium is found to convert essentially @xmath9 of the wind power into @xmath1-rays .
we comment on how this might impact theories of astrophysical transients for which sustained energy injection by a millisecond magnetar is invoked . in section
[ sec : crab - flares ] we discuss implications of our model for the crab nebula @xmath1-ray flares , and in section [ sec : particle - spec ] we propose a means by which our model may yield improved calculations of the non - thermal particle spectrum and evolution in pwne . throughout the paper ,
we utilize a notation in which @xmath10 denotes a speed ( normalized by @xmath11 ) while @xmath12 and @xmath13 denote the corresponding four - speed and lorentz factor . proper density of plasma is denoted by @xmath14 , which is implicitly multiplied by @xmath15 so it has dimensions of rest - mass energy per unit volume .
variables with a subscript zero , e.g. @xmath16 indicate values at the base of the wind .
here we develop equations for a stationary , relativistic plasma wind , subject to radiative losses and turbulent magnetic reconnection .
we adopt the toroidal mhd approximation , in which the flow is radial and the magnetic field is transverse . the flow is envisioned to contain a small - scale magnetic free energy which is in a state of turbulent relaxation . the free energy , or `` eddy '' scale is denoted by @xmath17 , and unlike earlier analyses of magnetic thermalization in relativistic outflows , we evolve that scale in a manner consistent with numerical simulations of turbulent magnetic relaxation in relativistic systems @xcite .
this picture is adapted to the freely expanding pulsar striped wind by initiating @xmath17 to the distance between magnetic reversals ( @xmath18 , where @xmath19 is the pulsar rotation period ) , and to the large - scale nebular magnetic field by initiating @xmath17 to the termination shock radius . in the equations that follow ,
the flow variables are understood to represent averages over scales that are larger than @xmath17 , yet smaller than the global system size .
the equations of relativistic mhd are given by conservation of mass , energy , momentum , and magnetic flux .
here , we will be using one - dimensional spherical coordinates and assuming a steady - state flow , so that only derivatives with respect to the radial coordinate @xmath20 are non - zero .
conservation of mass is given in general by @xmath21 where @xmath14 is the comoving density , and @xmath22 is the four - velocity . in
spherically symmetric flow , the rate of mass loss per steradian , @xmath23 , is a constant at each radius , where @xmath24 is the radial four - velocity . for optically thin plasma , conservation of energy and momentum
is given by @xmath25 where @xmath26 is the stress - energy tensor , and @xmath27 is the comoving emissivity , assumed to be isotropic in the plasma rest - frame .
the time component of equation [ eqn : rad - mhd ] is the energy conservation law , which dictates that the change in wind luminosity @xmath28 at each radius is balanced by the radiative power . here
, @xmath29 is the luminosity per unit mass loss , with @xmath30 denoting the total specific enthalpy .
@xmath31 is constant at each radius when radiation is neglected .
the @xmath20-component of equation [ eqn : rad - mhd ] is radial force balance , @xmath32 = -r^2 u \dot \epsilon + 2 p r \ , , \ ] ] which expresses cancellation of the total pressure gradient ( ram , gas , and magnetic ) and inward magnetic tension force , and is the bernoulli equation for this system . here
, @xmath33 denotes the comoving magnetic field ( for notational convenience , @xmath34 is normalized so that magnetic pressure is @xmath35 ) .
the term proportional to @xmath27 represents wind inertia carried away by photons .
magnetic flux transport in a flow containing magnetic free energy is complicated by non - ideal effects .
our approach here is to subsume the non - linear reconnection physics into a dissipation function , denoted by @xmath36 , which prescribes the loss of magnetic flux over distance . the ideal mhd induction equation
, @xmath37 implies that @xmath38 , when the flow is toroidal and stationary ( in equation [ eqn : ideal - faraday ] , @xmath39 ) .
dividing by @xmath40 , we see that @xmath41 is a non - dissipative invariant . here , @xmath42 is the ratio of the wind s electromagnetic to kinetic power .
note that this definition of @xmath0 does not include gas enthalpy in the denominator , and that the definition of @xcite , denoted as @xmath43 will be used later on in section [ sec : mean - field ] . to close the system
we adopt a @xmath44-law equation of state , for which the gas pressure @xmath45 , where @xmath46 is the thermal energy per unit rest - mass energy .
the corresponding specific entropy is @xmath47 . throughout we will use the value @xmath48 appropriate for relativistic gas particles .
the total specific enthalpy is given by @xmath49 , where @xmath50 is the thermal enthalpy per unit rest - mass energy .
the wind equations can now be written down as @xmath51 which are compact versions of the continuity equation , the energy equation , the radial force balance , the phenomenological flux transport law , and the thermodynamic entropy relation .
all terms appearing on the right - hand side would be zero when the wind is non - radiative and non - dissipative .
the derivative @xmath52 may be with respect to @xmath20 , or to the proper elapsed time of a fluid element .
primes denote @xmath53 , while dots are comoving time derivatives , e.g. @xmath54 .
equations [ eqn : df ] through [ eqn : ds ] may be combined is obtained starting with equation [ eqn : dp ] .
we then make the substitutions @xmath55 and @xmath56 , where the subscripts denote partial derivatives of the equation of state , written as @xmath57 .
we then insert expressions for @xmath58 and @xmath59 obtained from equations [ eqn : deta ] and [ eqn : dzeta ] .
this leaves an expression in which the only remaining differentials are @xmath60 , @xmath61 , @xmath62 , and @xmath63 .
we finally divide through by @xmath61 and set @xmath64 and @xmath65 . ] to yield the ordinary differential equation for the four - velocity , @xmath66 where @xmath67 , @xmath68 , and @xmath69 are given by @xmath70 and @xmath71 is the fast magnetosonic speed , @xmath72 numerical wind solutions are obtained by integrating the unknowns @xmath73 , @xmath74 , and @xmath31 simultaneously in @xmath20 , using equation [ eqn : wind - ode ] , and chosen prescriptions for @xmath75 and @xmath36 .
the wind equations may also be arranged to give the rate of entropy generation . by inserting the thermodynamic enthalpy relation @xmath76 ( which is equivalent to equation [ eqn : ds ] ) into equation [ eqn : dp ] , and then substituting equations [ eqn : deta ] and [ eqn : dzeta ]
, we obtain @xmath77 this reflects that any non - ideal mhd processes @xmath78 generate entropy at the expense of otherwise frozen - in magnetic flux , while radiative losses @xmath79 reduce entropy by cooling the plasma .
it is worth pointing out certain mathematical features of equation [ eqn : wind - ode ] .
first , the toroidal mhd wind can not go smoothly through a fast magnetosonic point , regardless of how dissipation occurs .
second , the post - shock flow is always subsonic , and decelerates to a terminal speed proportional to the net magnetic flux .
we briefly explain these points here
. the lack of critical points can be shown by inspecting the polynomials @xmath67 and @xmath68 .
for @xmath73 to increase smoothly through @xmath80 , the ratio @xmath81 would need to be finite there .
but @xmath82 when @xmath83 , so @xmath68 would have to vanish simultaneously .
the condition @xmath84 yields the quartic polynomial @xmath85 for equation [ eqn : q - polynomial ] to have a root at @xmath80 , the condition @xmath86 must also be met .
if equation [ eqn : degenerate - solution ] is satisfied , then the wind has zero temperature and coasts along at the fast magnetosonic speed .
if equation [ eqn : degenerate - solution ] is not satisfied , then @xmath67 and @xmath68 have no simultaneous roots , and thus @xmath87 can not be finite where @xmath83 . these conclusions do not depend on the value of @xmath69 , so however dissipation or radiation may occur , the flow will not go smoothly through a fast magnetosonic point . since @xmath88 when @xmath84 ( assuming for the moment that @xmath89 ) , the roots of equation [ eqn : q - polynomial ] represent asymptotic wind speeds .
the subsonic solution branch has a terminal speed given roughly by @xmath90 , so as mentioned before , the post - shock flow moves away with a constant speed proportional to the magnetic flux .
importantly , magnetic dissipation oppositely affects the subsonic and supersonic solution branches .
it is easily seen that @xmath91 ( since @xmath92 and @xmath93 ) , while @xmath67 is negative for supersonic flow and positive for subsonic flow . since @xmath94 ( now assuming that @xmath84 ) , it is clear that magnetic dissipation _ accelerates _ the supersonic ( pre - shock ) flow and _ decelerates _ the subsonic ( post - shock ) flow .
evolution of @xmath31 ( the wind luminosity per particle ) occurs only as the result of radiative losses .
if radiation were neglected , the plasma energy and momentum would be conserved and the right - hand - side of equation [ eqn : rad - mhd ] would be equal to zero . here
we assume that synchrotron is the dominant radiative mechanism , that the nebula is optically thin to the emitted photons , that gas pressure is isotropic , and that particles are mono - energetic , with the thermal lorentz factor given by @xmath95 .
the comoving synchrotron luminosity per particle is given by @xcite @xmath96 where @xmath97 is the thomson cross section and @xmath98 is the magnetic energy density .
the emissivity is given by @xmath99 , and recalling that @xmath100 ( equation [ eqn : deta ] ) , we have @xmath101 where @xmath102 is the pulsar s particle production rate , @xmath103 is the classical electron radius , and @xmath104 is the electron mass .
further useful diagnostics to be encountered later on in section [ sec : mean - field - procedure ] include the synchrotron frequency , @xmath105 where @xmath106 and @xmath107 is the comoving electron gyro - radius , and the dimensionless radiated power @xmath108 our assumption that particles are distributed mono - energetically , @xmath109 , may be dropped in a more sophisticated analysis . for example , if one assumes power - law distributed particle energies , @xmath110 then an additional parameter @xmath111 needs to be specified independently .
a reasonable choice for @xmath112 is the energy of a particle whose gyro - radius is marginally confined by the local turbulent eddies , @xmath113 . of course
, that may be an underestimate since particles experiencing bohm diffusion @xmath114 continue to be accelerated by the second - order fermi process . in modeling emission from the crab nebula
, @xcite assumed that particles reach the energy at which they would be marginally confined within the termination shock radius , @xmath115{cm}$ ] .
here we develop a prescription for the local rate @xmath36 of magnetic dissipation in the pulsar wind nebula .
our starting assumption is that magnetic free energy has a half - life @xmath116 , where @xmath17 is the proper scale of magnetic fluctuations ( the `` eddy '' scale ) and @xmath117 is the local alfvn speed .
given @xmath4 , the evolution of @xmath74 is given simply by @xmath118 .
this expression is found by first rearranging @xmath119 , and then separating @xmath120 into ideal and dissipative parts , @xmath121 equating the dissipative term @xmath122 with @xmath123 yields the expression @xmath118 .
so far , this procedure for treating dissipation is equivalent to that employed by @xcite and by @xcite for analysis of magnetic dissipation in gamma - ray burst outflows .
it is also similar to the formalism developed by @xcite and @xcite to characterize magnetic reconnection in the pre - shock pulsar striped wind .
the only significant difference is that the latter authors utilized various microphysical prescriptions to specify the speed @xmath124 of magnetic reconnection . in appendix
[ app : reconnection - fronts ] we show that our approach and theirs are equivalent if @xmath124 is instead taken to be the alfvn speed , and the comoving stripe separation is chosen as the eddy scale , @xmath125 .
a potentially significant effect , that was not accounted for in earlier studies , is evolution of the eddy scale .
in particular , growth of @xmath17 over time is now understood to be a general feature of turbulent magnetic relaxation @xcite .
previously , the `` inverse energy transfer '' was thought to operate only when the field had a significantly non - zero magnetic helicity measure ( see e.g. * ? ? ?
mhd simulations presented in @xcite demonstrate that a magnetic field , that is initially tangled isotropically at a small scale , relaxes according to @xmath126 , while @xmath127 .
turbulent motions are sustained at roughly the comoving alfvn speed . over time
, the cascade slows because the alfvn speed drops , and also because the eddy scale grows .
this process yields self - similar temporal and spectral evolution that persists while @xmath17 remains smaller than a global system scale ( the nebula in the present context ) .
the scaling behavior of turbulent relaxation reported in @xcite was limited to conditions where the plasma was overall at rest in a periodic box .
however , we are interested here in decaying turbulence that is embedded into a flow , and whose scale is thus influenced by the expansion or compression of that flow as it accelerates or decelerates . in other words ,
evolution of @xmath17 is driven by adiabatic effects , and simultaneously by the dissipative action of the cascade .
for this reason , we choose to characterize eddies based on their mass @xmath128 , which remains fixed under the adiabatic effects alone .
the eddy mass and scale are related through the ambient density @xmath14 , and the assumption that eddies are isotropic ; @xmath129 . when the flow is not expanding or contracting , the two laws @xmath130 and @xmath131 say the same thing ( with appropriate prefactors ) .
evolution of @xmath128 in this way simply states that magnetic structures intend to reduce their energy by merging with one another .
merging between eddies proceeds over a dynamical time , and results in an eddy with double the mass .
the process repeats itself hierarchically , while @xmath4 adjusts to the local values of @xmath17 and @xmath132 , @xmath133 more details on the turbulent magnetic relaxation picture , including a sensible explanation of the inverse energy transfer , are provided in appendix [ app : plasmoid - cascade ] .
( top ) , and magnetization @xmath0 ( bottom ) as a function of distance , for different models of a dissipative mhd wind with @xmath134 .
the thick solid curve is the `` fast - reconnection '' picture described in @xcite .
the vertical dashed line indicates the radius @xmath135 at which a dynamical time has elapsed in the comoving frame , and where the fast reconnection picture should break down .
the dashed line shows evolution according to the turbulent relaxation picture.,width=340 ] winds driven by obliquely rotating pulsars contain `` stripes '' around the equatorial region .
the stripes are reversals of the azimuthal magnetic field , separated by a half - wavelength @xmath136 @xcite .
magnetic reconnection across the reversals has been considered as a possible mechanism by which the wind s electromagnetic energy may be dissipated upstream of the termination shock @xcite . however , those treatments neglected feedback of the dissipated magnetic energy on the global flow . energy conserving analyses @xcite revealed that dissipated magnetic energy goes toward accelerating the flow , due to reduction of inward - pointing magnetic hoop stress .
as the flow moves faster , relativistic time dilation suppresses the dissipation rate as seen in the pulsar frame .
the crab pulsar s wind becomes weakly magnetized by the time it reaches the termination shock if it is at the upper end of believable mass - loading @xmath137{s^{-1}}$ ] .
the analysis given in @xcite was based on a plane - parallel reconnection model , in which cold , magnetized plasma fills the volume between hot , unmagnetized current layers .
hot current layers expand into the cold plasma at a speed @xmath124 , consuming magnetic energy as they proceed .
the fraction of volume occupied by hot plasma , denoted as @xmath138 , indicates how much of the initial magnetic energy has been consumed , @xmath139 .
to first order in both the expansion speed @xmath124 and the inverse lorentz factor @xmath140 , @xmath138 evolves according to ( see appendix [ app : reconnection - fronts ] ) @xmath141 thus , lyubarsky s prescription and ours are made equivalent by the identifications @xmath142 , @xmath143 , and @xmath144 , and by replacing @xmath139 with @xmath145 , so that @xmath138 is formally allowed to exceed unity .
conveniently , @xmath138 may here be interpreted as the number of elapsed alfvn times felt by a fluid parcel .
it was argued in @xcite that the planar geometry of the current layers would only persist until order one comoving alfvn times had elapsed , @xmath146 .
subsequently the striped magnetic field becomes isotropic and dissipation should be described by turbulent relaxation .
thus , the onset of isotropic turbulence accompanies the transition to particle - dominated flow , which again , occurs upstream of the shock only if the wind is well mass - loaded .
nevertheless , we will explore briefly evolution of the striped wind , if it were allowed to expand freely beyond the turbulence transition .
we do this by setting @xmath17 to @xmath147 until @xmath148 , and henceforth evolving @xmath17 according to the turbulent relaxation model . for as long as @xmath149
, @xmath0 remains large and thus @xmath150 .
we expect the asymptotic solutions to obey the same scaling as the `` fast reconnection '' picture presented in @xcite , because that also adopts constant reconnection speed ( that of sound @xmath151 in the hot plasma ) .
figure [ fig : fastvsturbulentreconnection ] illustrates this equivalence for a hypothetical wind having a magnetization @xmath134 , and which continues to @xmath152 without passing through a termination shock .
we see that indeed @xmath138 and @xmath73 both evolve @xmath153 . beyond @xmath148 , the solution is integrated in two ways : ( 1 ) by extending the fast reconnection picture , where @xmath17 remains equal to @xmath147 and ( 2 ) by using the turbulent relaxation picture where @xmath17 increases according to equation [ eqn : turbulent - relaxation ] . since roughly half of the magnetic energy has already been spent by the time @xmath146 , further evolution of @xmath73 is not affected much by how dissipation is prescribed where @xmath154 .
magnetization @xmath0 , on the other hand , evolves much more slowly when turbulent growth of @xmath17 is accounted for . , cascade number @xmath138 , magnetic field strength @xmath155 , electron temperature @xmath156 , photon energy @xmath157 , and radiated power @xmath158 , all as a function of distance downstream of the shock , for particular wind parameters @xmath159{erg / s}$ ] , @xmath160{s^{-1}}$ ] , and @xmath161 .
dissipation occurs in the `` fast reconnection '' regime upstream where @xmath162 , and in the turbulent regime once @xmath163 .
the horizontal dashed line in the top - most panel indicates the nominal nebula expansion speed @xmath164{km / s}$ ] , while the vertical dashed line marks the nominal nebula boundary @xmath165{pc}$].,width=340 ] now we consider dissipation of the striped wind beyond the termination shock , assuming that it arrives there well magnetized . at the shock ,
the flow is compressed and decelerated below the fast magnetosonic speed .
so , as discussed in section [ sec : general - properties ] , any dissipation that occurs beyond it only further decelerates the flow .
such `` forced reconnection '' has been studied in depth for its possible role in energizing at least some of the crab nebula s radio - emitting , non - thermal electrons . however , as pointed out by @xcite , the post - shock temperature is likely high enough that electron gyro - radii exceed the stripe wavelength , @xmath166 .
electron heating in this regime has been studied using kinetic simulations by @xcite . forced reconnection of the striped wind has also been examined in a regime where post - shock kinetic scales remain smaller than the stripe separation , in both kinetic @xcite and hydromagnetic @xcite settings . here
, we analyze the post - shock forced reconnection using our model , even though turbulent dissipation is probably not an appropriate treatment when the eddy scale is microscopic .
still , this approach extends the analysis of @xcite , by resolving the forced reconnection zone .
this approach also extends the analyses of @xcite , which do resolve the post - shock reconnection zone , but do not account for the flow s spherical geometry or its macroscopic evolution throughout the nebula .
our aim is to determine the magnetization and radiative efficiency of the nebular flow , at latitudes where stripes may be fully dissipated .
this applies to the equatorial belt in the crab nebula , or at all latitudes in a hypothetical nebula powered by an orthogonally rotating pulsar .
we choose a set of wind parameters , @xmath159{erg / s}$ ] , @xmath167{s^{-1}}$ ] , and @xmath161 , intended to illustrate extreme electron heating and photon production in the forced reconnection .
such small mass - loading implies that the each particle must bear a relatively greater portion of the wind power , such that the post - shock temperature and thus synchrotron frequency is higher .
the freely expanding wind is assumed to dissipate in the fast reconnection regime described in section [ sec : pre - shock - stripe - dissipation ] , because for these parameters , @xmath138 remains small ( @xmath168 ) out to the shock . at the distance @xmath169{cm}$ ] , we solve the kennel - coroniti jump conditions ( see equation [ eqn : kc - jump - conditions ] ) and continue integrating the solution on the downstream side of the shock .
there , @xmath170 increases dramatically such that @xmath138 goes through unity after a few wavelengths @xmath171{cm}$ ] .
once @xmath154 we switch to the turbulent relaxation model .
a number of points are illustrated by the solution , shown in figure [ fig : postshockstripedwind ] .
first , dissipation ( and thus deceleration ) occur rapidly on the downstream side of the shock , even though turbulent relaxation is slowed by the growth of @xmath17 . since we have assumed zero residual ( stripe - averaged ) magnetic field , reconnection proceeds toward arbitrarily small values of the magnetic field .
this has two immediate consequences ( 1 ) that the flow decelerates well below the nebula expansion velocity @xmath172{km / s}$ ] , such that the outer boundary condition can not be satisfied , and ( 2 ) that the magnetic field at large distance lies in the range of @xmath173{g}$ ] , much weaker than what is thought to exist in the crab nebula , @xmath174{g}$ ] .
thus , orthogonal rotation of the crab pulsar is incompatible with our dissipation model .
if a residual magnetic field is included , may been implemented by replacing equation [ eqn : turbulent - relaxation ] with @xmath175 .
such solutions have been examined but are not shown here , because they are identical , in the context of the striped wind , to those obtained with the dissipative shock jump condition of @xcite .
] then only that part remains after a short distance beyond the shock .
a residual field might well be interpreted as the magnetization invoked by @xcite to match the nebula expansion . however
, any such residual field at a given latitude should also be prone to reconnection by mixing across the nebula equator or succumbing to kink instabilities at high latitude .
we analyze relaxation of the large - scale residual magnetic field in section [ sec : mean - field ] .
figure [ fig : postshockstripedwind ] illustrates a number of points related to synchrotron radiation from the pulsar wind .
first , effectively no radiation comes from the freely expanding wind ; the synchrotron cooling time for streaming elections is vastly longer than their adiabatic cooling time .
this point is already well appreciated , and known empirically from the crab s `` underluminous '' zone ( e.g. * ? ? ?
second , electrons are only heated up to @xmath176{ev}$ ] by the shock itself , but may be further heated ( at least for the low mass - loading chosen here ) up to @xmath177{pev}$ ] in the forced reconnection immediately downstream .
the associated photon energies reach to the mev range . however , the magnetic field dissipates at short range to much lower values , and only @xmath178 of the pulsar luminosity is ultimately converted into @xmath1-rays .
the wind parameters chosen for this example are intentionally chosen to represent an extreme case of electron heating in the forced reconnection . for more realistic parameters ,
say @xmath179 , the post - shock electron temperatures are typically in the tev range , with associated photon energies in the uv to soft x - ray . however , the choice of wind parameters has little bearing on the profile of flow velocity and magnetic field strength at distances larger than @xmath180{cm}$ ] .
in particular , we have not identified any relevant wind parameters which yield a realistic radiative efficiency , again because the short - wavelength oscillating magnetic field dissipates too quickly ; producing the high radiative efficiency inferred for the crab requires a residual , or at least slowly dissipating component of the magnetic field . in the next section , we show that our turbulent relaxation model can match the outer boundary condition and radiative efficiency if applied to the large - scale toroidal magnetic field , rather than the striped wind .
we now turn to the question of how the large - scale toroidal magnetic field relaxes and dissipates in the volume of the nebula .
this analysis intends to characterize the post - shock flow at moderate to high latitudes , where the oscillating magnetic field is small or zero .
we thus evolve the freely expanding flow without any dissipation , while the post - shock flow is evolved using the turbulent relaxation prescription .
we fix the wind luminosity @xmath181{erg / s}$ ] , which now stands for power supplied to the nebula in the form of non - oscillating magnetic field , and should thus , strictly speaking , be smaller by a factor of @xmath182 ( * ? ? ?
* assuming @xmath183 obliquity ) than the pulsar spin - down power .
the radiative efficiency is set to a nominal value of @xmath184 , which is marginally higher than current estimates in the range of @xmath8 @xcite .
the eddy scale near the inner edge of the nebula , @xmath185 is left as a free parameter , to be determined by appropriate matching of the outer boundary condition .
the corresponding rate @xmath186 may be interpreted loosely as the growth rate of kink instabilities operating on the toroidal field around nebula s x - ray core @xcite . at larger distances
the dissipation time - scale will increase . if the model is realistic , the volume average @xmath187 will be close to @xmath188{yr}$ ] , estimated by @xcite from calorimetric considerations .
we identify a family of wind solutions , one for each value of @xmath189 , that matches the nebula expansion speed @xmath190{km / s}$ ] and radiative efficiency @xmath184 .
we do this by numerically integrating 144 wind solutions , tabulated on an evenly spaced @xmath191 grid in @xmath192 and @xmath193 .
one such grid is generated for each of 30 different values of @xmath189 .
on each grid , [ @xmath194 , @xmath195 is recorded at all lattice points @xmath196 .
we then prolong the grid by a factor of 8 in both variables using third - order splines .
next we construct a set of level surfaces @xmath197 , @xmath198 $ ] , parameterized in arc - length @xmath199 , along which @xmath194 , @xmath200 take on constant values .
finally , we determine the intersection between the level surfaces by minimizing the distance function @xmath201 ^ 2 + [ y_u(s_u ) - y_\varepsilon(s_\varepsilon)]^2}$ ] over ( @xmath202 , @xmath203 ) .
figure [ fig : expradcontourplot ] shows an example of one of the solution grids . of the post - shock flow in the crab nebula , as a function of the wind magnetization @xmath0 and particle production rate @xmath204{s^{-1}}$ ] .
the color indicates the fraction of pulsar spin - down power that has been converted into synchrotron photons once the flow reaches the nebula boundary at @xmath205{pc}$ ] .
dashed contours are the level surfaces of radiative efficiency @xmath200 .
the solid contours represent level surfaces of @xmath194 , the nebula expansion speed at its boundary .
the open circle at the intersection of the two heavier contours indicates the values of @xmath0 and @xmath206 , that are favored by the turbulent relaxation model , given the fiducial values @xmath184 and @xmath190{km / s}$].,width=340 ] between 10 and 697 , but that all satisfy the nebula boundary condition @xmath190{km / s}$ ] and have a radiative efficiency @xmath184 .
shown , from top to bottom , is the four velocity @xmath73 , photon energy @xmath207 , electron gyro - radius @xmath107 , magnetization @xmath43 , magnetic field strength @xmath155 , and local dissipation time - scale @xmath4 , all as a function of distance downstream of the shock.,width=340 ] exploration of the parameter space reveals that the outer boundary condition and the radiative efficiency can only be matched simultaneously when @xmath189 is quite near ( within @xmath208 of ) @xmath209{cm}$ ] , and when @xmath210{s^{-1}}$ ] to within a factor of @xmath211 . on the other hand ,
the pre - shock value of @xmath0 is found to be rather sensitive to @xmath189 , ranging from @xmath212 when @xmath213{cm}$ ] to @xmath214 when @xmath215{cm}$ ] .
figure [ fig : expradcontourplot ] shows an example solution grid for which @xmath216{cm}$ ] .
the level surfaces of @xmath194 and @xmath200 intersect at @xmath217{s^{-1}}$ ] and @xmath218 .
radiative efficiency increases with lower mass - loading because each particle must bear a greater portion of the wind s dissipated magnetic energy , and thus attains a higher thermal lorentz factor .
the nebula magnetic field strength is found to be relatively insensitive to @xmath206 .
figure [ fig : solutionpanels ] shows six of the matching solutions for different values @xmath189 , labeled for the value of @xmath219 that satisfies the boundary conditions .
the different solutions are largely distinguished by the post - shock speed , which for the higher magnetization value remains at @xmath220 over @xmath221{cm}$ ] beyond the shock .
for all solutions , the electron gyro - radius @xmath107 ( and indeed the skin depth , not shown ) is larger than the oscillating magnetic field wavelength , suggesting that any stripes would dissipate without energizing non - thermal electrons @xcite .
however , the kinetic scales remain everywhere much smaller than the eddy scale , indicating that mhd turbulent relaxation is an appropriate description of the mean field dissipation .
furthermore , non - thermal particle acceleration may be successful in the turbulent reconnection , since there the magnetic fluctuation scale is macroscopic .
solutions are indistinguishable in the profile of magnetic field strength , which remains between 100 and @xmath222{\mu g}$ ] out to @xmath223{cm}$ ] and then declines to @xmath224{\mu g}$ ] by the nebula s outer edge .
the volume average of the nebula magnetic field is very nearly @xmath225{\mu g}$ ] in all our models , about half that required by the one - zone model of @xcite .
the photon energy peaks at @xmath223{cm}$ ] in all models , and varies from @xmath226{ev}$ ] as @xmath219 increases ( and @xmath206 decreases ) .
the crab spectrum peaks at around @xmath227{ev}$ ] , so mass loadings at the lower end of our parameter range , @xmath228 , are weakly favored .
electron temperatures are from @xmath229{gev}$ ] around @xmath230{cm}$ ] , and the gyro - radii are from @xmath231{cm}$ ] .
meanwhile , the turbulent eddy scale ( not shown in the figure ) increases from @xmath232 by a factor of roughly 10 , to @xmath233 at the outer edge , so the relaxation picture becomes marginally inapplicable there .
the dissipation time - scale varies from about 4 months near the nebula core , to over 100 years at the outer edge .
the volume average is @xmath234 years for all models , slightly lower than the 80-year dissipation time scale determined in @xcite .
we have developed a model for turbulent magnetic reconnection in a pulsar wind nebula .
the model is based on stationary , one - dimensional mhd flow which embeds a freely relaxing , small - scale magnetic field .
such characterization of the internal magnetic field allows for a dissipation model that is supported by recent developments in the theory of relativistic magnetic reconnection and turbulent relaxation .
we examined separately dissipation of the striped wind in both the freely expanding and post - shock flow , as well as turbulent relaxation of the mean magnetic field beyond the termination shock . applied to the crab nebula
, this model reproduces the expansion speed , radiative efficiency , and peak photon frequency within a one - parameter family of solutions , and without invoking unrealistically low values of the upstream magnetization , as required in the historical non - dissipative models .
our model strongly favors a pulsar particle production rate @xmath235{s^{-1}}$ ] , and a magnetic thermalization time scale which evolves from roughly 4 months ( the light - travel time of the wind termination shock radius ) near the nebula core , to 100 years near its outer edge .
such slowing of the turbulent cascade rate with distance is not put in by hand , but results from scaling laws derived from three - dimensional numerical simulations of turbulent magnetic relaxation @xcite .
we have also introduced a simple formalism for incorporating optically thin synchrotron losses into steady - state mhd winds .
though not critical in the dynamics of the crab pulsar wind , we will point out in section [ sec : gamma - ray - revshock ] that the radiative losses may dominate the reverse shock dynamics in the striped wind of a `` millisecond magnetar '' .
the model we have developed here is overly simplistic in a number of ways .
first , we do not account for any non - spherical geometry . in this approximation ,
the equations of motion and the shock jump conditions are formally valid in transverse mhd only at the equator .
more realistic , two - dimensional shock jump conditions have been analyzed in connection with the crab s inner knot feature @xcite .
also , in treating the radiative losses we assumed that particles were mono - energetic , whereas @xcite produced integrated spectra by assuming that particles occupy a power - law in energy . in principle , we could have done the same here , but that would require a particle spectral index to be selected by hand .
given that the magnetic field strength and coherence scale are both predicted by the model , the particle spectrum could be truncated at the energy where gyro - radii would exceed the local eddy scale .
a more sophisticated treatment along these lines may be pursued in a future work .
the crab s @xmath1-ray flares @xcite have been widely interpreted as the signature of exceptionally powerful reconnection episodes @xcite .
reconnection is most promising to explain flares if it occurs where the plasma is strongly magnetized . as pointed out by @xcite , such a region
is expected to occur in the @xmath3-pinch at high latitudes . in his model ,
the core of the pinch would be of the order @xmath236{cm}$ ] , roughly the scale of the emitting region as implied by the @xmath237{hr}$ ] flare duration . despite the rather artificial geometry of our own model
, it also predicts a strongly magnetized region region , @xmath238 of scale @xmath239{cm}$ ] , when the upstream magnetization is high .
interestingly , in that region ( just past the termination shock , see figure [ fig : solutionpanels ] ) plasma moves radially outward with a lorentz factor @xmath240 .
the implied doppler beaming could help account for the spectral cutoff , which for some flares exceeds the radiation reaction limit @xmath241{mev}$ ] ( e.g. * ? ? ?
* ) . in our model ,
the post - shock flow is relativistic and strongly magnetized only where stripes are negligible . where they dominate , the magnetization decreases and the flow decelerates immediately beyond the shock in the forced reconnection ( see figure [ fig : postshockstripedwind ] )
this may still be consistent with a scenario discussed in @xcite , where the flaring region lies marginally upstream ( or perhaps within the kinetic structure ) of the termination shock , where the radial lorentz factor remains high .
we hope that the model developed here will find utility in multi - zone modeling of particle transport and acceleration in pwne
. this could be useful in modeling of pwne for which multi - band , spatially resolved observations are available , such as msh 15 - 52 ( e.g. * ? ? ?
* ) and 3c 58 , not to mention the crab . it could also help determine the rate of positron leakage from the nebula , and help assess the importance of pwne as galactic positron sources .
unlike one - zone models @xcite for which the spectral energy distribution of injected electrons is a model parameter , a multi - zone treatment could possibly predict the spectrum from humbler assumptions .
for example , one might solve an advection - diffusion equation , where the advective part is given by the flow velocity @xmath73 , and the spatial diffusion by the eddy scale @xmath17 and fluctuating velocity @xmath132 .
the latter yields a turbulent diffusivity , while the local specific heating rate @xmath36 may be used to normalize energy gains by the electron population . this may be taken up in a future work . rapidly rotating ,
strongly magnetized neutron stars have been invoked in connection with a variety of observed and hypothetical astrophysical transients .
many @xmath1-ray bursts , both long and short , exhibit an extended x - ray emission plateau , which has been widely interpreted as the signature of sustained energy injection by a nascent pwn . generally , one imagines the nebula to be confined by a massive shell , either of stellar ejecta in the case of long bursts and superluminous supernovae , or material dynamically expelled from a binary neutron star merger in the case of short bursts . in either scenario ,
the resulting emission is expected to be more isotropic than the @xmath1-ray burst itself , and is thus interesting as a possible electromagnetic counterpart to gravitational wave detections of binary neutron star coalescence .
conditions in these hypothetical pwne are qualitatively distinct in a couple of ways from those of the crab nebula .
the radiative energy density @xmath158 in the nascent nebula could be so high as to initiate a cascade of @xmath242 pair production , so that mass conservation @xmath243 does not apply .
as a result , the plasma density may be high enough to trap synchrotron photons , so that @xmath244 even when @xmath245 .
pair production is robust when the nebula compactness parameter @xmath246 is high . as we saw in section
[ sec : post - shock - stripe - dissipation ] the pre - shock flow is cold and non - radiative , so pair production should only commence in the region of width @xmath247 between the reverse shock and the confining shell . equation
[ eqn : cp ] indicates that if some photons were to leak out from the nebula , then pair production would be suppressed for two reasons .
the first , of course , is that leakage reduces @xmath248 and thus @xmath249 directly .
the second is that loss of radiation pressure in the shocked plasma allows the reverse shock to advance toward the shell , reducing @xmath247 .
this further decreases the nebula optical depth , meaning that a slightly `` deflated '' nebula only becomes more leaky . of course
, the details of radiation leakage from nascent pwne are beyond the present scope . nevertheless , we make two observations here that could motivate future more detailed calculations .
first , we point out that under conditions relevant to binary neutron star mergers ( those yielding a millisecond magnetar ) , the synchrotron efficiency of the reverse shock can go up to @xmath9 , with all the radiation produced above @xmath250{mev}$ ] .
photon losses may become relevant at such high energies because klein - nishina effects reduce the optical depth of the shocked plasma and confining shell .
also , the relevant shell albedo would be that of @xmath1-rays , while x - ray albedo has been adopted in earlier analyses @xcite .
second , we argue that the pair cascade might in some cases appear intermittently or not at all .
consider the special case of a stable millisecond magnetar formed in a binary neutron star merger .
we adopt nominal source parameters @xmath251{erg / s}$ ] , @xmath252{s^{-1}}$ ] , and @xmath253 , and place the ejecta at @xmath254{hr } \approx \unit[10^{13}]{cm}$ ] , moving outwards at @xmath255{km / s}$ ] .
now suppose the pulsar is an orthogonal rotator , so that its electromagnetic power is all in the striped wind as we analyzed in section [ sec : striped - wind ] .
with those parameters fixed , vary the shock location until the flow matches smoothly onto the ejecta .
if the plasma is optically thick , then synchrotron losses should be ignored as photons are trapped and contribute to the gas pressure with the same equation of state , @xmath256 as the plasma particles . in this case , the shock is found to lie about half way to the ejecta shell ( similar to figure [ fig : postshockstripedwind ] ) , so the assumption @xmath257 in calculating @xmath249 ( e.g. * ? ? ?
* ) appears to be well justified . however , unlike the result shown in figure [ fig : postshockstripedwind ] for crab parameters , the synchrotron frequency is @xmath258{mev}$ ] , and the efficiency @xmath158 in the forced reconnection zone is essentially 100% .
accounting for the klein - nishina suppression , @xmath259 at @xmath250{mev}$ ] , @xmath260 is found to be @xmath168 .
therefore , synchrotron photons produced at the reverse shock traverse the nebula , and their fate is determined by the albedo and optical depth of the shell . if insufficient radiation were reflected back to the nebula , then @xmath158 and thus @xmath249 would be small , and the pair cascade could fail .
of course , the preceding analysis ignores the enhancement of @xmath261 brought on by a pair cascade .
if one operates , then once again @xmath262 , photons are trapped , radiation pressure forces the reverse shock to recede inwards , the compactness parameter is kept large , and pair production continues .
but this illustrates the point made previously , that the pair cascade depends on itself to survive , and is thus a complex problem .
unraveling the non - linear physics may require a detailed calculation of fully coupled magnetic reconnection , synchrotron radiation , photon propagation , and pair production processes .
it is difficult to say whether such a pursuit would be fruitful in the present context .
nevertheless , similarly complex relativistic plasma systems were examined by @xcite , and also by @xcite with interesting applications to pulsars and @xmath1-ray burst prompt emission , respectively . continued research along these lines
was advocated for by @xcite .
j. zrake acknowledges yajie yuan , roger blandford , andrew macfadyen , dan kasen , andrei beloborodov , brian metzger , and lorenzo sironi for inspiring discussions .
natexlab#1#1 abdo , a. a. , ackermann , m. , ajello , m. , et al .
2011 , science ( new york , n.y . ) , 331 , 739 an , h. , madsen , k. k. , reynolds , s. p. , et al .
2014 , the astrophysical journal , 793 , 90 arnold , v. 1965 , comptes rendus hebdomadaires des seances de lacademie des sciences , 261 , 17 arons , j. 2002 , the astrophysical journal , volume 589 , issue 2 , pp .
871 - 892 . , 589 , 871 begelman , m. c. 1998 , the astrophysical journal , 493 , 291 begelman , m. c. , & li , z .- y .
1992 , the astrophysical journal , 397 , 187 . 1994 , the astrophysical journal , 426 , 269 beloborodov , a. m. 2016 , eprint arxiv:1604.02794 , arxiv:1604.02794 brandenburg , a. , kahniashvili , t. , & tevzadze , a. g. 2015 , physical review letters , 114 , 075001 bucciantini , n. , arons , j. , & amato , e. 2011 , monthly notices of the royal astronomical society , 410 , 381 buehler , r. , scargle , j. d. , blandford , r. d. , et al .
2012 , the astrophysical journal , 749 , 26 campanelli , l. 2016 , the european physical journal c , 76 , 1 cerutti , b. , werner , g. r. , uzdensky , d. a. , & begelman , m. c. 2012 , the astrophysical journal , 754 , l33 . 2014 , physics of plasmas , 21 , 056501 . 2014 , the astrophysical journal , 782 , 104 chi , x. , cheng , k. s. , & young , e. c. m. 1996 , the astrophysical journal , 459 , l83 chiueh , t. , li , z .- y . , & begelman , m. c. 1991 , the astrophysical journal , 377 , 462 clausen - brown , e. , & lyutikov , m. 2012 , monthly notices of the royal astronomical society , 426 , 1374 coroniti , f. v. 1990 , the astrophysical journal , 349 , 538 dessart , l. , hillier , d. j. , waldman , r. , livne , e. , & blondin , s. 2012 , monthly notices of the royal astronomical society : letters , 426 , l76 drenkhahn , g. 2002 , astronomy and astrophysics , 387 , 714 east , w. e. , zrake , j. , yuan , y. , & blandford , r. d. 2015 , physical review letters , 115 , 095002 finn , j. m. , & kaw , p. k. 1977 , physics of fluids , 20 , 72 frisch , u. , pouquet , a. , leorat , j. , & mazure , a. 1975 , journal of fluid mechanics , 68 , 769 gao , h. , ding , x. , wu , x .- f .
, zhang , b. , & dai , z .- g . 2013 , the astrophysical journal , 771 , 86 gelfand , j. d. , slane , p. o. , & zhang , w. 2009 , the astrophysical journal , 703 , 2051 giannios , d. , & spruit , h. c. 2007 , astronomy and astrophysics , 469 , 1 goldreich , p. , & julian , w. h. 1969 , the astrophysical journal , 157 , 869 . 1970 , the astrophysical journal , 160 , 971 harding , a. k. , stern , j. v. , dyks , j. , & frackowiak , m. 2008 , the astrophysical journal , 680 , 1378 hester , j. j. 2008 , annual review of astronomy and astrophysics , 46 , 127 hester , j. j. , mori , k. , burrows , d. , et al .
2002 , the astrophysical journal , 577 , l49 kasen , d. , & bildsten , l. 2010 , the astrophysical journal , 717 , 245 kasen , d. , metzger , b. d. , & bildsten , l. 2016 , the astrophysical journal , 821 , 36 kennel , c. f. , & coroniti , f. v. 1984 , the astrophysical journal , 283 , 694 . 1984 , the astrophysical journal , 283 , 710 kennel , c. f. , fujimura , f. s. , & okamoto , i. 1983 , geophysical and astrophysical fluid dynamics , 26 kirk , j. g. , & skjaraasen , o. 2003 , the astrophysical journal , 591 , 366 komissarov , s. s. 2012 , monthly notices of the royal astronomical society , 428 , 2459 liu , l. d. , wang , l. j. , & dai , z. g. 2016 , astronomy & astrophysics , 592 , a92 lyubarsky , y. , & kirk , j. g. 2001 , the astrophysical journal , 547 , 437 lyubarsky , y. , & liverts , m. 2008 , the astrophysical journal , 682 , 1436 lyubarsky , y. e. 2002 , monthly notices of the royal astronomical society , 329 , l34 . 2003 , monthly notices of the royal astronomical society , 339 , 765 . 2003 , monthly notices of the royal astronomical society , 345 , 153 . 2012 , monthly notices of the royal astronomical society , 427 , 1497 lyutikov , m. , komissarov , s. s. , & porth , o. 2016 , monthly notices of the royal astronomical society , 456 , 286 lyutikov , m. , sironi , l. , komissarov , s. , & porth , o. 2016 , arxiv:1603.05731 matthaeus , w. h. , & montgomery , d. 1980 , annals of the new york academy of sciences , 357 , 203 matthaeus , w. h. , montgomery , d. c. , wan , m. , & servidio , s. 2012 , journal of turbulence , 13 , n37 metzger , b. d. , giannios , d. , thompson , t. a. , bucciantini , n. , & quataert , e. 2011 , monthly notices of the royal astronomical society , 413 , 2031 metzger , b. d. , & piro , a. l. 2014 , monthly notices of the royal astronomical society , volume 439 , issue 4 , p.3916 - 3930 , 439 , 3916 metzger , b. d. , vurm , i. , hascoet , r. , & beloborodov , a. m. 2013 , monthly notices of the royal astronomical society , volume 437 , issue 1 , p.703 - 720 , 437 , 703 meyer , m. , horns , d. , & zechlin , h .- s .
2010 , astronomy & astrophysics , 523 , a2 michel , f. c. 1969 , the astrophysical journal , 158 , 727 . 1971 , comments on astrophysics and space physics , 3 . 1982 , reviews of modern physics , 54 , 1 . 1994 , the astrophysical journal , 431 , 397 mignone , a. , striani , e. , tavani , m. , & ferrari , a. 2013 , monthly notices of the royal astronomical society , volume 436 , issue 2 , p.1102 - 1115 , 436 , 1102 mizuno , y. , lyubarsky , y. , nishikawa , k .- i .
, & hardee , p. e. 2011 , the astrophysical journal , 728 , 90 nalewajko , k. , zrake , j. , yuan , y. , east , w. , & blandford , r. 2016 , astrophysical journal , 826 , doi:10.3847/0004 - 637x/826/2/115 porth , o. , komissarov , s. s. , & keppens , r. 2013 , monthly notices of the royal astronomical society : letters , 431 , l48 . 2013 , monthly notices of the royal astronomical society , 438 , 278 rees , m. j. , & gunn , j. e. 1974 , monthly notices of the royal astronomical society , 167 , 1 rybicki , g. b. , & lightman , a. p. 1979
, new york sironi , l. , & spitkovsky , a. 2011 , the astrophysical journal , 741 , 39 takamoto , m. , inoue , t. , & inutsuka , s .- i .
2012 , the astrophysical journal , 755 , 76 tavani , m. , bulgarelli , a. , vittorini , v. , et al .
2011 , science ( new york , n.y . ) , 331 , 736 taylor , j. b. 1974 , physical review letters , 33 , 1139 timokhin , a. n. , & arons , j. 2013 , monthly notices of the royal astronomical society , 429 , 20 uzdensky , d. a. 2015 , magnetic reconnection , astrophysics and space science library , volume 427 .
isbn 978 - 3 - 319 - 26430 - 1 .
springer international publishing switzerland , 2016 , p. 473
, 427 , 473 uzdensky , d. a. , cerutti , b. , & begelman , m. c. 2011 , the astrophysical journal , 737 , l40 woosley , s. e. 2010 , the astrophysical journal letters , 719 , l204 yuan , y. , & blandford , r. 2015 , monthly notices of the royal astronomical society , volume 454 , issue 3 , p.2754 - 2769 , 454 , 2754 yuan , y. , nalewajko , k. , zrake , j. , east , w. e. , & blandford , r. d. 2016 , the astrophysical journal , 828 , 92 zhang , b. 2013 , the astrophysical journal letters , 763 , l22 zrake , j. 2014 , the astrophysical journal , 794 , l26 . 2016 , the astrophysical journal , 823 , 39 zrake , j. , & east , w. e. 2016 , the astrophysical journal , 817 , 89 zrake , j. , & macfadyen , a. i. 2013 , the astrophysical journal , 769 , l29
coherent magnetic structures ( eddies , or flux tubes ) grow over time by merging with one another as a result of coalescence instability ( e.g. * ? ? ?
* ; * ? ? ?
* ) , and they also grow or shrink due to the expansion or compression of comoving volume . for the latter reason it is convenient to utilize the eddy mass @xmath128 as a proxy for its scale . in three dimensions ,
magnetic structures are cylindrical flux tubes , having mass @xmath263 ( assuming their length and radius are comparable ) .
the flux tubes are locally relaxed equilibria and thus generally helical @xcite , having comparable axial and azimuthal field strength .
when net magnetic helicity is zero , left and right polarized flux tubes exist in equal number .
pairs can join by reconnecting their azimuthal field when their axial electric current vectors are parallel , but they only form a stable structure when their axial magnetic fields are also parallel ; otherwise the axial field annihilates and renders the merged flux tube kink - unstable
. thus only half of the merging episodes ( those occurring between like - polarized tubes ) yield stable structures , and so the total number of eddies @xmath199 decreases by a factor of four in each stage of coalescence , @xmath264 .
stages proceed at the cascade rate @xmath265 where @xmath124 is given roughly by the local alfvn speed @xmath266 . each merging event conserves mass and magnetic helicity , so we have @xmath267 and @xmath268 . meanwhile , the magnetic energy per eddy drops according to @xmath269 . the volume - averaged magnetic energy per unit mass
is given by @xmath270 . here
, the cascade number @xmath271 is a discrete version of the variable @xmath138 used throughout the paper .
when expansion is neglected ( @xmath272 ) , this heuristic yields a decay law in which the characteristic eddy scale increases over time as @xmath273 , and the magnetic energy decays as @xmath274 . such behavior was seen by @xcite in simulations of freely decaying , non - helical relativistic mhd turbulence in three dimensions .
however , the heuristic s applicability to an expanding volume has not yet been explored directly with numerical simulations .
arguably , one expects the eddy mass to be a good proxy for its scale whenever the turbulence cascade rate @xmath275 is fast compared with the secular evolution . comoving volume in the pulsar wind expands at a rate @xmath276 in the transverse direction , and at @xmath277 in the longitudinal direction .
if either of the expansion rates was faster than @xmath278 , the magnetic field pattern would become frozen into the flow , and get stretched in whichever direction expands faster .
this situation corresponds to the magnetic free energy scale exceeding that of the local horizon . provided
the cascade is faster than both expansion rates , eddies maintain their isotropy in the comoving frame , and use of the mass coordinate in place of scale is well justified .
here we review the conditions under which the plane - parallel reconnection picture of @xcite is equivalent to the turbulent relaxation model . near the pulsar , the striped wind consists of cold , well - magnetized plasma .
the azimuthal magnetic field alternates in direction every half - wavelength @xmath136 .
magnetic reconnection operates around the reversals , causing slabs of hot plasma to expand at a speed @xmath279 .
the hot regions occupy a fraction @xmath138 of the wind volume , and so @xmath280 is the surviving fraction of the wind s magnetic energy .
consider a cold plasma volume which is bounded by current layers centered at @xmath281 . as the current layers expand , the one centered at @xmath282 extends forward to @xmath283 , while the one centered at @xmath284 extends backward to @xmath285 , so we have @xmath286 now , @xmath287 and to linear order @xmath288 , where @xmath289 and @xmath290 here denotes time in the pulsar frame .
the velocity gradient is given by @xmath291 where @xmath292 is the acceleration rate .
the reconnection fronts advance according to the relativistic velocity addition of the flow and @xmath124 , @xmath293 where @xmath294 .
we also assume that @xmath295 . since the wind is ultra - relativistic
, we expand @xmath296 in powers of @xmath140 , @xmath297 \gamma^{-2 } + \mathcal{o}(\gamma^{-3})\right\ } \ , .
\label{eqn : delta - dot - full}\ ] ] further dropping terms of order higher than @xmath298 , we arrive at @xmath299 which is the same as equation [ eqn : reconnection - front - speed ] , and equivalent to equation c3 of @xcite . the second order term in equation [ eqn : delta - dot - full ] inhibits the progress of reconnection fronts by stretching the background flow , but remains small provided @xmath300 , where @xmath301 is the local horizon scale with respect to the reconnection speed .
sticking to first order in @xmath140 is thus sufficiently accurate unless reconnection were somehow to proceed ultra - relativistically .
the kennel - coroniti jump conditions are given by conservation of mass , magnetic flux , energy , and momentum across the shock , @xmath302 where subscripts 1 and 2 refer to values just ahead of , and just behind the shock respectively .
together with the @xmath44-law equation of state , these yield the following equation for the density jump @xmath303 across the shock , @xmath304
|
we present a model for magnetic energy dissipation in a pulsar wind nebula .
better understanding of this process is required to assess the likelihood that certain astrophysical transients may be powered by the spin - down of a `` millisecond magnetar . ''
examples include superluminous supernovae , gamma - ray bursts , and anticipated electromagnetic counterparts to gravitational wave detections of binary neutron star coalescence .
our model leverages recent progress in the theory of turbulent magnetic relaxation to specify a dissipative closure of the stationary magnetohydrodynamic ( mhd ) wind equations , yielding predictions of the magnetic energy dissipation rate throughout the nebula .
synchrotron losses are treated self - consistently . to demonstrate the model s efficacy ,
we show that it can reproduce many features of the crab nebula , including its expansion speed , radiative efficiency , peak photon energy , and mean magnetic field strength . unlike ideal mhd models of the crab ( which lead to the so - called @xmath0-problem ) our model accounts for the transition from ultra to weakly magnetized plasma flow , and for the associated heating of relativistic electrons .
we discuss how the predicted heating rates may be utilized to improve upon models of particle transport and acceleration in pulsar wind nebulae .
we also discuss implications for the crab nebula s @xmath1-ray flares , and point out potential modifications to models of astrophysical transients invoking the spin - down of a millisecond magnetar .
|
smooth - particle hydrodynamics ( sph ) is a lagrangian method for modelling fluid dynamics , pioneered by @xcite and @xcite . instead of discretising the fluid quantities , such as density , velocity , and temperature , on a fixed grid as in eulerian methods ,
the fluid is represented by a discrete set of moving particles acting as interpolation points . due to its lagrangian nature , sph models regions of higher density with higher resolution with the ability to simulate large dynamic ranges .
this makes it particularly useful in astrophysics , where it is used to model galaxy and star formation , stellar collisions , and accretion discs .
the core of sph is the kernel estimator : the fluid density is _ estimated _ from the masses @xmath1 and positions @xmath2 of the particles via to denote a local _ estimate _ in many sph - related publications the distinction between actual and estimated quantities is not clearly made , confusing the discussion . ]
@xmath3 where @xmath4 is the kernel function and @xmath5 the sph smoothing length for @xmath6 . ] for the @xmath7th particle .
similar estimates @xmath8 for the value of any field @xmath9 can be obtained from its discretised values @xmath10 . by applying these estimators to the fluid equations governing mass , momentum and energy , discrete equations for the sph particle positions @xmath2 and other properties ( such as internal energy @xmath11 ) can be obtained .
together with an appropriate time integration method , these constitute a concrete sph scheme .
unfortunately , this process is not unique and since its inception the sph method has undergone many refinements such as individual particle smoothing lengths and viscosities , as well as many alternative derivations of the sph equations , leading to a plethora of sph methods .
while formally these various schemes differ only in their error terms , their conservation and stability properties can be quite different .
this has lead to the unfortunate situation that the shortcomings of a few such implementations are often blamed on the general sph concept per se .
however , @xcite have pointed out that sph equations derived from a variational principle are not only unique , but also conservative .
such sph equations are most simply obtained as the euler - lagrange equations derived from an sph lagrangian @xmath12 representing the lagrangian of the fluid system .
once @xmath12 is chosen , the sph equations follow uniquely ( see appendix [ app : pressure ] for a typical example ) . complementing these with a symplectic integrator , such as the standard leap - frog , results in a sph scheme which by construction conserves the total mass , momentum , angular momentum , energy , and entropy .
the conservation of entropy means that sph is _ dissipation - less _ , as demonstrated in fig .
[ fig : sound_wave ] . in real fluids , however , entropy increases in shocks , where particle collisions randomise their velocities generating heat and entropy .
this basic collisional mechanism is inherent to all fluids ( except for dust and collisionless plasma , which therefore may not be considered fluids ) and prevents the flow from becoming multi - valued . in sph
_ artificial _ viscosity is needed to dissipate local velocity differences and convert them into heat , which generates entropy and prevents inter - penetration of sph particles and thus a multi - valued flow .
since the artificial viscosity required for this goal is usually much stronger than the actual physical viscosity , it also causes unphysical dissipation away from shocks .
while it may be possible for certain simulations to select the magnitude of the viscosity to minimise such undesired dissipation , in general the adverse effect of artificial viscosity is unknown prior to any simulation and , possibly , even afterwards .
for example , when simulating the effect of a perturbing massive body on a pulsating star , it may be very difficult to distinguish this effect from that induced by artificial viscosity .
another example is the case of a differentially rotating disc , where artificial viscosity causes spurious angular momentum transport .
since viscosity is a dissipative process , the corresponding sph equations can not be derived from a variational principle , and we are back to ad - hoc methods for deriving them .
most sph simulations to date still use a rather simple artificial viscosity , which effectively amounts to modelling a viscous fluid and quickly damps away any oscillations , such as sound waves or stellar pulsations , and impedes shear flows .
while suggestions have been made to _ reduce _ such unwanted dissipation , our goal here is to _ eliminate _ it . to this end
we introduce a novel method of controlling the amount of artificial viscosity , such that away from shocks the modelled flow is virtually inviscid .
section [ sec : av ] describes sph artificial viscosity and previous efforts to reduce its adverse effects , while our new method is outlined in section [ sec : av : novel ] .
the ability of the new scheme to reduce artificial viscosity but also to capture shocks is demonstrated in sections [ sec : test : suppress ] and [ sec : test : shock ] , respectively .
finally , section [ sec : summ ] concludes our study .
the traditional form of artificial viscosity ( e.g. * ? ? ? * ) adds the following terms to the momentum and energy equations , allowing the conversion of kinetic energy into heat .
@xmath13 with the average kernel @xmath14 here , @xmath15 and @xmath16 , while @xmath5 is the individual adaptive smoothing length of each sph particle ( for details on how @xmath5 is adapted see appendix [ sec : adapt ] ) .
the artificial viscosity term is given by @xcite @xmath17 with @xmath18 ( @xmath19/2 $ ] and likewise for the average sound speed @xmath20 and estimated density @xmath21 ) . since @xmath22 for receding particle pairs ,
artificial viscosity does not affect expanding flows .
this functional form of sph artificial viscosity may seem rather ad - hoc , but it is reasonably well motivated and emerged as the most useful one amongst several methods @xcite .
moreover , it is equivalent to the form of dissipation implicit in riemann solvers @xcite . by expanding density and velocity in a taylor series around @xmath2 , it is straightforward to show that these terms correspond to both a shear and a bulk viscosity .
more quantitatively , if one assumes that , other than in equation ( [ eq : pi ] ) , artificial viscosity acts between approaching and receding neighbours and that @xmath23 , the corresponding shear and bulk viscosity coefficients are ( e.g. * ? ? ?
* ) @xmath24 and @xmath25 , respectively , where the factor @xmath26 is of order unity and depends on the functional form of the kernel .
this implies that artificial viscosity decreases with increasing resolution ( smaller @xmath27 ) .
thus , a straightforward though expensive way to reduce unwanted dissipation is to increase the resolution .
in fact , one motivation for reducing artificial viscosity is to avoid this purely numerical necessity for high resolution .
most sph applications to date use the above treatment with @xmath28 .
the widely used code gadget-2 @xcite employs a fixed @xmath29 chosen at the start of the simulation ( though @xcite , @xcite , have implemented into gadget-2 the improved method described in [ sec : av : mm ] below ) .
clearly , in complex situations , where strong and weak shocks are present as well as converging flows , any choice for @xmath29 is unsatisfactory , leading to bad treatment of strong shocks , over - damping of converging flows , or both .
the purpose of artificial viscosity is to allow for entropy generation across shocks and to stop particle interpenetration . to this end
, only bulk viscosity is required , but the inherent shear viscosity is unnecessary .
what is worse , this shear viscosity may seriously compromise simulations of shear flows , such as in a differentially rotating gas disc . in an effort to reduce the resulting artificial angular momentum dissipation
, @xcite proposed to multiply @xmath30 with a reduction factor @xmath31 with @xmath32 ( with velocity divergence and curl estimated using the sph kernel estimator ) .
this term diminishes the effect of artificial viscosity whenever the vorticity dominates the convergence .
however , this method only reduces ( but does not eliminate ) unwanted dissipation in the presence of a rotating shear flow .
standard sph artificial viscosity acts whenever the flow of the fluid converges , even if only weakly .
for example , when a pulsating star contracts artificial viscosity damps its pulsation .
exactly the same happens to ordinary sound waves : standard sph viscosity damps them , as demonstrated in fig .
[ fig : damp_sound ] , the faster the shorter the wave length ( because these are more poorly resolved ) . with this in mind
, @xcite proposed to adapt the strength of artificial viscosity to the local convergence of the flow . to this end , they introduced the concept of individual adaptive viscosities @xmath33 for each particle , replaced @xmath29 in equation ( [ eq : pi ] ) by @xmath34 , and set @xmath35 .
the individual viscosities are adapted according to the differential equation @xmath36 with the velocity - based source term @xmath37 and the decay time ) accounts for the difference in the definition of the smoothing length @xmath27 between us and @xcite . ]
@xmath38 here , @xmath39 constitutes a lower limit for the artificial viscosity such that @xmath40 for non - convergent flows . for a convergent flow , on the other hand , @xmath33 grows above that value , guaranteeing the proper treatment of shocks . in the post - shock region ,
the flow is no longer convergent and @xmath33 decays back to @xmath41 on the time scale @xmath42 ( typically @xmath43 ) .
this method reduces the artificial viscosity away from shocks by an order of magnitude compared to standard sph and gives equally accurate post and pre - shock solutions @xcite .
more recently , @xcite proposed to alter the adaption equation ( [ eq : davdt ] ) to ) but multiplying the source term ( [ eq : s : mm ] ) by @xmath44 , which is what @xcite actually did.]@xmath45 with @xmath46 , while @xcite advocated @xmath47 .
the effect of this alteration is first to prevent @xmath33 to exceed @xmath48 and second to increase @xmath49 for small @xmath33 , which ensures a faster viscosity growth , resulting in somewhat better treatment of shocks @xcite .
this method may also be combined with the balsara switch by applying the reduction factor ( [ eq : balsara ] ) either to @xmath30 @xcite or to @xmath50 @xcite .
the scheme of equations ( [ eq : s : mm ] ) , ( [ eq : tau ] ) and ( [ eq : davdt : alt ] ) with @xmath39 , @xmath47 and @xmath51 is the current state of the art for sph and is implemented in the codes phantom ( by daniel price ) and vine @xcite . in sections [ sec :
test : suppress ] and [ sec : test : shock ] , we will frequently compare our novel scheme ( to be described below ) with this method and refer to it as the ` m&m method ' or the ` @xcite version of the m&m method ' as opposed to the ` original m&m method ' , which uses equation ( [ eq : davdt ] ) instead of ( [ eq : davdt : alt ] ) .
the m&m method certainly constitutes a large improvement over standard sph , but low - viscosity flows , typical for many astrophysical fluids , are still inadequately modelled . after studying this and related methods in detail , we identify the following problems .
first , any @xmath52 results in unwanted dissipation , for example of sound waves ( see fig . [ fig : damp_sound ] ) or stellar pulsations ( see [ sec : test : poly ] ) , yet the m&m method requires @xmath53 . this necessity has been established by numerous tests ( most notably of * ? ? ?
* ) and is understood to originate from the requirement to ` maintain order amongst the particles away from shocks ' @xcite .
second , there is a delay between the peak in the viscosity @xmath29 and the shock front ( see fig .
[ fig : source ] ) : the particle viscosities are still rising when the shock arrives .
one reason for this lag is that integrating the differential equation ( [ eq : davdt : alt ] ) increases @xmath33 too slowly : the asymptotic value @xmath54 is hardly ever reached before the shock arrives ( and @xmath50 decreases ) .
third , the source term ( [ eq : s : mm ] ) does not distinguish between pre- and post - shock regions : for a symmetrically smoothed shock it peaks at the exact shock position ( in practice the peak occurs one particle separation in front of the shock , * ? ? ?
* see also fig .
[ fig : source ] ) .
however , immediately behind the shock ( or more precisely the minimum of @xmath55 ) , the ( smoothed ) flow is still converging and hence @xmath29 continues to increase without need .
a further problem is the inability of the source term ( [ eq : s : mm ] ) to distinguish between velocity discontinuities and convergent flows .
finally , in strong shear flows the estimation of the velocity divergence @xmath55 , needed in ( [ eq : s : mm ] ) , often suffers from substantial errors ( see appendix [ app : divv : fail ] for the reason ) , driving artificial viscosity without need .
this especially compromises simulations of differentially rotating discs even when using the balsara switch .
our aim is a method which overcomes all the issues identified in [ sec : mm : crit ] above and in particular gives @xmath56 away from shocks . to this end
, we introduce a new shock indicator in [ sec : av : a ] , a novel technique for adapting @xmath33 in [ sec : av : new ] , and a method to suppress false compression detections due to the presence of strong shear in [ sec : shear ] .
we need a shock indicator which not only distinguishes shocks from convergent flows , but , unlike @xmath55 , also discriminates between pre- and post - shock regions .
this requires ( at least ) a second - order derivative of the flow velocity and we found the total time derivative of the velocity divergence , @xmath57 , to be most useful . as is evident from differentiating the continuity equation , @xmath58 @xmath59 indicates an non - linear density increase and a steepening of the flow convergence , as is typical for any pre - shock region .
conversely , in the post - shock region @xmath60 .
this suggests to consider only negative values and , in analogy with equation ( [ eq : s : mm ] ) , we define the new shock indicator @xmath61 here , @xmath62 is a limiter , detailed in [ sec : shear ] below , aimed at suppressing false detections of compressive flows in multi - dimensional flows . instead of increasing @xmath33 by integrating a differential equation , we set @xmath33 directly to an appropriate local value @xmath63 whenever this exceeds the current value for @xmath33 . after extensive experimenting
, we settled on the following simple functional form @xmath64 with the signal velocity . ] @xmath65 at the moment of passing through a shock ( more precisely through a maximum of the flow convergence ) , @xmath66 and hence @xmath67 return to zero and whenever @xmath68 we let @xmath33 decay according to @xmath69 we use @xmath70 rather than @xmath71 in the decay time @xmath42 for internal consistency ( this is of little practical relevance as @xmath72 in the post - shock region ) .
we use @xmath73 , such that the viscosity decays twice as slowly as in previous methods , avoiding some occasional minor post - shock ringing not present in methods with @xmath74 .
however , the traditional @xmath51 also gives satisfactory results for most of our test problems .
as explained in detail in appendix [ app : divv : fail ] , in multi - dimensional flows strong shear induces false detections of @xmath55 with the standard sph estimator even in the absence of particle disorder ( noise ) . as shown in appendix [ app : divv : good ] , these errors can be reduced by first estimating the velocity gradient matrix @xmath75 and then obtaining @xmath55 as its trace ( we employ a similar method to estimate @xmath76 , see appendices [ app : divv : dot ] ) .
unfortunately , even with this improved method false detections for @xmath55 ( and @xmath76 ) remain , for example in the situation of a differentially rotating disc .
these still induce artificial viscosity , which may be significant in particular if @xmath77 is small compared to the shear .
the limiter @xmath62 in equation ( [ eq : a ] ) is aimed at suppressing such false detections by @xmath78 whenever the shear is much stronger than the convergence _ and _ no shock is present .
having obtained the velocity gradient matrix @xmath79 , the shear is easily obtained as its traceless symmetric part @xmath80 ( with @xmath81 the number of spatial dimensions ) , while the presence of a shock is indicated by @xmath82 since near a shock @xmath83 for all particles .
after some experimenting , we found the following functional form for the limiter suitable @xmath84 this functional form is similar to the balsara limiter ( [ eq : balsara ] ) in that it compares the flow convergence to a measure of the traceless part of the velocity gradient ( the shear or the vorticity ) . alternatively ,
if one can be sure that no strong shear flows occur during the simulation , one may use the standard sph estimator for @xmath55 and estimate @xmath76 from its change over the last time step .
however , the limiter is still desirable and one may use @xmath85 instead of @xmath86 in equation ( [ eq : xi ] ) .
we do not use this simplified version in the tests presented below , but our experiments indicate that such a method would pass all our tests except those of
[ sec : test : disc ] and [ sec : test : shear ] , both involving strong shear .
before considering 2d and 3d test problems , we now assess the behaviour of our novel scheme , as well as that of the m&m method , in simple yet typical situations .
first , consider a sound wave of velocity amplitude @xmath87 and wave number @xmath88 as example of a well - resolved weakly convergent flow . in this case ,
@xmath89 and @xmath90 at their respective maxima .
since @xmath91 we have @xmath92 , while for the m&m method the asymptotic value @xmath93 . in the limit @xmath94 of a well - resolved wave , @xmath95 faster than @xmath96 , such that even with @xmath97 the m&m method would be more viscous than our new scheme .
[ fig : damp_sound ] shows 1d sound - wave sph runs , demonstrating that our new scheme behaves quasi - inviscid in this situation .
following @xcite , we may also consider a simple homologous flow @xmath98 with @xmath99 , which approximates certain astrophysical problems involving collapse and does not require artificial dissipation .
for this situation @xmath100 but @xmath101 ( a direct consequence of the ability of @xmath76 to distinguish shocks from convergent flows ) , such that our new scheme remains inviscid , while the m&m method does not even for @xmath97 .
next , consider a strong shock with velocity discontinuity @xmath102 .
assuming that it is smoothed over one kernel width , we find maximum amplitudes of @xmath103 and @xmath104 ( the exact values depend on the shock conditions and the functional form of the smoothing kernel ) .
since @xmath105 , our new scheme gives @xmath106 , while the asymptotic value ( [ eq : a : asym ] ) for the m&m method also approaches @xmath48 . while 3d simulations of strong shocks are presented in
[ sec : test : shock : strong ] , fig . [ fig : source ] presents weak ram - shock simulations with @xmath107 ( top ) and @xmath108 ( bottom ) for our new scheme , the m&m method , and standard sph . in both regimes
the peak in , respectively , @xmath67 and @xmath109 is one particle farther in front of the shock with our new method than with the m&m method , which reflects the superiority of @xmath66 over @xmath110 to detect an incoming shock .
this , combined with setting the viscosity parameter directly to the required value , results in the peak in @xmath29 to occur two particle separations _ before _ the shock for our new method , while for the m&m method it peaks a similar length _ behind _ the shock . with our new method , the viscous deceleration ( bottom panels in fig .
[ fig : source ] ) sets in about three particle separations before the weak and the strong shock , yielding good shock capturing properties in both cases .
the m&m method , on the other hand , decelerates the flow much earlier for a weak shock than for a strong shock and results in significant over - damping of weak shocks ( which also pertains to density and internal energy not shown in fig .
[ fig : source ] ) , while our method smoothes both shocks over four particle separations ( top panels in fig .
[ fig : source ] ) , the optimal sph resolution in 1d . note that standard sph is hopeless : it over - smoothes the strong shock and is completely incapable of dealing with the weak shock .
the main point of our method is the absence of artificial viscosity away from shocks .
hence , if @xmath74 was indeed required to maintain particle order , as previously argued in context of the m&m method , our method should fail in this regard .
noise in sph can emerge from shocks or carelessly generated initial conditions .
let us first consider the time evolution of noisy initial conditions , generated by adding random displacements to particle positions representing noise - free hydrostatic equilibrium ( the vertices of a face - centred - cubic grid , i.e. densest - sphere packing ) .
we consider two cases with the displacements in each dimension drawn from a normal distribution with rms amplitude equal to the nearest - neighbour distance or a tenth of it , respectively .
the time evolution of such noisy initial conditions can be distinguished by monitoring @xmath111 there are three possible scenarios .
either the particles settle back close to the original grid ( @xmath112 approaches its grid value @xmath113 ) , form a glass ( @xmath112 approaches a finite value @xmath114 ) , or form dense clumps ( ` clumping instability ' , @xmath115 ) .
[ fig : noise ] plots the evolution of @xmath112 for @xmath116 sph neighbours ( see also appendix [ sec : adapt ] ) when @xmath117 .
clumping only occurs when @xmath118 , while for any viscous scheme tested the particles settle back onto the grid or form a glass with roughly similar time evolutions .
post - shock noise occurs because the shock - induced compression disrupts the original particle order , but other than in the above test the viscosity is already switched on . in fig .
[ fig : sod : noise ] , we plot the amplitudes of the velocity and density noise in 3d simulations of the standard @xcite shock tube test ( see also [ sec : sod ] ) .
the three methods have similar levels of density noise , but standard sph is less noisy in the velocities , which is not surprising given its stronger viscosity . however , between the two viscosity suppressing schemes there is little difference , even though @xmath97 for our method . similar results obtain for other shock tests and we conclude that our method is no worse than m&m s for maintaining particle order .
we now present some tests of low - mach - number flows , where previous methods give too much unwanted dissipation .
the steepening of sound waves is a simple example demonstrating the importance of distinguishing between converging flows and shocks . as the wave propagates , adiabatic density and pressure oscillations result in variations of the sound speed , such that the density peak of the wave travels faster than the trough , eventually trying to overtake it and forming a shock . in our test , a 1d sound wave with a velocity amplitude @xmath119 of the sound speed is used ( ideal gas with @xmath120 ) .
[ fig:1d_sound ] compares the velocity field at the moment of wave steepening for various sph schemes , each using 100 particles , with a high - resolution grid simulation .
the new method resolves the shock better than the m&m scheme , let alone standard sph . in fig .
[ fig:1d_sound ] , we also show results from gph ( godunov - type particle hydrodynamics , * ? ? ?
* ) , which differs from sph by using the pressure @xmath121 , found by solving the riemann problem between particle neighbours , in the momentum and energy equations and avoids the need for explicit artificial viscosity .
this substitution does not affect the energy or momentum conservation @xcite , and indeed we find that both are well conserved .
while the first - order gph scheme is comparable to standard sph and also to an eulerian godunov grid code using the same riemann solver without interpolation ( not shown ) , the second - order gph scheme resolves the discontinuity almost as well as our novel method .
similar to sound - wave steepening , this test requires good treatment of convergent flows and weak shocks .
the initial conditions are uniform pressure and density and a continuous flow velocity @xmath122 as there is no analytical solution , we compare the results to a high - resolution grid - code simulation .
we run tests for @xmath123 and @xmath124 as shown in the top and bottom panels of fig .
[ fig : converging ] .
while the m&m switch certainly improves upon standard sph , it still over - smoothes the velocity profile as the viscosity is increased before a shock has formed .
this is particularly evident in the velocity profile of the @xmath124 case ( bottom ) near @xmath125 .
the new switch keeps the viscosity low , in the @xmath124 case an order of magnitude lower than the m&m method .
in fact , the agreement between our method and the high - resolution grid code is as good as one can possibly expect at the given resolution , in particular the velocity plateau and density amplitude around @xmath125 in the @xmath124 case ( bottom ) are correctly modelled . in this test ,
a gaseous ring orbits a central point mass , neglecting the self - gravity of the gas .
initially , the ring is in equilibrium : pressure forces , attraction by the point mass , and centrifugal forces balance each other .
the keplerian differential rotation implies that the flow is shearing and any viscosity causes the ring to spread @xcite .
this is indeed what @xcite found in sph simulations without pressure forces .
@xcite also found an instability to develop from the inner edge , which quickly breaks up the ring .
they argue convincingly that this is the viscous instability @xcite , which causes eccentric orbits at the inner edge of the ring to become more eccentric due to the viscous deceleration peaking at apo - centre .
@xcite performed sph simulations of the same problem but including pressure forces .
they find a similar break - up of the ring after only few rotations and blame it on an inadequacy of the sph scheme itself .
we strongly suspect that @xcite encountered a form of the clumping instability , which appears to be particularly strong in 2d simulations of strong shear flows ( though it may have been a dynamical instability inherent to gaseous keplerian rings , e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
this numerical instability grows on a local hydrodynamical time and may therefore be suppressed by choosing the sound speed @xmath71 much lower than the rotation speed @xmath126 .
indeed , @xcite and @xcite , who repeated these and similar experiments with a very low sound speed , found no such numerical instabilities .
a detailed investigation of these issues is clearly beyond the scope of our study and we merely compare our new scheme to previous methods for pressure forces with @xmath127 when the viscous instability should strike after few rotations depending on the strength of the artificial viscosity . in our test , @xmath128 for the central point mass , while the gas ring has gaussian surface density centred on @xmath129 with width ( standard deviation ) @xmath130 represented by @xmath131 particles initially placed according to the method of @xcite .
this implies an orbital period of @xmath132 and velocity of @xmath133 at the ring centre .
we choose a sound speed of @xmath134 to ensure that any dynamical instabilities of inviscid rings become important only after many periods .
figure [ fig : ring ] shows the particle distributions at various times for different sph schemes . only with our new method ,
the rings stay in their initial equilibrium configuration over at least five periods , while for the other methods , the inner parts of the ring soon become disordered leading to a catastrophic break - up after a few periods .
it is noteworthy that this failure occurs despite the balsara switch , which was designed specifically for applications like this .
note that without the viscosity limiter @xmath135 of equation ( [ eq : xi ] ) , our novel method fails , precisely because of shear causing false detections of flow compression ( as highlighted in
[ sec : shear ] and appendix [ app : divv ] ) .
we also run similar tests with the central point mass replaced by a mass distribution ( plummer sphere or kuzmin disc ) with gravitational potential @xmath136 with @xmath137 , such that the rotation curve of the disc also contains a rising part , similar to the situation in galactic discs .
the outcome of these simulations ( not shown ) is essentially identical to that for the pure keplerian rings : only our new method with viscosity limiter does not fall prey to the viscous instability .
the pulsations of a polytropic sphere are a good test for the adverse effects of artificial dissipation @xcite .
we set up a polytropic sphere of @xmath138 particles and induce oscillations in its fundamental mode ( e.g. * ? ? ?
* ) with relative amplitude of @xmath139 in radius and a period of @xmath140 . in the absence of viscosity
we expect the radial oscillations to continue with the initial amplitude and period over many oscillations .
however , as with any numerical method some small amount of numerical dissipation may appear .
nonetheless , such effects should be small compared to the dissipation caused by artificial viscosity . since the size of the radial perturbations increases with radius , we expect the oscillations to be small at the centre of the polytrope and therefore our new method to keep the viscosity low there .
however , at the edge the size of the oscillations is more significant , and we may see an increase in viscosity at this point . in order to track the oscillations , we monitor in fig .
[ fig : polyosc ] the time evolution of the virial ratio @xmath141 where @xmath142 , @xmath143 , and @xmath4 , are the kinetic , internal , and the gravitational energies , respectively . at maximum contraction
the virial ratio is at its peak and at maximum expansion the virial ratio is lowest . with no artificial viscosity ( solid curves in fig .
[ fig : polyosc ] ) the wave remains at constant amplitude barring a slight initial drop .
the period averaged over 25 oscillations is @xmath144 , only slightly larger than the expected value .
the reason for this discrepancy is most likely the unavoidable deviation of the ( finite - resolution ) sph model from a perfect polytropic sphere .
this deviation also means that our sph model is not exhibiting a pure eigenmode , but in addition contains some higher - order modes at low amplitudes , resulting in some beating between them . the m&m method results in a slow but continuous decay of the oscillations , though the period is hardly affected .
this damping can be blamed largely on the finite @xmath41 ( standard sph damps the oscillation ten times faster ) .
conversely , our new method , hardly damps the oscillations at all , because @xmath29 is kept very small ( except for the outermost layers where @xmath29 is still below the m&m values ) .
we also run simulations where the oscillating polytropic sphere is on a circular orbit 20 times the radius of the sphere around a point mass 100 times that of the sphere ( corresponding to a period of 56 time units ) . with this choice ,
the tidal radius is approximately four times the radius of the gas sphere , implying that tides are strong but not catastrophic .
since the orbital accelerations are much larger than those due to the polytropic oscillations , this is a tough test for any numerical scheme . in particular
, eulerian methods should have severe problems ( this does exclude using co - rotating coordinates , which do not allow for tidal evolution of the orbit and are unavailable for eccentric orbits ) .
the time evolution of the virial ratio and the viscosity parameter @xmath29 are shown in fig . [
fig : polypoint ] for the same viscosity schemes as for the isolated case in fig .
[ fig : polyosc ] .
first note that the undamped simulations ( solid curves ) behave differently from the isolated case , exhibiting variations and a slight decay , both of which are most likely caused by the tidal field .
as to be expected for any lagrangian scheme , both sph methods perform very similar to the isolated case , because neither @xmath55 nor @xmath76 are affected by the orbital acceleration .
in this section , we subject our method to situations where artificial viscosity is required , mainly high - mach number shocks , and our aim is to demonstrate that it performs at least as well as the m&m method . the @xcite shock tube test is a standard test for any shock capturing method and consists of an initial discontinuity in pressure and density leading to the production of a rarefaction wave , contact discontinuity and shock wave , which forms from the steepening of a subsonic wave .
the whole system is subsonic with a maximum mach number of @xmath145 in the pre - shock region .
we perform the test in 3d at a resolution of 200 particle layers in the high - density region .
the density , energy , velocity , and viscosity for standard sph as well as the m&m and our method are shown in fig .
[ fig : sod ] . as for the 1d ram test ( see fig [ fig : source ] ) , our new method switches on viscosity already in the pre - shock region peaking about one smoothing length before the actual shock front ( which travels to the right in fig .
[ fig : sod ] ) , whereas the m&m switch turns on viscosity later , lagging our method by about four particle separations . as a consequence ,
the transition of the fluid values across the shock front is slightly smoother with our method than with the m&m method .
note that the irregularities around the contact discontinuity at @xmath146 common to all schemes tested are not related to artificial viscosity ( the irregularities in @xmath29 at that point could be removed by choosing non - zero initial @xmath29 at the initial discontinuity ) ; they can be alleviated by artificial conductivity @xcite . in [ sec
: av : behaviour ] and fig .
[ fig : source ] we already demonstrated that our new method is superior to the m&m scheme in resolving ` subsonic shocks ' ( velocity discontinuities smaller than the sound speed ) and comparable in resolving shocks of mach number @xmath147 . here , we extend this comparison to high mach numbers .
[ fig:1dram50 ] shows the result for the 1d ram test with @xmath148 . the @xcite version of the m&m method , which uses equation ( [ eq : davdt : alt ] ) with @xmath47 , is implemented in some contemporary sph codes , and has been used in our tests so far , fails this test : @xmath29 remains too low and as a consequence the velocity discontinuity is not correctly smoothed and some post - shock ringing occurs .
to give credit to @xcite , we also tested their original method and find it to work well ( stars in fig .
[ fig:1dram50 ] ) .
our new method works about as well as the original m&m scheme , with @xmath29 reaching the same level , though our scheme detects the coming shock much earlier : @xmath29 is ahead of the original m&m method by about four particle separations . whilst the main role of artificial viscosity is to resolve shocks by transferring entropy ,
a secondary but vital role is to prevent particle penetration , which requires strong viscosity in high mach number shocks .
@xcite performed many tests to determine the value of the parameters @xmath29 and @xmath149 needed to prevent particle penetration in ram shock tests of various mach numbers . for particles arranged in face - centred - cubic or cubic grids , bate found that appropriate values for the viscosity parameters can prevent particle penetration for shocks up to @xmath150 .
most sph practitioners opt for a value of @xmath151 @xcite . to determine the correct value of @xmath149 required for the new scheme , we perform high - resolution 3d runs of ram shocks with @xmath152 and various values for @xmath153 .
we smooth the initial velocity discontinuity , as suggested by @xcite , to provide the method with a situation realistic for sph , such as would have arisen for a shock forming from continuous initial conditions . for different values of @xmath153 with our viscosity scheme and the two variants of the m&m switch , we plot in fig .
[ fig:3dram ] the @xmath154 and @xmath155 positions ( for all values of @xmath156 ) of particles near the shock front at a late time .
the colour coding distinguishes particles which at that time should be up- ( red ) or downstream ( green ) .
our scheme prevents particle penetration with @xmath157 ( for @xmath23 there is one layer of overlap ) .
the original m&m scheme with the standard choice @xmath151 also avoids particle penetration , but not the @xcite version , again a consequence of too little viscosity .
this test combines a shock with a perpendicular shear and presents a difficult test for any sph scheme .
we use periodic boundary conditions and start from a face - centred cubic grid and velocities @xmath158 in fig .
[ fig : shear : shock ] , we present results for various sph simulations as well as a grid - code run for @xmath159 .
the m&m method produces a large viscosity due to the shear - induced errors in @xmath55 , leading to spurious results .
using the balsara limiter with either m&m or standard sph gives in much better results , though the shock is clearly over - smoothed .
the new scheme is able to limit the viscosity to the correct level , allowing good capturing of the shock and retaining particle order in the post - shock region .
note that this is a difficult test for any sph implementation : without viscosity reduction ( as in standard sph ) the shear flow is strongly damped , while viscosity reduction schemes ( m&m as well as ours ) suffer from the problem of shear - induced errors .
these potentially result in too much viscosity and over - smoothing of the shock .
our limiter was able to control this problem , but for yet larger ratios @xmath160 of shear to shock amplitude this problem becomes too difficult for any sph implementation . in this test
the inward gravitational pull of a gas cloud exceeds its outward pressure force causing the cloud to collapse under its own self - gravity .
the initial conditions consist of a gas sphere with density profile @xcite @xmath161 for @xmath162 and @xmath163 for @xmath164 .
initially the gas is at rest and has constant specific internal energy @xmath165 , which corresponds to a virial ratio @xmath166 .
the initial gravitational inward pull is the same at each radius , while the pressure forces decline outwards , leading to collapse and , as a consequence , formation of a shock , which steepens and evolves into a strong shock propagating outwards as more incoming material joins the jam . even though the problem is initially spherically symmetric , the sph realisation of initial conditions can not be exactly spherically symmetric and
the system may well evolve away from sphericity , for instance driven by dynamical instabilities .
we use a unit system such that @xmath167 and represent the cloud by 100280 sph particles , initially placed on a face - centred - cubic grid which is then radially stretched to match the density .
[ fig : evrard ] compares the simulation results for our method , the original m&m method , and a 1d calculation by @xcite using the piece - wise parabolic method ( ppm ) . at early times
( @xmath168 , left column ) the results from all three methods match very well , but the m&m scheme already shows a large viscosity . at later times a shock forms ( at @xmath169 by @xmath170 ) , which moves outwards until it reaches the end of the sphere , when a significant fraction of the gas still has outwards velocities ( by @xmath171 ) .
the most obvious difference between the two sph schemes is the amount of ( artificial ) dissipation : the m&m method is much more viscous , resulting in significant over - smoothing of the shock front by @xmath170 accompanied by unphysical pre - shock heating as visible in the entropy ( @xmath172 ) profile .
our new scheme agrees better with the 1d calculation , in particular in the inner ( post - shock ) regions .
note that with our new method @xmath29 peaks well before the shock arrives ( at @xmath173 ) , while for the m&m method the peak in @xmath29 appears actually slightly after the shock .
we found this a valuable test as early versions of our scheme tended to be far too viscous , while our final version passes this test ahead of the m&m switch .
standard sph ( not shown in the figure ) shows similar results , though the shock at @xmath170 appears less smoothed than with the m&m method but more smoothed than with the new scheme .
any hydrodynamical numerical method requires some form of artificial viscosity in order to resolve shocks ( in grid methods , artificial viscosity is implicit in the riemann solver , * ? ? ?
* ) . in grid codes , such as @xcite ,
interpolation methods are employed to effectively suppress artificial viscosity away from shocks .
most sph simulations to this date hardly use such precautions and , as a consequence , adiabatic oscillations and shear - flows are damped . note
that this affects state - of - the - art simulations of , e.g. galaxy formation , which usually only employ @xcite s ( @xcite ) rather inefficient method to reduce some adverse effects of artificial viscosity on rotation discs .
the method of @xcite , which reduces the default amount of artificial viscosity by an order of magnitude compared to standard sph practice , has only recently been recognised as advantageous . in this method ,
explained in detail in
[ sec : av : mm ] , individual artificial viscosities @xmath33 are adapted by integrating a differential equation . though constituting a major improvement
, this method remains unsatisfactory , because it still damps adiabatic oscillations and over - smoothes weak shocks , as we argued in
[ sec : av : novel ] and demonstrated in [ sec : test : suppress ] . in [ sec
: av : novel ] , we present a novel method , which improves upon that of @xcite in four important ways . *
we set @xmath97 enabling @xmath56 away from shocks and effectively modelling the fluid as inviscid .
* we use @xmath174 rather than @xmath83 as shock indicator .
this distinguishes pre - shock from post - shock regions ( where @xmath60 but @xmath83 ) and discriminates much better between converging flows and weak shocks .
* we set @xmath33 directly to an appropriate local value @xmath67 , instead of growing it by integrating a differential equation .
* we use an improved estimator for @xmath55 and @xmath76 and employ a limiter to avoid viscosity driven by shear - induced errors .
together these novelties result in a significantly improved artificial viscosity method , in particular the viscosity is increased to an appropriate level well before an incoming shock .
the implementation details , i.e.the precise way of setting @xmath67 from @xmath76 and the exact form of the limiter , may well be subject to improvements . any reader who considers modifying these details is advised to consider the behaviour of the resulting method for a test suite comprising noise suppression as well as shear and strong shocks , for example the tests of figures [ fig : noise ] , [ fig : ring ] , and [ fig : shear : shock ] . for static equilibria @xmath175 and @xmath176 , and our new shock indicator ( as well as the m&m shock indicator )
are only triggered by velocity noise .
as long as particle order is maintained , such noise triggers only negligible amounts of viscosity , unlike the situation with the m&m method , whose minimum viscosity @xmath39 is often sufficient to affect the simulations ( as demonstrated in
[ sec : test : suppress ] ) . nonetheless , the noise - induced viscosity is sufficient to suppress particle disorder , as demonstrated in [ sec : av : amin ] . for dynamic equilibria @xmath175 ( and @xmath177 ) but @xmath178 .
however , in multi - dimensional flows strong shear induces false detections of @xmath55 ( and @xmath76 ) , even with best possible particle order , for reasons explained in appendix [ app : divv : fail ] . in simulations of differentially rotating discs
, this problem strongly affects the m&m method ( even when using the balsara switch ) .
we avoid this problem by applying a limiter ( see [ sec : shear ] ) as well as using improved estimators for @xmath55 and @xmath76 , see appendix [ app : divv : good ] for details .
( alternatively , if no strong shear flows are present , the standard estimators should suffice , though still in conjunction with a limiter using @xmath179 as a proxy for the shear amplitude . )
these improved estimators also provide the full velocity and acceleration gradient matrices for each particle ( and increase the computational costs by @xmath180 ) .
the knowledge of the velocity gradient matrix @xmath181 and its traceless symmetric part , the shear @xmath182 , is also useful for implementing physical viscosity @xmath183\ ] ] ( with @xmath184 and @xmath185 the shear and bulk viscosity coefficients ) in sph . in sections
[ sec : av : amin ] , [ sec : test : suppress ] , and [ sec : test : shock ] , we demonstrate convincingly that our technique successfully deals with the following four situations . * shocks * : : are resolved at least as well , if not better , than with any previous technique ; * adiabatic oscillations * , : : such as sound waves or stellar pulsations , remain undamped over many periods , which was not possible with any previous sph implementation ; * strong shear flows * , : : such as in accretion discs , are modelled virtually inviscid , while shearing shocks are well resolved without being over - smoothed ; * particle disorder * : : is suppressed at least as well as with the m&m method . in particular , in the regime of convergent flows and weak shocks our new method is far superior to any previous scheme , which all required a significant increase in resolution just to suppress adverse effects of artificial viscosity . research in theoretical astrophysics at leicester is supported by a stfc rolling grant .
we thank all members of the leicester theoretical astrophysics group , in particular graham wynn and justin read , for helpful discussions , seung - hoon cha for contributing the gph data to fig .
[ fig:1d_sound ] , matthias steinmetz for providing the ppm data of the evrard test for fig .
[ fig : evrard ] , and the anonymous referee for helpful comments and enhanced scrutiny .
d. , 1995 , journal of computational physics , 121 , 357 m. , 1995 , phd thesis , university of cambridge , uk a. , stamatellos d. , whitworth a. p. , 2009
, , 395 , 2373 s .- h .
, 2002 , phd thesis , cardiff university , uk s .- h . , whitworth a. p. , 2003
, , 340 , 73 j. p. , 1980
, theory of stellar pulsation .
princeton university press , princeton , nj , usa w. , 2002 , journal of computational physics , 179 , 27 k. , vazza f. , brunetti g. , tormen g. , 2005 , , 364 , 753 a. e. , 1988 , , 235 , 911 r. a. , monaghan j. j. , 1977 , , 181 , 375 r. a. , monaghan j. j. , 1983 , , 204 , 715 p. , narayan r. , 1985 , , 213 , 7p y. , inutsuka s .- i .
, 2002 , , 569 , 501 l. b. , 1977 , , 82 , 1013 d. , pringle j. e. , 1974 , , 168 , 603 y. e. , postnov k. a. , prokhorov m. e. , 1994 , , 266 , 583 s. t. , murray j. r. , monaghan j. j. , 1996 , publications of the astronomical society of australia , 13 , 66 z. , wickramasinghe d. , bicknell g. v. , 1993 , , 264 , 691 j. j. , 1992 , , 30 , 543 j. j. , 1997 , journal of computational physics , 136 , 298 j. j. , 2006 , , 365 , 199 j. j. , lattanzio j. c. , 1985 , , 149 , 135 j. p. , monaghan j. j. , 1997 , journal of computational physics , 136 , 41 j. c. b. , pringle j. e. , 1984 , , 208 , 721 j. c. b. , pringle j. e. , 1985 , , 213 , 799 d. j. , 2004 , phd thesis , university of cambridge , uk d. j. , 2008 , journal of computational physics , 227 , 10040 s. , davies m. b. , thielemann f .- k .
, piran t. , 2000 , , 360 , 171 g. a. , 1978 , journal of computational physics , 27 , 1 v. , 2005 , , 364 , 1105 v. , hernquist l. , 2002 , mnras , 333 , 649 m. , mller e. , 1993 , , 268 , 391 r. , 2002 , , 385 , 337 m. , nelson a. f. , naab t. , burkert a. , 2009 , , 184 , 298
for completeness , we give here a brief description of our sph method , which is largely similar to previous methods , but may differ in some details .
let @xmath81 denote the number of spatial dimensions , then we adapt the individual smoothing lengths @xmath5 such that @xmath186 with @xmath187 a global constant , defined in terms of the number @xmath188 of neighbours , the mass @xmath189 of each sph particle , and the volume @xmath190 of the unit sphere .
in this work , we use @xmath191 and 40 for @xmath192 and 3 dimensions , respectively . inserting the density estimator ( [ eq : est : rho ] ) , we find @xmath193 where we have re - written the sph kernel as @xmath194 with the dimensionless function @xmath195 . for this work , we employ the usual cubic spline kernel @xcite @xmath196 2(1-r)^3 & 1/2<r<1 , \\[0.5ex ] 0 & \mathrm{otherwise}. \end{array}\right.\ ] ] at each time step , the @xmath5 are adjusted by performing one newton - raphson step in @xmath197-@xmath198 space , i.e.@xmath199 with a factor of order unity @xmath200 where @xmath201 and @xmath202 .
this method converges extremely well , except when @xmath5 was much too small . in this case , faster convergence can be achieved by subtracting the self - contribution ( which does not depend on @xmath5 ) .
thus , whenever @xmath203 we use instead of ( [ eq : h : adapt ] ) @xmath204 the time derivatives @xmath205 are obtained by demanding @xmath206 : @xmath207 we use sph equations of motion derived from the simple sph lagrangian @xmath208 . together with the relation in the lagrangian with @xmath209 and consider the entropy function @xmath210 to be constant @xcite . ]
@xmath211 , this gives @xmath212 where the factors @xmath213 and @xmath214 ( equation [ eq : f ] ) arise from the fact that the derivatives @xmath215 have to be taken at fixed @xmath216 .
the work done by these pressure forces has to be balanced by @xmath217 for the artificial viscosity drag and heating we actually use @xmath218 with @xmath219 , where @xmath220 while the parameter @xmath221 has the meaning of @xmath153 for traditional sph .
note that equations ( [ eq : dv : av ] ) and ( [ eq : dv ] ) differ only by @xmath222 .
the difference between equations ( [ eq : du : av ] ) and ( [ eq : du ] ) is more pronounced since , similarly to equation ( [ eq : udot ] ) , we do not symmetrise the contributions w.r.t . @xmath7 and @xmath223 .
our scheme employs a kick - drift - kick leap - frog time integrator , which is second - order accurate . with this scheme ,
a full ( global ) time step of size @xmath224 consists of the following sub - steps ( ` @xmath225 ' means ` is replaced by ' ) .
* initial kick * : : compute @xmath226 and @xmath11 at half step @xmath227 \tilde{u}_i & = & u_i + \tfrac{1}{2}\,\delta t\,\dot{u}_i .
\end{array}\ ] ] * full drift * : : advance @xmath228 and @xmath2 by a full step : @xmath229 * prediction * : : predict @xmath226 , @xmath11 , and @xmath5 at full step : @xmath230 * sweep 0 * : : compute @xmath231 and @xmath213 ( equations [ eq : mh ] and [ eq : f ] ) . * adapt * : : adjust @xmath5 ( equation [ eq : h : adapt ] or [ eq : h : adapt : better ] ) . * sweep 1 * : : compute @xmath232 , @xmath213 , and @xmath205 ( eqs .
[ eq : mh ] , [ eq : f ] , and [ eq : hdot ] ) as well as @xmath233 , @xmath234 , and @xmath235 ( eqs .
[ eq : divv : new ] , [ eq : dtdivv ] , and [ eq : r ] , using @xmath236 and @xmath55 from the previous time step ) . * between sweeps * : : obtain @xmath237 and @xmath238 from @xmath232 and @xmath11 via the equation of state , and adapt @xmath33 via ( using eqs .
[ eq : tau ] and [ eq : a : loc ] ) @xmath239 * sweep 2 * : : compute @xmath240 ( eqs .
[ eq : vdot ] and [ eq : dv : av ] plus gravitational forces ) and @xmath241 ( eqs .
[ eq : udot ] and [ eq :
du : av ] plus external heating or cooling ) . * final kick * : : set @xmath226 and @xmath11 at full step : @xmath242 u_i & = & \tilde{u}_i + \tfrac{1}{2}\,\delta t\,\dot{u}_i .
\end{array}\ ] ] in the initial kick and prediction steps , the time derivatives are known from the previous time step ( in case of the very first time step , they need to be precomputed ) .
note that the quantities predicted in ( [ eq : pred ] ) enter the final @xmath226 and @xmath11 only indirectly via the computation of the time derivatives .
we use an oct - tree , generated just before sweep 0 , to find all interacting particle pairs , which are then remembered in an interaction list , whereby allowing for the fact that @xmath5 may grow slightly during adjustment ( just after sweep 0 ) . utilising this interaction list in sweeps 1 and 2
is much faster than further tree walks .
the same oct - tree is also used in computing gravitational forces , as outlined by @xcite .
our scheme can also be implemented with adaptive individual time steps organised in a hierarchical block - step scheme , though we have not used this in the tests presented in this study .
our constraint that @xmath231 be constant ( see [ sec : adapt ] ) implies @xmath243 . together with the continuity equation @xmath244 and equation ( [ eq : hdot ] ) this yields the simple velocity - divergence estimate @xmath245 while this estimate satisfies the continuity equation for the sph density estimate @xmath232 , it is not necessarily accurate . to see this ,
consider the matrix ( @xmath246 denotes the outer or dyadic vector product ) @xmath247 with @xmath248 some weighting factor . assuming a smooth velocity field , we may replace @xmath249 in equation ( [ eq : d ] ) with its taylor expansion @xmath250 , where @xmath251 is the gradient of @xmath252 at position @xmath2 , and obtain @xmath253 with the symmetric matrix @xmath254 comparing ( [ eq : d ] ) and ( [ eq : t ] ) to the simple estimator ( [ eq : divv : std ] )
, we see that the latter corresponds to ( conveniently dropping the index @xmath7 ) @xmath255 and the weights @xmath256 .
if we split @xmath181 into its isotropic part ( divergence ) , the symmetric traceless part @xmath182 ( shear ) , and the antisymmetric part @xmath257 ( vorticity ) , @xmath258 and insert it into ( [ eq : dvt ] ) , we find for the simple estimator ( [ eq : divv : std ] ) @xmath259 where @xmath260 denotes the anisotropic ( traceless ) part of @xmath261 .
thus , the simple estimator ( [ eq : divv : std ] ) contains an @xmath262 error term , which originates from anisotropy of @xmath261 in conjunction with velocity shear ( owing to the symmetry of @xmath261 the vorticity is harmless ) . for perfectly symmetric particle distributions @xmath263 , but in general @xmath264 such that in the presence of strong shear even a small residual @xmath260 results in a failure of the simple estimator ( [ eq : divv : std ] ) .
this typically happens in differentially rotating discs , where ( i ) the velocity field is divergent - free but contains shear and ( ii ) even in the absence of noise @xmath264 owing to the shearing particle distribution . from equation ( [ eq : dvt ] ) , we can also estimate @xmath265 which allows an improved divergence estimator @xmath266 in order to assess the error of this estimator , let us expand the flow to second order , replacing equation ( [ eq : dvt ] ) with ( dropping the index @xmath7 and using suffix instead of matrix notation ) @xmath267 with the symmetric tensor @xmath268 . inserting this into ( [ eq : v : est ] )
we find @xmath269 thus , while this estimator avoids an @xmath262 error , we still have an @xmath270 error term ( since @xmath271 is one order higher in @xmath27 than @xmath261 ) .
we can _ reduce _ the @xmath270 error by a careful choice of the weights @xmath248 .
if , for instance , @xmath272 then @xmath273 to leading order in the _ continuum limit _ by virtue of the isotropy of the kernel .
this limit , which is commonly used to assess sph estimators , replaces @xmath274 with @xmath275 under the assumption of a smooth density without particle noise , which causes the @xmath262 error term in the simple @xmath55 estimator , vanishes . ] .
as these conditions are hardly ever truly satisfied , we can only reduce but not eliminate the @xmath270 error term as we do not even try to avoid the @xmath276 error ( hidden in ` h.o.t . '
above ) , such a reduction should be okay in most cases .
we can estimate @xmath76 either from the change in the estimated @xmath55 over the last time step or as the trace of @xmath277 , the total time derivative of @xmath181 . since ( with @xmath278 the gradient of the acceleration ) @xmath279 ( a good exercise for your undergraduate students ) , we can estimate @xmath280 here , the estimate @xmath281 is obtained from the accelerations at the previous time step in a way analogous to estimating @xmath282 , in particular we need to compute the matrix @xmath283 and its inverse only once . the lowest - order error in this estimate
again is @xmath284 , such that reducing @xmath285 by careful choice of the weights remains a good idea .
note that , by virtue of equation ( [ eq : vdot = a - v^2 ] ) , we could estimate @xmath76 also as @xmath286 with the acceleration divergence @xmath287 estimated using the standard divergence estimator , in the hope that its @xmath262 error term is small since the acceleration is hardly sheared .
|
in smooth - particle hydrodynamics ( sph ) , artificial viscosity is necessary for the correct treatment of shocks , but often generates unwanted dissipation away from shocks .
we present a novel method of controlling the amount of artificial viscosity , which uses the total time derivative of the velocity divergence as shock indicator and aims at completely eliminating viscosity away from shocks .
we subject the new scheme to numerous tests and find that the method works at least as well as any previous technique in the strong - shock regime , but becomes virtually inviscid away from shocks , while still maintaining particle order . in particular sound waves or oscillations of gas spheres
are hardly damped over many periods .
[ firstpage ] hydrodynamics methods : numerical methods : @xmath0-body simulations
|
two neutron - capture processes are important for astrophysical nucleosynthesis of heavy elements .
the first one is a rapid neutron - capture process ( r - process ) that is considered to occur in supernova ( sn ) explosions @xcite , and the other is a slow neutron - capture process ( s - process ) in low - mass asymptotic giant branch ( agb ) stars @xcite or massive stars @xcite .
long - lived radioactive nuclei are used as nucleo - cosmochronometers , which are useful for an investigation of nucleosynthesis process history along the galactic chemical evolution ( gce ) before the solar system formation .
a general idea of the nucleo - cosmochronometer was proposed by rutherford about 70 years ago @xcite .
radioactive nuclei of cosmological significance are rare and only six chronometers with half - lives in the range of the cosmic age 1 @xmath11 100 gyr are known .
they are @xmath12k @xcite and @xmath13rb @xcite for the s - process or explosive nucleosynthesis in sne , @xmath14lu @xcite for the photodisintegration reaction nucleosynthesis in sne ( @xmath3-process ) , and @xmath1re @xcite , @xmath15th and @xmath16u @xcite for the r - process . recently
two actinoid elements u and th in a very metal - poor star were detected for the first time @xcite .
the actinoid nuclei in the old metal - poor star were perhaps created in a single r - process of a sn explosion .
since the initial abundances of @xmath16u and @xmath15th can not be measured directly these abundances should be calculated in the framework of the r - process models .
however the calculated results by many models @xcite are different from one another and thus a large uncertainty still remains in the estimation with the u - th chronometer .
a pair of nuclei , @xmath1re-@xmath1os , was proposed to be an intentional cosmochronometer to date the r - process @xcite .
1 shows a nuclear chart and flows of nucleosynthesis processes around the nucleus @xmath1re .
the nucleus @xmath1re is predominantly produced by @xmath17-decay after the freezeout of the r - process and is located outside the main path of the s - process .
the s - process nuclei @xmath18os are not directly produced by the r - process because they are shielded by the stable isobars of @xmath0w and @xmath1re against the @xmath19-decay after the r - process .
the pure s - process nucleus @xmath0os is used for the normalization of the s - process abundances in this mass region . the s - process nucleus @xmath1os is also produced by the cosmoradiogenic @xmath20-decay of @xmath1re .
the ground state of @xmath1re decays to @xmath1os with a half - life of 4.35@xmath2110@xmath22 yr @xcite , which is longer than the age of the universe .
it should be noted that the @xmath1re-@xmath1os chronometer has the advantage that applying this chronometer is free from the uncertainty of the initial abundances calculated by the r - process models .
the epoch from an r - process nucleosynthesis event to the present can be evaluated by the present abundances of @xmath1re and @xmath1os after the subtraction of the s - process contributions to these nuclei .
although the initial abundances of @xmath1re and @xmath1os should be calculated using the s - process models , the uncertainty of the calculated s - process abundances is generally smaller than those in the r - process models .
the @xmath1re-@xmath1os chronometer has been applied for studying the gce @xcite .
this chronometer was also used in analyses of meteorites @xcite .
the measurements of the @xmath1re and @xmath1os abundances in iron meteorites and ordinary chondrites were used for evaluating the ages of the meteorites , which were formed from a metal reservoir at the solar system formation . these results lead to an upper - limit estimation of the age of the galaxy @xcite .
it was pointed out that the half - life of the @xmath20-decay of @xmath1re is affected by stellar environments @xcite .
the half - life depends on the temperature and the electron density of the stellar environments .
the half - life becomes shorter in high temperature such as @xmath23 @xmath24 10@xmath25 k due to the enhanced bound - state @xmath19-decay in ionized atoms .
this phenomena was verified in a measurement of the change of the half - life of the ionized @xmath0re @xcite .
however the @xmath1re-@xmath1os system is still an important chronometer in the astronomical observations of old stars and in the analyses of pre - solar grains in primitive meteorites for the chronology of the r - process .
the chemical compositions of the extremely metal - poor stars are enhanced by the production of heavy elements in neutron - capture nucleosynthesis in the first generation stars .
recently astronomical observations with isotope separations have been carried out for several heavy elements such as ba @xcite and eu @xcite .
likewise , the pre - solar grains in the primitive meteorites provide samples affected strongly by a single nucleosynthesis event .
heavy elements such as mo , te and ba in the pre - solar grains have already been separated into isotopes .
the origins of these pre - solar grains are considered to be the ejecta of core collapse sn explosions @xcite .
the purpose of this paper is to present a contribution of a new s - process path to @xmath1re and @xmath1os in order to improve the accuracy of this cosmochronometer . in the previous studies the s - process contributions via two weak paths , @xmath26w(n,@xmath3)@xmath2w(n,@xmath3)@xmath0w(n,@xmath3 ) @xmath1w(@xmath27,e@xmath28)@xmath1re and @xmath2re(n,@xmath3)@xmath0re@xmath5(n,@xmath3 ) @xmath1re , were studied .
measured the neutron capture cross sections of @xmath2re and @xmath1re at @xmath8 = 25 kev @xcite .
the neutron capture cross section of an unstable nucleus @xmath2w is evaluated from the inverse reaction @xmath0w(@xmath3 , n)@xmath2w @xcite . here
we propose another new weak s - process branch to produce @xmath1re : a path through a @xmath0re isomer which has a half - life of 2.0@xmath2110@xmath29 yr .
this isomer was found by lindner about 50 years ago @xcite , and then seegmiller _ et al . _
measured the half - life of this isomer by a mass - spectrometric analysis and @xmath3-ray spectroscopy of samples after neutron irradiation @xcite .
however the absolute value of the neutron - capture cross section for the @xmath2re(n,@xmath3)@xmath0re@xmath30 reaction has not been reported .
et al . _ reported a ratio of the neutron - capture cross section of @xmath2re leading to the isomer and the ground state , but the uncertainty was not given in the article .
therefore the neutron capture cross section to the isomer has not been established @xcite . the half - life of the ground state of @xmath0re is only 3.718 days , while that of the isomer , 2.0@xmath2110@xmath29 yr , is longer than the timescale of the s - process , 10@xmath31 @xmath11 10@xmath32 yr . the s - process contribution should be , thus , estimated with high accuracy . in the present work we measure the neutron - capture cross section ratio @xmath2re(n,@xmath3)@xmath0re@xmath30 / @xmath2re(n,@xmath3)@xmath0re@xmath5 at the thermal energy .
an implication of this path in the abundances of @xmath1re and @xmath1os is discussed using our data in a classical steady - flow model , and an effect to the chronometry is evaluated in a sudden approximation .
in order to measure the ratio of the neutron - capture cross - sections @xmath2re(n,@xmath3)@xmath0re@xmath30 and @xmath2re(n,@xmath3)@xmath0re@xmath5 at the thermal energy , two thin natural re foils were irradiated with thermal neutrons from the nuclear reactor jrr-4 at the japan atomic energy research institute ( jaeri ) .
two al - co wires were also irradiated simultaneously to monitor the neutron flux .
the samples were irradiated by the reactor neutrons for six hours .
after the irradiation the samples were cooled in a water pool for about four months in order to reduce the background of the unstable isotopes with short half - lives .
four and eight months after the irradiation , @xmath3-rays emitted from the activated samples were measured by two hpge detectors . in addition
sixteen months later , the @xmath3-rays from one sample were measured by one hpge detector .
the weights of the re foils are 43.8 mg and 58.1 mg , whereas the weights of the al - co wires are 1.74 mg and 4.58 mg .
the co element composition is 0.475% .
the natural re sample consists of two stable isotopes , namely @xmath2re ( 37.4% ) and @xmath1re ( 62.6% ) , and thus two unstable isotopes , @xmath0re and @xmath33re , are mainly produced by the neutron capture reactions .
the half - lives of the ground states of these isotopes are @xmath34 days ( @xmath0re ) and @xmath35 hr ( @xmath33re ) , which are much shorter than the @xmath0re isomer .
the average flux of the neutrons is estimated to be 4.37@xmath2110@xmath36 @xmath37s@xmath38 , which is evaluated by a measurement of decay @xmath3-rays after @xmath19-decay from the activated co in the wires .
the @xmath3-rays from @xmath39al can not be measured because a cooling time of four months is much longer than the half - life of @xmath39al .
the total efficiency of the hpge detectors is approximately equal to 20% relative to a 3@xmath403@xmath41 nai detector and the energy resolution is 2.2 kev for a 1.3 mev @xmath3-ray in @xmath42co .
the efficiency is calibrated with standard sources of @xmath43eu and @xmath44ba .
the @xmath3-rays from the samples are measured at a 5 cm or 3 cm distance from the end caps of the detectors in the first or second ( third ) measurement , respectively , with a lead - shielded low - background environment .
2 shows a partial decay scheme of the @xmath0re isomer @xcite .
the isomer with @xmath45=(8@xmath46 ) locates at an excitation energy of 149 kev .
this state decays to the ground state through a cascade of @xmath3-decay .
the ground state subsequently decays to @xmath0os or @xmath0w with @xmath47=3.718 days .
the @xmath3-ray transitions from the isomer to the ground state can not be distinguished from @xmath3-rays emitted from other activities .
strong @xmath48-rays are irradiated from @xmath26re ( @xmath47=38 days ) , which is produced by a ( n,2n ) reaction with fast neutrons existing in the nuclear reactor ( see fig .
its half - life is comparable to the cooling time , and hence the intensities are enhanced .
however we observe a 137 kev transition of @xmath0os ( see fig . 4 ) , which is the strongest @xmath3-ray in the decay scheme associated with the isomer .
the yield of the 137-kev @xmath3-ray is contributed not only from the isomer but also from the ground state produced directly by the @xmath2re(n,@xmath3)@xmath0re@xmath5 reaction . in order to obtain the ratio of the neutron - capture cross section between the isomer and the ground state , we select the cooling time of four months after the irradiation for the first measurement . after the cooling , the ground state of @xmath0re is still remained because of its large capture cross - section relative to the isomer .
although the surviving number of the ground state is small , the contribution to the 137 kev @xmath3-ray yield is competitive to the isomer because the decay rate of the ground state is much higher than the isomer . in the second and third measurements
the ground states are not remained .
the yields of the 137-kev @xmath3-ray at the second and third measurements are , thus , proportional to the neutron capture cross section to the isomer .
the number of the nuclei produced by the neutron - capture reaction is proportional to the @xmath19-decay events , which are evaluated from the @xmath3-ray intensity in the measured spectra by applying the following equation of @xmath49 where @xmath50 means the total number of the @xmath19-decays , @xmath51 denotes the peak counts corresponding to the decay @xmath3-rays , @xmath52 the emission probability of the measured @xmath3-ray per decay , @xmath53 the efficiency of the hpge detectors and @xmath54 the correction of the @xmath19-decay during the measurement time .
the @xmath54 is expressed by @xmath55 where @xmath56 and @xmath57 stand for the start and stop time of the measurement , respectively .
the experimental uncertainty consists mainly of the statistical error of the yields of the @xmath3-rays and the efficiency of the hpge detectors .
the @xmath3-ray energy of 137 kev is near the peak of the efficiency curve of the detectors .
we evaluate the efficiency uncertainty to be approximately 7% from the deviation of the @xmath3-ray efficiency obtained with the calibration sources from the efficiency curve estimated by @xmath58 fitting .
the pileup effect of the @xmath3-rays is negligibly small because the absolute efficiency is lower than 5% .
the self - absorption of the @xmath3-rays is also negligibly small .
an energy spectrum of neutrons produced by a nuclear reactor consists of thermal flux with energy below 0.2 ev and epithermal one above .
the rate of the neutron capture reaction is expressed by @xmath59 where @xmath60 denotes the reaction rate for the reaction @xmath61 , @xmath62 the neutron density at the velocity @xmath63 , @xmath64 the cross section of the reaction @xmath61 at the velocity @xmath63 , @xmath65 and @xmath66 the thermal and epithermal flux , @xmath67 and @xmath68 are the cross section at the thermal energy and the resonance integral for the reaction @xmath61 , respectively .
a measurement for a sample with a cd absorber is not carried out in the present experiment and we can not obtain the thermal neutron flux .
instead we use an average neutron flux , which is defined by @xmath69 where @xmath70 the average neutron flux and @xmath71 the effective cross section of the reaction @xmath61 in the reactor neutron spectrum .
the effective neutron - capture cross sections with the average neutron flux are defined by @xmath72 applying these equations , we obtain the effective neutron capture cross sections leading to the isomer of @xmath0re as given in table 1 , where the mean value derived from the second and third measurements is 0.74 @xmath7 0.05 barn .
the yield of the @xmath3-ray at the first measurement consists of contributions from the ground - state decay as well as the isomer decay .
the effective neutron capture cross section to the ground state can be evaluated by subtracting the contribution of the isomer decay , which is calculated from the average cross section of 0.74 @xmath7 0.05 barn , from the total yield at the first measurement ( see table 2 ) .
the obtained cross section to the ground state is 132 @xmath7 26 barn .
this value is consistent with the thermal values of 118 barn @xcite and 112 barn @xcite within the uncertainty and thus the effective cross section derived here is essentially the thermal cross section . to verify this conjecture
, we compare the thermal cross sections and resonance integrals leading to the ground state and isomer of @xmath0re and their ratio in table [ table : jendl ] .
these values are calculated from japanese evaluated nuclear data library ( jendl ) activation cross section file 96 @xcite .
the ratios , @xmath73/@xmath74 , for both the ground state and isomer are approximately equal ( see table 3 ) .
this fact indicates that the resonance integral are written by @xmath75 with the common coefficient @xmath76 .
the isomer / ground ratio of the capture cross sections at the thermal energy can be derived as @xmath77 which verifies our conjecture .
the isomer / ground ratio obtained by applying this equation is @xmath6 = 0.54 @xmath7 0.11% .
this value is a factor of 1.8 larger than the value of 0.3% reported in a previous report @xcite .
note that experimental uncertainty was not reported in this report @xcite .
there are strong @xmath3-rays from @xmath26re produced by the @xmath2re(n,2n)@xmath26re reaction with fast neutrons in fig .
[ fig : largenenough ] .
the half - life of @xmath26re ( @xmath47=38 days ) is long enough to remain after the cooling time .
this fact suggests the possibility that the isomer of @xmath0re might be also produced by the @xmath1re(n,2n)@xmath78re reaction . to estimate the contribution by the @xmath1re(n,2n)@xmath0re@xmath4 reaction
, we evaluate the effective cross section of the @xmath2re(n,2n)@xmath26re reaction by @xmath79 we obtain the result that the effective cross section of the @xmath2re(n,2n)@xmath26re reaction is approximately equal to 0.00078 barn .
since the decay rate of @xmath26re is much higher than @xmath0re@xmath30 , the @xmath3-ray intensities of @xmath26re are stronger than the @xmath0re isomer . assuming that the effective cross section of the @xmath1re(n,2n)@xmath0re is equal to that of the @xmath2re(n,2n)@xmath26re reaction , the contribution to the @xmath0re isomer can be calculated by @xmath80 where @xmath52 means the production ratio of the isomer to the ground state of @xmath0re
, @xmath81 means the isotope abundance ratio in the sample and @xmath82 means the number of the nuclei produced by the neutron - induced reactions .
the contribution of the @xmath1re(n,2n)@xmath0re@xmath30 reaction is approximately equal to 0.0007% relative to that of the @xmath2re(n,@xmath3)@xmath0re@xmath30 reaction .
we therefore conclude that the contamination due to the @xmath1re(n,2n)@xmath0re@xmath4 reaction is negligibly small .
we here estimate the effect of the new path in the s - process using a classical steady - flow model @xcite .
the steady - flow model was extended by ward , newman and clayton @xcite to apply to a branching point where the s - process nucleus is produced by a neutron capture and/or a feeding by @xmath19-decay of a parent nucleus and also destroyed by a neutron capture and/or @xmath19-decay . the abundance of the residual nucleus can be calculated by @xmath83 where @xmath84 means the neutron capture rate @xmath85 @xmath86 @xmath87 or the @xmath19-decay rate @xmath85 ln2/t@xmath88 @xmath87 , @xmath89 is the neutron number density , t@xmath88 is the half life of the @xmath19-decay , @xmath90 means the isotope abundance , @xmath91 and @xmath92 are the nuclear mass numbers which have the relationship , @xmath91=@xmath92 ( for the @xmath19-decay ) or @xmath92 - 1 ( for the neutron capture ) . in the limit that the timescale of the s - process is much longer than a mean time of neutron capture , @xmath93 vanishes and hence the `` steady flow '' becomes a good approximation of the s - process .
the equation in this limit is expressed by @xmath94 applying this equation to the re - os branching point displayed in fig . 1
, we obtain a set of the following equations : @xmath95 @xmath96 @xmath97 @xmath98 @xmath99 @xmath100 we here assume that the internal transition rate between the ground state and the isomer is negligibly lower than the neutron capture rate .
the validity of this assumption should be discussed later .
a typical astrophysical nucleosynthesis environment of the s - process is @xmath8 = 30 kev and @xmath89 = 10@xmath25 @xmath101 .
since the following two relations hold , @xmath102(to @xmath33re ) @xmath103 @xmath104(@xmath1re ) at @xmath1re and ( @xmath104 + @xmath105 ) ( @xmath0re@xmath5 ) @xmath103 @xmath84(to @xmath1re ) at @xmath0re@xmath5 , on the environmental condition for t@xmath88(@xmath1re ) = 43.5 gyr and t@xmath88(@xmath0re@xmath5 ) = 3.718 days , we omit those two slower reaction rates in eqs .
( [ eq:14 ] ) and ( [ eq:15 ] ) .
the situation does not change when the atom of @xmath1re is ionized in the stellar interior during the s - process and the half - life decreases by 9@xmath1110 orders of magnitude @xcite .
two weak s - process contributions are known @xcite .
the unstable nucleus @xmath2w is a branching point where it decays to @xmath2re with a half - life of 75.1 days and it is transmuted to @xmath0w by the neutron capture reaction .
the nucleosynthesis flow from @xmath0w reaches to @xmath1re .
the @xmath19-decay rate at @xmath2w is higher than the neutron capture rate by an order in the case of @xmath89=10@xmath106 @xmath101 , but it depends on the neutron density in the s - process environment .
the ground state of @xmath0re is the second branching point , which decays to both @xmath0os and @xmath0w . using eqs .
( [ eq:12])-([eq:17 ] ) , the s - process abundance of @xmath1re relative to @xmath0os that is produced by the s - process including the new s - process path is calculated by @xmath107 where @xmath52 is defined by @xmath84(to @xmath0re@xmath30)/@xmath84(to @xmath0re@xmath5 ) and @xmath91 is the branching ratio of the @xmath19-decay of @xmath0re@xmath5 , namely @xmath91 = @xmath108/(@xmath108+@xmath109 ) .
we use the branching ratio of 0.9253 for @xmath91 @xcite .
it should be noted that the s - process abundance of @xmath1re has two terms proportional to the ratio of the neutron capture rates leading to the isomer and the ground state of @xmath0re .
the maxwellian - average neutron - capture cross section ( macs ) with quasi - stellar neutrons corresponding to an energy range of @xmath8=8@xmath1130 kev is important for calculating nucleosynthesis under typical s - process environments .
we measure only the neutron - capture cross - sections with the thermal neutrons in the present experiment . in order to estimate the s - process abundance of @xmath1re , we assume first that the ratio of neutron capture rates ,
@xmath102(to @xmath0re@xmath4)/@xmath102(to @xmath0re@xmath5 ) = 0.54 @xmath7 0.11% , does not depend on the neutron energy .
the macs at @xmath8=30 kev for re and os isotopes are taken from the best set evaluated previously @xcite .
the values are 1160 and 418 mbarn for @xmath1re and @xmath0os , respectively .
we assume the neutron density of 10@xmath106 @xmath101 to estimate a neutron capture rate of @xmath84(to @xmath0w ) . substituting the measured cross section ratio , @xmath6 = 0.54% , into eq .
[ eq : final ] , we obtain the s - process abundance of @xmath1re of @xmath10 = 0.23 @xmath7 0.05% relative to @xmath0os .
as an alternative method to estimate the isomer / ground ratio at stellar energy , we calculate the energy dependence of the ratio by the hauser - feshbach statistical - model , which was used in jendl activation cross section file @xcite . in jendl the total capture cross section is calculated using a resonance formula and the statistical - model ( see fig .
we first calculate the capture cross section to the ground state , which is normalized at 100 kev to the value given in jendl .
the isomer / ground ratio is also given in jendl , but its value at the thermal energy is smaller than the presently measured value by a factor of 23 . in order to reproduce the measured isomer / ground ratio at the thermal energy , we adjust the parameters in the present statistical - model calculation in two different ways . in the first method we adjust the @xmath110 transition probability in the @xmath3-decay cascade to enhance the production of the isomer . in the second method
we adjust the level density parameters ( ldp ) which affects the branching of the neutron and @xmath3 emission and affects the average number of cascade @xmath3-transitions .
the calculated isomer / ground ratios are shown in fig .
6 as a function of the incident neutron energy .
the fact that the energy dependence of the two calculated results is almost identical indicates that the calculated ratio is not sensitive to the choice of the parameters .
finally both results give the isomer / ground ratio of the macs at @xmath8=30 kev to be @xmath9 = 1.3 @xmath7 0.8% .
although the two calculated ratios are identical , we assign an error of 50% to the calculated result as a conservative estimate of the uncertainty . by inserting this ratio , @xmath9 = 1.3 @xmath7 0.8% , to eq .
[ eq : final ] we obtain the s - process abundance of @xmath1re , @xmath10 = 0.56 @xmath7 0.35% , relative to @xmath0os .
this value is larger than that at the thermal energy by a factor of 2.4 .
this difference shows that the energy dependence of the ratio is important for the study of the s - process and an experimental measurement of the cross section at the stellar energy is a further subject .
6 shows that the calculated ratio increases drastically above 10 kev .
this tendency indicates that the abundance ratios for the re and os isotopes can be used as a nucleo - thermometer of the s - process due to the strong temperature dependence of the isomer / ground ratio .
finally we would like to discuss a problem concerning the transition probability between the isomer and the ground state .
we assume that this transition probability on a typical s - process condition is ignored . a case of the nucleus
@xmath111ta gives a clue for this problem .
the nucleus @xmath111ta has a similar nuclear structure : the isomer is meta - stable with @xmath45=9@xmath46 ( @xmath47 @xmath24 10@xmath112 yr ) existing in the solar system , while the ground state with @xmath45=1@xmath46 is unstable against the @xmath19-decay with a half - life of 8.15 hr .
this nucleus is proposed to be synthesized through a weak branch of the s - process @xcite and by photodisintegration reactions in supernova explosions @xcite .
the isomer synthesized by either process may decay to the ground state through intermediate states ( ims ) located above the isomer by ( @xmath3,@xmath113 ) reactions in high temperature environment , and subsequently the ground state decays to daughter nuclei through @xmath19-decay .
the residual abundance of @xmath111ta is affected by the transition probability between the ground state and the isomer , which depends on the excitation energy of the ims @xcite .
it is pointed out that the @xmath111ta abundance contributed by the s - process should be smaller than the present solar abundance if the ims might exist below excitation energy of 1 mev @xcite .
this suggests that there is no ims lower than 1 mev .
the ims have been investigated experimentally , but the ims below 1 mev has not been found .
recently the transition probability from the isomer to the ground state has been directly measured in photo - excitation experiment @xcite and the result supports the previous investigation . in the case of @xmath0re , the ims at low excitation energy also play an essential role for the destruction of the synthesized @xmath0re isomer , but these ims of @xmath0re have not been found .
the transition probability between the ground state and isomer via ims depends on the quantum number @xmath114 @xcite .
the quantum number @xmath114 is a projected component of the total angular momentum on the nuclear symmetry axis .
the transition probability increases with increasing the deformation of the nuclear shape @xcite , because the coriolis interaction in high spin states of the deformed nuclei causes the mixing of the states with the different @xmath114 values .
since @xmath111ta is more distant from the magic number , the deformation of @xmath111ta is larger than @xmath0re @xcite .
this fact indicates that the effect of the coriolis interaction on @xmath0re is smaller than that of @xmath111ta .
we therefore assume that the transition between the ground state and the isomer via the ims can be ignored in the typical s - process environment .
the measurement of the transition probability of @xmath0re may be important for the study of the s - process abundance of @xmath1re .
the direct measurement of the transition probability is difficult because the isomer of @xmath0re does not exist in the solar system .
a search for excited states by in - beam @xmath3-ray spectroscopy such as a measurement of excited states of @xmath1re @xcite would be a further subject .
the @xmath1re - @xmath1os chronometer is useful for the study of the age estimate of an r - process event .
the purpose of this subsection is to discuss the effect of the new s - process path on the @xmath1re - @xmath1os chronology .
the contribution from the new s - process path through the @xmath0re isomer has been totally ignored in the past .
we first present the age estimate without the effects of all weak s - process branchings .
in such a case an age of an object which exhibits both the @xmath1re and @xmath1os abundances is calculated by @xmath115 @xmath116 where @xmath117 means the observed abundance and @xmath118 means the abundance contributed from the s - process .
note that we here adopt a simple model taking a sudden approximation that the r - process occurred at the look back time at @xmath119 = @xmath23 only once .
although the sudden approximation is not generally a good approximation for analyses of the solar materials , it is still useful in analyses of primitive meteorites that have recently been found to be affected strongly by a single supernova r - process episode @xcite .
this approximation also applies to an analysis of astronomical date such as the abundances of the r - process elements detected in metal - poor stars with isotope separation because the elements on the surface of the old low - metallicity stars are probably generated in a single nucleosynthesis event @xcite .
we present the age estimation by taking into account the new s - process path to @xmath1re . in this case , eq . ( [ eq : age1 ] ) should be replaced by @xmath120 @xmath121 where we assume that the s - process occurred at different look back time @xmath122 from the r - process episode , i.e. @xmath123 , @xmath118 means the contribution by the s - process , @xmath117 means the observed abundance .
the s - process contribution to @xmath1os should be modified by the change of the nucleosynthesis flow at @xmath0re . the s - process contribution to @xmath1re presented in eq .
( [ eq : final ] ) is calculated from eqs .
( [ eq:12 ] ) @xmath11 ( [ eq:17 ] ) , which contain all the weak s - process paths . in eq .
( [ eq : final ] ) the contribution of the new s - process path is expressed by the term of @xmath52 .
substituting the estimated macs ratio of @xmath9 = 1.3% at @xmath8= 30 kev , we obtain the results shown in tables [ table : age1 ] and [ table : age2 ] .
the neutron capture cross sections at @xmath8 = 30 kev are taken from the previous article @xcite .
those of @xmath18os and @xmath1re are 418 , 874 , 1160 mbarn , respectively .
we use the solar abundances for @xmath124(@xmath1re ) and @xmath124(@xmath1os ) .
@xmath125(@xmath1os ) is calculated from the observed abundance @xmath124(@xmath0os ) = @xmath125(@xmath0os ) . the @xmath124(@xmath0os)/@xmath124(@xmath1re )
ratios in the primitive meteorites or the metal - poor stars may be different from the solar abundance ratio because @xmath124(@xmath1re ) should be enhanced in the sample which is strongly affected by the single r - process event .
we thus introduce a parameter , @xmath126 , which is defined by @xmath127 for allowing the variation of @xmath124(@xmath0os)/@xmath124(@xmath1re ) .
the difference between the calculated ages for @xmath52 = 0 and 1.3% is at most @xmath11 1% in either case @xmath122=5@xmath2110@xmath128 yr ( table [ table : age1 ] ) or 10@xmath22 yr ( table [ table : age2 ] ) . the ages are slightly sensitive to the neutron density .
aoki et al . reported isotope ratios of eu in s - process element rich metal - poor stars and show the possible neutron density range of the s - process environment by comparing their data with the s - process calculation : 10@xmath129 @xmath130 n @xmath130 10@xmath128 @xmath101 @xcite .
thus we also present the age for the neutron density of 10@xmath129 @xmath101 as well as 10@xmath106 @xmath101 in tables [ table : age1 ] and [ table : age2 ] .
the difference between the ages for two different neutron densities is at most @xmath11 2% .
several metal - poor halo stars are thought to form from either the s - process element rich gas @xcite or the r - process element rich material @xcite , which are ejected from progenitor stars .
the parameter @xmath126 can be even larger or smaller than unity .
the primitive meteorites , which are affected by a single supernova r - process , can have the @xmath126 values smaller than unity .
the column of @xmath126 in tables [ table : age1 ] and [ table : age2 ] shows this parameter .
the estimated ages for @xmath52 = 0 and 1.3% under the conditions of @xmath131 1.0 show almost the same results because the absolute nucleosynthesis flow from the s - process decreases with decreasing the @xmath126 value .
in particular , the ages become slightly different with increasing @xmath126 value for the case of @xmath122=10@xmath22 yr ( table [ table : age2 ] ) .
although the effect of the new s - process path seems to be small , the effect is almost the same as those of the other s - process paths under the condition of @xmath126 @xmath130 1.0 .
therefore the direct measurement of the absolute neutron capture cross section of @xmath2re leading to the isomer of @xmath0re at @xmath8=8@xmath1130 kev is important for the @xmath1re - @xmath1os chronometer .
we propose a new s - process path that leads to synthesize @xmath1re and @xmath1os , which form an important nuclear - pair as a nucleo - cosmochronometer of the r - process .
the new path consists of a neutron capture reaction chain from @xmath2re to the isomer of @xmath0re and from the @xmath0re isomer to @xmath1re , namely @xmath2re(n,@xmath3)@xmath0re@xmath4(n,@xmath3)@xmath1re , which has been ignored in all previous studies .
we measure the ratio of the reaction cross sections @xmath2re(n,@xmath3)@xmath0re@xmath4 and @xmath2re(n,@xmath3)@xmath0re@xmath5 with thermal neutrons provided by a nuclear reactor at the jaeri .
we measure @xmath3-rays after the @xmath19-decay from the re samples by hpge detectors and thereby we obtain the ratio of @xmath6 = 0.54 @xmath7 0.11% .
the maxwellian averaged cross section reaction ratio at the stellar energy @xmath8 = 30 kev is estimated with the help of the statistical model calculation .
the result is used to calculate the s - process abundance of @xmath1re relative to the abundance of @xmath0os in the classical steady - state flow model , which yielded a value of @xmath10 = 0.56 @xmath7 0.35% .
the proposed new s - process path does not make any remarkable change in the @xmath1re-@xmath1os chronometer in the sudden approximation when all other possible s - process branchings are included in the calculation .
this confirms the robustness of the @xmath1re-@xmath1os chronometer which applies to the age estimated by the analyses of primitive meteorites that are presumed to be affected strongly by a single supernova r - process episode .
since the @xmath1re atom in these meteorites most likely has not been astrated in the stellar interior after it was produced in the r - process , the chronometer is free from the change of the half - life of ionized @xmath1re .
we however need further experimental study to measure the cross section precisely at the stellar energy @xmath8=8@xmath1130 kev .
we would like to thank k. takahashi , h. utsunomiya and m. fujiwara for valuable discussions .
we also thank the crew of the nuclear reactor at jaeri . this work has been supported in part by grants - in - aid for scientific research ( 15740168 ) of the ministry of education , culture , sports , science and technology of japan , and the mitsubishi foundation .
amari , s. , hoppe , p. , zinner , e. & lewis , r.s .
1992 , apj , 394 , l43 aoki , w. , honda , s. , beers , t. c. & sneden , c. 2003 , apj , 586 , 506 aoki , w. , et al .
2003 , apj , 592 , l67 arlandini , c. et .
al . 1999 , apj .
525 , 886 arnould , m. & goriely , s. 2003 , phr .. , 384 , 1 arnould , m. , takahashi , k. & yokoi , k. 1984 , a&a , 137 , 51 bao , z. y. & k"appeler , f. 1987 , adndt , 36 , 411 baglin , c.m .
2003 , nucl.data sheets * 99 * , 1 burbidge , e. m. , burbidge , g. r. , fowler , w. a. , & hoyle , f. 1957 , remp . , 29 , 548 beer , h. , kppeler , f. , wisshak , k. , & ward , r. a. 1981 , apjs , 46 , 295 belic , d. , et al . 1999 ,
phrel , 83 , 5242 birck , j. l. & allegre , c. j. 1998 , m&ps . , 33 , 647 bosch , f. , et al .
1996 , phrel , 77 , 5190 brown , j.c . &
berman , b.l .
1981 , phrec 23 , 1434 carroll , j.j . , et al .
1991 , phrec , 43 , 1238 cayrel , r. , et al . 2001 , nature , 409 , 91 clayton , d.d . 1964 , apj , 139 , 637 clyton , d. d. 1969 , nature , 224 , 56 conser , k. m. & truran , j. w. 1981 , ap .
space sci .
, 78 , 85 cowan , j.j . , thielemann , f .- k . , &
truran , j.w .
1991 , phre . , 208 , 267 cowan , j.j . ,
et al . 1999 ,
apj , 521 , 194 firestone , r.b .
1991 , nds .. , 62 , 159 firestone , r. b. , et al .
1998 , table of isotopes , eighth edition , wiley interscience fowler , w.a . & hoyle .
f. 1960 , ann .
10 , 280 gallino , r. , _ et al_. 1998 , apj 497 , 388 goriely , s. , & arnould , m. 2001 , a&a , 379 , 1113 hayakawa , t. , _ et al .
_ 2004 , prl 93 , 161102 hershberger , r.l . , _ et al .
_ 1983 , phrec 28 , 2249 hoffman , r.d . , woosely , r.d . & weaver , t.a .
2001 , apj 549 , 1085 holmes , j.a . ,
woosley , s.e . ,
fowler , w.a .
& zimmerman , b.a .
1976 , adndt , 18 , 305 honda , s. , aoki , w. , kajino , t. , ando , h. , beers , t. c. izumiura , h. , sadakane , k. , takada - hidai , m. 2004 , 604 , 474 nuclear data center , japan atomic energy research institute : `` chart of the nuclides 2000 , '' http://wwwndc.tokai.jaeri.go.jp/cn00/index.html ( 2001.12.16 ) .
kppeler , f . ,
gallino , r . ,
busso , m . ,
picchio , g . , ratteri , c.m .
1990 , apj 354 , 630 kppeler , f. , et al . 1991 , apj , 366 , 605 lambert , d.l . & prieto , c.a .
2002 , mon . not .
r. astron . soc .
335 , 325 lee , l .- s . ,
eleid , m.f . & meyer , b.s .
2000 , apj 533 , 998 lindner , m. 1951 , phys .
, 84 , 240 lindner , m. , et al .
1986 , nature , 320 , 256 luck , j. m. , brich , j.l .
& allegre , c.j .
1980 , nature , 283 , 256 luck , j. m. & allegre , c. j. 1983 , nature , 302 , 130 mcellistrem , m.t . , et al . , 1989 , phrec 40 , 591 meyer , b.s . ,
mathews , g.j . ,
howard , w.m .
, woosley , s.e . &
hoffman , r.d .
, 1992 , apj 399 , 656 mohr , p. et al .
2004 , phrec , 69 , 2801 mughabghab , s.f . , 1984 , `` neutron cross sections , vol.1 , neutron resonance parameters and thermal cross sections '' , academic press inc .
nakajima , y. , 1991,jndc wg on activation cross section data : `` jendl activation cross section file , '' proc .
the 1990 symposium on nuclear data , jaeri - m 91 - 032 , p. 43
nazarewicz , w. , riley , m. a. & garrett , j. d. nupha , 512 , 61 otsuki , k . ,
tagoshi , h . ,
kajino , t. , & wanajo , s.y .
2000 , apj 533 , 424 otsuki , k. , mathews , g. j. , kajino , t. 2003 , new astronomy , 8 , 767 pellin , m.j .
2000 , lunar and planet .
31 , 1917 prantzos , n. , hashimoto , m. & nomoto , k. 1990 , apj 234 , 211 qian ,
vogel , p .
, & wasserburg , g.j .
1998 , apj 494 , 285 richter , s. , ott , u. & begemann , f. 1998 , nature , 391 , 261 rutherford , e. 1929 , nature , 123 , 313 ryan , s.g . ,
norris , j.e . & beers , t.c .
1996 , apj , 471 , 254 schaty , h. , _ et al . _ 2002 , apj , 579 , 626 seegmiller , d. w. , et al .
1972 , nupha , 185 , 94 shizuma , t. , et al .
2003 , eur .
j. a , 17 , 159 sneden , c. et al .
2002 , apj 566 , l25 sneden , c. , mcwilliam , a. , preston , g. w. , cowan , j. j. , burris , d. l. , armosky , b. j. 1996 , apj .
467 , 819s sonnabend , k. et al .
2003 , apj , 583 , 506 stolovt , a. , namenson , a.i . & berman , b.l .
1976 , phrec 14 , 965 straniero , o. _ et al .
_ 1995 , apj 440 , l85 takahashi , k. & yokoi , k. 1983 , nupha , 404 , 578 utsunomiya , h. , et al .
2003 , phrvc 67 , 015807 wanajo , s. , et al .
2002 , apj , 577 , 853 ward , r.a . ,
newman , m.j .
& clayton , d.d .
1976 , apj , 31 , 33 winters , r.r . ,
carlron , r.f .. , harvery , j.a . & hill , n.w .
1986 , phrec , 34 , 840 woosley , s.e . &
fowler , w.a .
1979 , apj 233 , 411 woosley , s. e. and weaver , t. a. 1995 , apj * 101 * , 181 woosley , s.e . , wilson , j.r . , mathews , g.j .
, hoffman , r.d .
, & meyer , b.s . , 1994 , apj .
433 , 229 yin , q. , jacobsen , s.b . &
yamashita , k. 2002 , nature 415 , 881 yokoi , k. , takahashi , k. & arnould , m. 1983 , a&a , 117 , 65 yokoi , k. & takahashi , k. 1983 , nature , 305 , 198 .effective neutron - capture cross sections of @xmath2re leading to the @xmath0re isomer .
two samples were irradiated for about six hours by reactor neutrons .
time means the cooling time after the irradiation .
since the cooling time is long enough , the ground state with a short half - life does not survive .
[ cols= " < , < , < , < , < " , ] re .
the nuclei in this mass region are synthesized by the s- and r - processes .
the ground state of @xmath1re decays to @xmath1os with a half - life of about 44 gyr .
the dotted - line is the new s - process path to @xmath1re proposed in this work .
, title="fig : " ] re isomer .
relative @xmath48-ray intensities are indicated .
large number means the energy of the @xmath3-ray and small number means the relative intensity .
the isomer of @xmath0re decays to the ground state through the internal transitions and subsequently the ground state disintegrates to @xmath0os or @xmath0w.,title="fig : " ]
|
we study a new s - process path through an isomer of @xmath0re to improve a @xmath1re-@xmath1os nucleo - cosmochronometer . the nucleus @xmath1re is produced by this new path of @xmath2re(n,@xmath3)@xmath0re@xmath4(n,@xmath3)@xmath1re .
we measure a ratio of neutron capture cross - sections for the @xmath2re(n,@xmath3)@xmath0re@xmath4 and @xmath2re(n,@xmath3)@xmath0re@xmath5 reactions at thermal neutron energy because the ratio with the experimental uncertainty has not been reported . using an activation method with reactor neutrons ,
we obtain the ratio of @xmath6 = 0.54 @xmath7 0.11% . from this ratio
we estimate the ratio of maxwellian averaged cross sections in a typical s - process environment at @xmath8 = 30 kev with a help of the temperature dependence given in a statistical - model calculation because the energy dependence of the isomer / ground ratio is smaller than the absolute neutron capture cross - section .
the ratio at @xmath8=30 kev is estimated to be @xmath9 = 1.3 @xmath7 0.8% .
we calculate the s - process contribution from the new path in a steady - flow model .
the additional abundance of @xmath1re through this path is estimated to be @xmath10 = 0.56 @xmath7 0.35% relative to the abundance of @xmath0os .
this additional increase of @xmath1re does not make any remarkable change in the @xmath1re-@xmath1os chronometer for an age estimate of a primitive meteorite , which has recently been found to be affected strongly by a single supernova r - process episode .
|
based on the full - mission _ planck _
observations of temperature and polarization anisotropies of the cosmic microwave background ( cmb ) radiation , _ planck_-2015 results show that the temperature and polarization power spectra are consistent with the standard spatially - flat six - parameter @xmath5cdm cosmology with a primordial power - law spectrum of adiabatic scalar perturbations .
hereafter we shall call this model the base-@xmath5cdm . on the other hand , the same data , especially the temperature - temperature ( tt ) spectrum reveals some tension with the cmb lensing deflection angle ( @xmath6 ) spectrum reconstructed from the same maps . in details , the lensing amplitude in cmb temperature and polarization spectra , @xmath7 , is in @xmath1 tension with the amplitude of the cmb trispectrum reconstructed lensing deflection angle spectrum , @xmath8 while it is expected that in the base-@xmath5cdm model both these quantities should be equal to unity .
the _ planck _ collaboration finds that , compared with the base-@xmath5cdm model , the base-@xmath5cdm+@xmath3 model can reduce the logarithmic likelihood ( @xmath9 ) and provide a better fit to the data sets with @xmath10 or marginalized constraint @xmath7 @xcite .
more importantly , they find that there is roughly equal preference for high @xmath3 from intermediate and high multipoles ( _ i.e. _ , the ` plik ` likelihood ; @xmath11 ) and from the low-@xmath12 likelihood ( @xmath13 ) with a further small change coming from the priors .
this means that the base-@xmath5cdm+@xmath3 model can provide a better fit than base-@xmath5cdm model against both tt and lowp data sets .
however , the increase in @xmath3 will induce changes on the full sets of cosmological parameters as mentioned in the reference @xcite .
for example , compared with the base-@xmath5cdm fit , the scalar index , @xmath14 , is increased by @xmath15 , the primordial scalar spectrum amplitude , @xmath16 , is reduced by @xmath17 and the effective amplitude of the tt spectrum , @xmath18 , is reduced by @xmath15 . through
the complicated relationship between parameters and their degeneracy the re - ionization optical depth parameter , @xmath19 , falls to @xmath20 , which is roughly in @xmath1 tension with _
planck_-2013 temperature + wmap low-@xmath12 polarization data results of @xmath21 .
inspired by these observations , in this paper , we investigate whether some modifications of gravity can relieve the tension between _ planck _ cmb anisotropy spectra and cmb lensing results .
to do so we will simulate a tension cmb data set that resembles the tension present in the _
planck_-2015 results and we will try to fit the resulting power spectra with different models to have a glimpse of the changes in the parameter that arise because of this tension . we shall show that the modified gravity models that we consider are not capable of alleviating this kind of tension without affecting substantially the growth of structure .
these models will then have problems in fitting simultaneously other cosmological probes .
moreover we will show that , if the amplitude of the reconstructed cmb lensing power spectrum is in tension with the amount of cmb lensing in the anisotropy spectra , these kind of models struggle to relieve it . as we will elaborate later
this happens because modified gravity models predict the same amplitude for the reconstructed cmb lensing potential and the lensing effect on the cmb so that the tension is usually moved to other parameters of the model .
we analyse several modified gravity models against two sets of simulations of cmb spectra ( tt , te , ee ) and cmb lensing spectra ( dd and td ) with the fiducial cosmological parameter equal to the _ planck_-2015 data release best - fit values and _ planck _ blue book beam and noise specifications . the motivation of doing these simulations are mainly two .
first of all the _ planck_-2015 likelihood code and the corresponding spectrum data are not yet publicly available .
second , because the cosmological parameters are degenerate with each other in a complicated way , by using simulations , we can efficiently isolate and study the effects coming from different parameters and their combinations . in the following section
, we will briefly review the futurcmb package @xcite that we used to produce our simulations .
the construction of the power spectrum simulations are mainly made by two essential parts , firstly the mock data generation and the likelihood construction .
for the mock data generation we assume , for simplicity , the raw maps to be composed of gaussian signal ( @xmath22 ) and a uniform gaussian white noise ( @xmath23 ) , @xmath24 where the superscript @xmath25 stands for temperature and e - mode polarization as we do not consider b - mode polarization . the observation phase
then convolves this signal with a gaussian beam characterized by the full width half maximum parameter @xmath26 . on the top of it , we also need to add a gaussian white noise .
up to a normalization , the noise power spectrum can be approximated as @xmath27\;,\ ] ] where @xmath28 models the root mean square of the instrumental noise . in this paper
, we shall adopt the _ planck _
bluebook @xcite beam and noise parameters .
the exact values used in our simulations are listed in table [ tab : experiment ] .
the mock data generator package futurcmb automatically computes the noise power spectra of the lensing deflection angle based on the hu - okamoto @xcite quadratic estimator algorithms from the given temperature and polarization spectra and noises .
then , we feed the fiducial cosmological model into futurcmb . in this paper , we generate the fiducial spectra @xmath29 , @xmath30 , @xmath31 , @xmath32 as well as @xmath33 from the public boltzmann code camb @xcite . in order to mimic _
planck_-2015 results , the cosmological parameter used to produce the fiducial spectra are set to the best - fit values of the base-@xmath5cdm+@xmath3 model to the tt+lowp and tt+lowp+lensing data set of the _ planck_-2015 data release ( see table [ tab : fiducialcp ] ) . with these input
the futurcmb code will generate gaussian - distributed random fields for @xmath34 and estimate the corresponding spectra @xmath35 .
for further details about the numerical implementation of the futurcmb code we refer the reader to @xcite .
the second step is to construct the likelihood for the mock data sets . since it is generated from a gaussian realization , we can write the likelihood function of the data given the theoretical template as @xcite @xmath36\bf a\right)\;,\ ] ] where @xmath37 is the data vector , @xmath38 is the parameter vector , and @xmath39 is the theoretical data covariance matrix . for the detailed construction of the spectrum likelihood
we refer to @xcite .
first of all we check whether these models can reproduce the input amplitude of lensing .
to do so we define an effective lensing amplitude as @xmath40 and we plot it for several choices of the parameters defining our modified gravity models in figure [ fig : aleff ] . from both panels we can see that the effective lensing amplitude is generally a function of scale so that the accuracy of using a scale independent approximation for it is limited by the accuracy of observations .
that is , if observations are not precise enough then the scale independent approximation can work while if the observations have enough sensitivity we would bias parameter estimation by considering it scale independent .
this point will be more and more relevant as the experimental accuracy improves .
the @xmath41 case , shown in panel ( a ) , in particular , displays an @xmath42 that is monotonically increasing as a function of scale . at low multipoles
the agreement with a scale independent model is as good as few percent while in the high-@xmath12 range it is off by @xmath43 depending on the present value of the compton wavelength of the scalaron .
+ the constant @xmath44 and @xmath45 models are shown in panel ( b ) .
unlike the @xmath41 case , @xmath46 has a wide bump or dip around multipoles of few tenths while the scale dependence is somewhat weaker than the previous case .
as expected , in the @xmath47 model the amplitude of lensing is decreased as a consequence of the suppression of growth while the opposite behaviour shows when @xmath48 is smaller than zero .
we fit these models to our mock - a and mock - c data sets and we show the marginalized bound on parameters of interest in figure [ fig : mg ] . from panel
( a ) we can see that there is a strong positive correlation between the scalaron present compton wavelength parameter @xmath49 and @xmath4 .
this effect is well known and corresponds to the fact that as the effective newton constant is increased the growth of matter perturbations is increased as well .
when considering just mock - a our results show that in order to mimic @xmath50 , the marginalized constraint on @xmath49 have to point significantly toward large values of the scalaron compton wavelength .
the marginal bound is then @xmath51 at @xmath52 c.l .. due to the significant degeneracy between @xmath49 and @xmath4 such values of @xmath49 will lead to strong enhancement of the growth rate at relatively small scales making the @xmath4 value too large so that the tension between _ planck _ and and lss surveys , such as cfhtlens @xcite , would become even worse . for this reason when lensing is added in mock - c , the tension in @xmath3 , that for the lensing data set is smaller , pushes the posterior of @xmath49 toward smaller values , making it closer to gr . a similar effect
was also observed for the _
planck_-2015 data set @xcite and in the _
planck_-2013 data @xcite .
as expected a similar result is found also in the @xmath45 model , as shown in panel ( b ) . here
the correlation between @xmath48 is negative due to the fact that negative values of @xmath53 correspond to significant deviations from gr , and consequently to stronger enhancement of cmb lensing .
all the conclusions , previously discussed in the @xmath41 case , apply also to this model with the relevant exception that the enhancement of the growth is not so dramatic as in @xmath41 .
the bound on @xmath4 from the @xmath3 fit , shown in figure [ fig : mg ] as a gray band , is in fact almost compatible with the one in the @xmath45 case .
this is due to the fact that deviations from scale independence of @xmath54 , in this model , are weaker and a constant @xmath53 is more efficient at mimicking a scale independent @xmath3 than @xmath41 models . since a weaker effective newton constant
is disfavoured by the _
planck_-2015 cmb anisotropy data , the best - fit parameter in the @xmath47 model mimic those in the base-@xmath5cdm model and the model result to be compatible with gr .
this conclusion is similar to the one found in @xcite . at last in table [ tab : chisq ]
we list the best - fit @xmath55 for the one parameter extensions of the base-@xmath5cdm model that we investigated in this paper .
from there we can see that the best - fit @xmath55 values in the base-@xmath5cdm+@xmath3 , @xmath41 , @xmath45 models are similar and are noticeably smaller than those from base-@xmath5cdm or @xmath47 models with @xmath56 from the mock - a and @xmath57 from the mock - c data sets .
the best - fit @xmath55 in the @xmath45 case is slightly better than the @xmath41 case and closer to the @xmath3 one because of the weaker scale dependence of @xmath42 .
+ in conclusion , _
planck_-2015 results revealed some interesting tensions between cmb temperature and polarization anisotropies and cmb lensing .
these tensions add to the one reported by lss surveys , like cfhtlens , that seem to favour a smaller @xmath4 .
if this is not due to some unaccounted systematic effects , this might be an indication of exotic physics beyond the base-@xmath5cdm model and as such should be investigated in great detail .
+ in this paper , using a simulation of the _
planck_-2015 data set , we tried to reconcile this tension with some modified gravity models .
in particular we studied whether this tension can be mitigated by @xmath41 models or models characterized by a constant conformal coupling between gravitational and matter perturbations .
we found that the @xmath41 and the @xmath45 models can mimic the role of @xmath3 even if , generally , the amplitude of lensing , with respect to the gr case , is scale dependent . in order to provide a good fit to the _
planck_-like cmb anisotropy spectra , however , these models predict an enhancement of the growth on smaller scales that make the tension in @xmath4 even worse . from these results
we can conclude that the tension between the growth of matter perturbations assessed from cmb power spectra , cmb lensing and lss surveys can be mimicked by modified gravity models but is hardly relieved by simple models beyond the standard @xmath5cdm one .
this in turn suggests to investigate more complicated models , possibly with different time dependencies , to allow for different regimes of growth at the times at which each of these observations is more sensitive .
we thank carlo baccigalupi , julien lesgourgues , matteo martinelli , sabino matarrese , laurence perotto , alessandra silvestri , wessel valkenburg and matteo viel for helpful correspondences and discussions .
is supported by the dutch foundation for fundamental research on matter ( fom ) .
acknowledges partial support from the infn - indark initiative .
thanks the instituut lorentz ( leiden university ) for hospitality while this work was being completed .
p. a. r. ade _ et al .
_ [ planck collaboration ] , arxiv:1502.01589 [ astro-ph.co ] .
l. perotto , j. lesgourgues , s. hannestad , h. tu and y. y. y. wong , jcap * 0610 * , 013 ( 2006 ) [ astro - ph/0606227 ]
. j. tauber _ et al .
_ [ planck collaboration ] , astro - ph/0604069 .
w. hu and t. okamoto , astrophys .
j. * 574 * , 566 ( 2002 ) [ astro - ph/0111606 ] .
a. lewis , a. challinor and a. lasenby , astrophys .
j. * 538 * , 473 ( 2000 ) , [ astro - ph/9911177 ] .
http://camb.info .
m. tegmark , a. taylor and a. heavens , astrophys .
j. * 480 * , 22 ( 1997 ) [ astro - ph/9603021 ] .
a. lewis , phys .
d * 71 * , 083008 ( 2005 ) [ astro - ph/0502469 ] .
a. lewis and a. challinor , phys .
rept . * 429 * , 1 ( 2006 ) [ astro - ph/0601594 ] .
p. a. r. ade _ et al .
_ [ planck collaboration ] , astron .
astrophys . * 571 * , a16 ( 2014 ) [ arxiv:1303.5076 [ astro-ph.co ] ] .
e. calabrese , a. slosar , a. melchiorri , g. f. smoot and o. zahn , phys .
d * 77 * , 123531 ( 2008 ) [ arxiv:0803.2309 [ astro - ph ] ] .
p. a. r. ade _ et al .
[ planck collaboration ] , arxiv:1502.01591 [ astro-ph.co ] .
p. a. r. ade _ et al .
_ [ planck collaboration ] , astron .
astrophys . * 571 * , a17 ( 2014 ) [ arxiv:1303.5077 [ astro-ph.co ] ] .
g. gubitosi , f. piazza and f. vernizzi , jcap * 1302 * , 032 ( 2013 ) [ jcap * 1302 * , 032 ( 2013 ) ] [ arxiv:1210.0201 [ hep - th ] ] .
j. k. bloomfield , .
flanagan , m. park and s. watson , jcap * 1308 * , 010 ( 2013 ) , [ arxiv:1211.7054 [ astro-ph.co ] ] .
y. s. song , w. hu and i. sawicki , phys .
d * 75 * , 044004 ( 2007 ) [ astro - ph/0610532 ] .
l. pogosian and a. silvestri , phys . rev .
d * 77 * , 023503 ( 2008 ) [ erratum - ibid .
d * 81 * , 049901 ( 2010 ) ] [ arxiv:0709.0296 [ astro - ph ] ] .
r. bean , d. bernat , l. pogosian , a. silvestri and m. trodden , phys .
d * 75 * , 064020 ( 2007 ) [ astro - ph/0611321 ] .
a. de felice and s. tsujikawa , living rev .
rel . * 13 * , 3 ( 2010 ) [ arxiv:1002.4928 [ gr - qc ] ] .
b. hu , m. raveri , n. frusciante and a. silvestri , phys .
d * 89 * , no . 10 , 103530 ( 2014 ) [ arxiv:1312.5742 [ astro-ph.co ] ] . m. raveri , b. hu , n. frusciante and a. silvestri , phys . rev .
d * 90 * , no . 4 , 043513 ( 2014 ) [ arxiv:1405.1022 [ astro-ph.co ] ] .
b. hu , m. raveri , n. frusciante and a. silvestri , arxiv:1405.3590 [ astro-ph.im ] . a. lewis and s. bridle , phys .
d * 66 * , 103511 ( 2002 ) [ astro - ph/0205436 ] . c. heymans , l. van waerbeke , l. miller , t. erben , h. hildebrandt , h. hoekstra , t. d. kitching and y. mellier _ et al .
_ , mon . not .
soc . * 427 * , 146 ( 2012 ) [ arxiv:1210.0032 [ astro-ph.co ] ] .
t. erben , h. hildebrandt , l. miller , l. van waerbeke , c. heymans , h. hoekstra , t. d. kitching and y. mellier _ et al .
_ , mon . not .
astron . soc .
* 433 * , 2545 ( 2013 ) [ arxiv:1210.8156 [ astro-ph.co ] ] .
p. a. r. ade _ et al .
_ [ xxx collaboration ] , arxiv:1502.01590 [ astro-ph.co ] .
b. hu , m. raveri , a. silvestri and n. frusciante , arxiv:1410.5807 [ astro-ph.co ] .
j. dossett , b. hu and d. parkinson , jcap * 1403 * , 046 ( 2014 ) [ arxiv:1401.3980 [ astro-ph.co ] ] .
b. hu , m. liguori , n. bartolo and s. matarrese , phys . rev .
d * 88 * , no .
12 , 123514 ( 2013 ) [ arxiv:1307.5276 [ astro-ph.co ] ] .
a. marchini and v. salvatelli , phys .
d * 88 * , no . 2 , 027502 ( 2013 ) [ arxiv:1307.2002 [ astro-ph.co ] ] .
|
_ planck_-2015 data seem to favour a large value of the lensing amplitude parameter , @xmath0 , in cmb spectra .
this result is in @xmath1 tension with the lensing reconstruction result , @xmath2 . in this paper , we simulate several cmb anisotropy and cmb lensing spectra based on _ planck_-2015 best - fit cosmological parameter values and _ planck _ blue book beam and noise specifications .
we analyse several modified gravity models within the effective field theory framework against these simulations and find that models whose effective newton constant is enhanced can modulate the cmb anisotropy spectra in a way similar to that of the @xmath3 parameter . however , in order to lens the cmb anisotropies sufficiently , like in the _
planck_-2015 results , the growth of matter perturbations is substantially enhanced and gives a high @xmath4 value .
this in turn proves to be problematic when combining these data to other probes , like weak lensing from cfhtlens , that favour a smaller amplitude of matter fluctuations .
|
the electrostatic coupling between the @xmath9 shell of rare earths having nonzero orbital angular momentum ( @xmath10 ) and its environment , known as crystalline electric field ( cef ) , greatly influences the physical properties of a rare earth system .
the cef modifies the energy levels of rare earth atoms in a solid and tends to remove the ( @xmath11)-fold degeneracy associated with total angular momentum @xmath12 of the @xmath9 ground state multiplet . the magnetic ordering in compounds having cef - split
nonmagnetic singlet ground state are of particular interest .
the magnetic properties of such systems critically depend on the relative strength of the crystal electric field and the exchange field between the rare earth ions . the magnetic moment associated with the @xmath9 electrons
can be completely quenched if the former dominates over the latter @xcite .
no magnetic ordering can occur in singlet ground state system unless the exchange interaction @xmath13 exceeds a certain critical value relative to the crystal electric field splitting energy @xmath14 between the ground state and the excited state coupled by the matrix element @xmath15 @xcite . the critical value of @xmath13 for the case of singlet ground state and singlet first excited state is given by the condition @xmath16 @xcite .
the system undergoes a self - induced spontaneous ordering above this critical value of @xmath17 .
pr@xmath18tl , tmni@xmath19 , pr@xmath18 in and prru@xmath19si@xmath19 are few of the well known cef - split singlet ground state systems undergoing the self - induced spontaneous ordering @xcite . with a @xmath12 value of 4 for pr@xmath20 ion , the pr - based compounds offer an interesting platform to explore the singlet ground state system .
the action of crystal electric field on the @xmath21-fold degenerate multiplet of pr@xmath20 ion splits it in such a way that a cef - split singlet state is always present for all type of crystallographic symmetry @xcite .
some of the singlet ground state pr - based systems are known to exhibit exotic behavior , such as the observation of unconventional heavy - fermion superconductivity in pros@xmath22sb@xmath23 @xcite .
heavy fermion behavior in pros@xmath22sb@xmath23 is proposed to arise by an unconventional mechanism of the excitonic mass enhancement by low lying crystal field excitations . a similar mechanism is also believed to be the origin of the heavy fermion behavior in singlet ground state system prrh@xmath24b@xmath24c @xcite .
in recent work on induced moment cef singlet ground state system prau@xmath24si@xmath24 , an interesting aspect of competition between the crystal electric field and exchange field was noted . a systematic study of the electronic ground state in prau@xmath19(si@xmath25ge@xmath26)@xmath19 revealed that the cef effect plays an important role in introducing a frustration to the magnetic ground state of the system @xcite .
a novel mechanism due to the dynamic fluctuations of the crystal field levels was proposed to be the origin of the spin - glass behavior in prau@xmath24si@xmath24 @xcite .
inelastic neutron scattering study revealed a cef - split singlet ground state in prau@xmath24si@xmath24 with an excited state doublet at 0.72 mev @xcite .
the exchange interaction in prau@xmath24si@xmath24 is found to be very close to the critical value required for the induced moment magnetic ordering .
as such the dynamic fluctuations of cef levels destabilize the induced moment magnetism in prau@xmath24si@xmath24 leading to spin - glass behavior in this compound .
this novel aspect of competition between the crystal field effect and exchange interaction brings a new perspective to the mechanism of spin - glass behavior .
similar frustrated magnetic ground states and spin - glass behaviors are also observed in cef - split singlet ground state systems prrusi@xmath0 and prrhsn@xmath0 @xcite .
here we report another cef - split singlet ground state system prirsi@xmath0 that exhibits a possible magnetically ordered ground state despite the dominant cef interaction and large splitting energy between the ground state and the first excited state .
the physical properties of prirsi@xmath0 have been investigated by dc magnetic susceptibility @xmath27 , isothermal magnetization @xmath28 , thermo - remnant magnetization @xmath28 , ac magnetic susceptibility @xmath29 , specific heat @xmath30 and electrical resistivity @xmath31 as function of temperature @xmath32 and magnetic field @xmath33 .
the magnetic ground state of prirsi@xmath0 is further probed by muon spin - relaxation ( @xmath7sr ) and inelastic neutron scattering ( ins ) measurements .
the @xmath1 shows a sharp anomaly due to magnetic phase transition at @xmath8 k. however , the @xmath5 data exhibit only a weak anomaly near 12 k. the @xmath7sr finds possible magnetic ordering below 30 k , apparently due to the presence of short range magnetic correlations well above the magnetic phase transition temperature .
a singlet ground state is inferred both from @xmath5 and ins data .
the ins spectra show two sharp inelastic excitations at 8.5 mev and 18.5 mev which are well represented by a cef model .
the analysis of ins data reveal a very large cef splitting energy of 92 k between the ground state singlet and the first excited state doublet . as such the criterion of induced moment magnetism
is not easily fulfilled in prirsi@xmath0 unless there is another kind of interaction in addition to the rkky interaction .
thus , the observation of long range magnetic ordering in singlet ground state prirsi@xmath0 is very surprising and may bring further insight to the interplay of crystal field effect and exchange interaction in singlet ground state system .
a polycrystalline sample of prirsi@xmath0 was prepared by the standard arc melting technique . the stoichiometric mixture of high purity elements ( pr : 99.9% , ir : 99.99% , si 99.999% ) were arc melted on a water cooled copper hearth under the titanium gettered inert argon atmosphere .
the sample was homogenized by flipping and re - melting several times .
the arc melted buttons were wrapped in tantalum foil and annealed at 900 @xmath34c for a week under the dynamic vacuum .
the phase purity and crystal structure of the sample was determined by the powder x - ray diffraction ( xrd ) using cu k@xmath35 radiation .
the dc magnetization measurements were performed by using a commercial superconducting quantum interference device magnetometer ( mpms , quantum design inc . ) and the ac magnetic susceptibility measurements were performed by using a physical properties measurement system ( ppms , quantum design inc . ) .
the specific heat measurements were performed by the relaxation method using the ppms .
the electrical resistivity measurements were performed by the standard four probe ac technique using the ppms .
the muon spin relaxation and inelastic neutron scattering measurements were performed at the isis facility of the rutherford appleton laboratory , didcot , u.k .
the @xmath7sr measurements were carried out on the musr spectrometer with the detectors in a longitudinal configuration .
the powdered sample was mounted on a high purity silver plate using diluted ge varnish and covered with kapton film which was cooled down to 1.0 k in a standard he-4 cryostat with he - exchange gas .
the ins measurements were carried out on the mari time of flight spectrometer .
the powdered sample was wrapped in a thin al - foil and mounted inside a thin walled al - can which was cooled down to 4.5 k inside a top - loading closed cycle refrigerator with he - exchange gas .
the neutrons with incident energies @xmath36 , 20 and 40 mev were used and the scattered neutrons were detected between 3@xmath34 and 135@xmath34 scattering angle .
recorded at room temperature .
the solid line through the experimental points is the two - phase rietveld refinement profile calculated for banisn@xmath0-type tetragonal ( space group @xmath37 ) and thsi@xmath19-type tetragonal ( space group @xmath38 ) structures .
the short vertical bars indicate the bragg peak positions .
the lowermost curve represents the difference between the experimental and calculated intensities.,width=321 ] llccc & + & + & 2 + + & 4.2367(2 ) + & 9.8050(4 ) + & 176.00(1 ) + + + atom & wyckoff & @xmath39 & @xmath40 & @xmath41 + & position + pr & 2a & 0 & 0 & 0 + ir & 2a & 0 & 0 & 0.6568(3 ) + si1 & 2a & 0 & 0 & 0.412(2 ) + si2 & 4b & 0 & 1/2 & 0.2582(9 ) + figure [ fig : xrd ] shows the x - ray diffraction pattern of prirsi@xmath0 which was collected on the powdered sample at room temperature .
the xrd data were analyzed by rietveld structural refinement using the software fullprof @xcite which revealed a banisn@xmath0-type body - centered tetragonal structure ( space group @xmath37 ) for prirsi@xmath0 . the rietveld refinement profile for this structure
is shown in figure [ fig : xrd ] .
the crystallographic parameters obtained from the refinements are listed in table [ tab : xrd ] .
the single phase nature of the sample is evident from the refinement .
a few weak impurity peaks are also present most of which could be accounted by two phase refinement , the impurity phase being prsi@xmath19 ( 1.35 wt% ) . during refinement
the occupancies were kept fixed to unity and the thermal parameters @xmath42 were set to zero .
no significant improvement in the refinement quality parameters or in the @xmath41-positions of atoms was obtained upon allowing variations in the occupancies or @xmath42 .
the impurity phase prsi@xmath19 is reported to order ferromagnetically below 11 k followed by another transition at 7 k @xcite .
the dc magnetic susceptibility of prirsi@xmath0 exhibits sharp anomaly at 12.2 k and a broad hump at 7.5 k ( figure [ fig : chi - lowh ] ) which are very close to the transition temperatures of prsi@xmath19 .
therefore , anomalies observed in magnetic data of prirsi@xmath0 may at first glance be naively connected with the traces of prsi@xmath19 .
however , a closer look at the temperature dependence of the ac magnetic susceptibility , resistivity and specific heat of prsi@xmath19 and prirsi@xmath0 in the vicinity of transition show different behavior and the distinctness of magnetic behavior of prirsi@xmath0 is evident . the ac susceptibility of prsi@xmath19 shows two well pronounced peaks at 11 k and 7 k @xcite .
in contrast the ac susceptibility of prirsi@xmath0 shows sharp peak at 12.2 k and a kink at 13.5 k , and there is no peak near 7 k. the isothermal magnetization of prsi@xmath19 exhibits a rapid increase up to 1.5 t whereas the isothermal magnetization of prirsi@xmath0 exhibits a rapid increase only up to 0.2 t. further , prsi@xmath19 which forms in thsi@xmath19-type tetragonal ( space group _ i4@xmath43/amd _ ) at room temperature undergoes a structural phase transition to gdsi@xmath19-type orthorhombic structure ( space group _ imma _ ) below 180 k and a well defined anomaly due to structural transition is observed in resistivity of prsi@xmath19 @xcite .
however , no such anomaly is observed in our resistivity of prirsi@xmath0 ( see figure [ fig : rho ] ) .
furthermore in our inelastic neutron scattering spectra we do not see the reported @xcite excitations near 9.7 mev , 5.5 mev and 2 mev which would otherwise be due to the presence of prsi@xmath19 , suggesting that the contribution from the impurity phase is of no consequence .
all of these lead to the compelling conclusion that our observed magnetic behavior is intrinsic to the title compound prirsi@xmath0 .
we therefore believe that the presence of tiny amount of impurity does not affect the results and conclusions reached on the bulk properties of prirsi@xmath0 presented here .
the reproducibility of anomalies in the magnetic data have been checked on two different batches of prirsi@xmath0 .
dependence of zero field cooled ( zfc ) and field cooled ( fc ) dc magnetic susceptibility @xmath27 of prirsi@xmath0 in the temperature range 2 30 k measured in a magnetic field @xmath44 t. ( b ) the zfc @xmath1 data measured at different @xmath33.,title="fig:",width=288 ] dependence of zero field cooled ( zfc ) and field cooled ( fc ) dc magnetic susceptibility @xmath27 of prirsi@xmath0 in the temperature range 2 30 k measured in a magnetic field @xmath44 t. ( b ) the zfc @xmath1 data measured at different @xmath33.,title="fig:",width=288 ] dependence of zero field cooled dc magnetic susceptibility @xmath27 of prirsi@xmath0 and its inverse @xmath45 in the temperature range 2300 k measured in a magnetic field @xmath46 t. the solid line represents the fit to curie - weiss law for @xmath47 k and the dashed line is an extrapolation.,width=307 ]
figures [ fig : chi - lowh ] and [ fig : chi ] show the temperature dependence of zero field cooled ( zfc ) and field cooled ( fc ) dc magnetic susceptibility @xmath27 of prirsi@xmath0 measured at different applied magnetic field @xmath33 .
a magnetic phase transition at temperature @xmath8 k is inferred from the low-@xmath32 @xmath1 data . at low @xmath33 ,
e.g. at @xmath48 t , the low-@xmath32 @xmath1 [ figure [ fig : chi - lowh](a ) ] exhibits a rapid increase below 15 k and peak at 12.2 k with a broad hump near 7.5 k. an irreversibility is also observed between the zfc and fc @xmath1 data below 12.2 k. such a behavior of @xmath1 reflects the presence of ferromagnetic correlation . an increase in @xmath33 shifts
the @xmath1 peak position towards the lower temperature , e.g. , at @xmath49 t the 12.2 k peak shifts to 11.1 k and 7.5 k anomaly to @xmath50 k , and reduces the magnitude of @xmath27 [ figure [ fig : chi - lowh](b ) ] . at @xmath51 t
both the anomalies merge together and appear as a very broad anomaly centered around 7 k. at high temperatures the @xmath1 data follow the curie - weiss behavior , @xmath52 . from a linear fit of the inverse magnetic susceptibility @xmath53 data measured at @xmath54 t in the temperature range 150
k @xmath55 300 k ( the solid line in figure [ fig : chi ] ) we obtain the curie constant @xmath56 emuk / mol and paramagnetic weiss temperature @xmath57 k. the effective moment @xmath58 obtained from @xmath59 using the relation @xmath60 .
the effective moment is slightly smaller than the theoretically expected value of effective moment for pr@xmath20 ions ( @xmath61 ) .
the positive @xmath62 reflects a dominant ferromagnetic interaction .
of prirsi@xmath0 as a function of magnetic field @xmath33 measured at the selected temperatures of 2 , 8 , and 15 k. inset : an expanded plot of @xmath2 isotherm at 2 k for @xmath63 t.,width=288 ] figure [ fig : mh ] shows the @xmath33 dependence of isothermal magnetization @xmath28 of prirsi@xmath0 measured at selected temperatures of 2 , 8 , and 15 k. at low-@xmath32 ( e.g. , at 2 k and 8 k ) the @xmath2 isotherm exhibits ferromagnetic - like spontaneous magnetization .
it is seen that at 2 k the @xmath28 increases very rapidly initially for @xmath64 t and then the increase rate gradually decreases and eventually for @xmath65 t the @xmath2 isotherm is almost linear in @xmath33 .
the @xmath2 isotherm at 15 k is almost linear in @xmath33 for @xmath66 t. the nonlinearity for @xmath67 t can be due to the presence of short range correlations above @xmath68 and/or crystal field effect .
further we see that at 2 k and 7 t , the magnetization attains a value of @xmath69 , which is very low compared to the theoretical value of saturation magnetization @xmath70 = 3.2@xmath71 for pr@xmath20 ions .
this reduction in @xmath28 can be attributed to the crystal field effect inferred from the specific heat and resistivity data in next section . in the absence of cef ,
the magnetization is given by @xcite @xmath72 where @xmath73 is the brillouin function given by @xmath74 with @xmath75 for pr@xmath20 land @xmath76 factor @xmath77 . from eqs .
( [ eq : m1 ] ) and ( [ eq : m2 ] ) for @xmath78 k and @xmath79 t we obtain @xmath80 which is much higher than the observed @xmath81 .
this suggests that the cef is responsible for the reduced @xmath28 in prirsi@xmath0 .
a reduced value of @xmath82 was also observed in isostructural prrusi@xmath0 @xcite and @xmath83 is observed in isostructural prrhsi@xmath0 at @xmath78 k and @xmath79 t @xcite . on the other hand for isostructural prcual@xmath0 @xmath84 at @xmath78 k and @xmath79 t @xcite and that for cubic prrhsn@xmath0 is @xmath85 @xcite .
dependence of thermo - remnant magnetization @xmath28 of prirsi@xmath0 measured at 2 k after switching off the cooling magnetic field of 0.05 t. the solid curve is the fit to the superposition of a stretched exponential and a constant term .
inset : @xmath3 data plotted on a semi - log scale.,width=288 ] figure [ fig : trm ] shows the time @xmath86 dependence of thermo - remnant magnetization @xmath28 of prirsi@xmath0 . the time dependence of @xmath28 was recorded by field cooling the sample in a magnetic field of 0.05 t from 50 k to 2 k , and then the field cooled isothermal remnant magnetization was measured at 2 k after switching off the magnetic field .
as seen from figure [ fig : trm ] the @xmath3 decays very slowly with time and remains nonzero even after 62000 s. a semi - logarithmic plot of @xmath3 data is shown in the inset of figure [ fig : trm ] which clearly shows that @xmath3 data do not follow the logarithmic decay , @xmath87 .
however , the @xmath3 data could be well fitted by a superposition of a stretched exponential and a constant term , @xcite @xmath88 \label{eq : trm}\ ] ] where @xmath89 is the mean relaxation time .
the constant term @xmath90 is expressed as the longitudinal spontaneous magnetization coexisting with the frozen transverse spin component @xcite .
a fit of @xmath3 data by eq .
( [ eq : trm ] ) is shown by the solid curve in figure [ fig : trm ] which is obtained for @xmath91 emu / mole , @xmath92 emu / mole , @xmath93 s and @xmath94 . a similar analysis of @xmath3 for spin - glass system prrusi@xmath0 have yielded @xmath95 emu / mole , @xmath96 emu / mole , @xmath97 s and @xmath98 @xcite .
the mean relaxation time @xmath89 of prirsi@xmath0 is extremely large and reflect a very slow decay of @xmath28 .
dependence of ( a ) real @xmath99 and ( b ) imaginary @xmath100 parts of the ac magnetic susceptibility @xmath29 of prirsi@xmath0 measured at different frequencies from 11 hz to 919 hz in an applied ac magnetic field of 1.5 mt .
inset in ( b ) : an expanded plot of @xmath101 at 11 hz.,title="fig:",width=288 ] dependence of ( a ) real @xmath99 and ( b ) imaginary @xmath100 parts of the ac magnetic susceptibility @xmath29 of prirsi@xmath0 measured at different frequencies from 11 hz to 919 hz in an applied ac magnetic field of 1.5 mt .
inset in ( b ) : an expanded plot of @xmath101 at 11 hz.,title="fig:",width=288 ] figure [ fig : chiac ] shows the @xmath32 dependence of the real @xmath99 and imaginary @xmath100 parts of ac susceptibility @xmath29 data of prirsi@xmath0 measured at three different frequencies @xmath102 under an applied ac magnetic field of 1.5 mt . as seen from figure [ fig : chiac](a ) a sharp anomaly at 12.2 k is observed in @xmath103 at 11 hz together with a kink at 13.5 k and a hardly detectable hump near 7.5 k. no noticeable change is observed in the anomaly temperature with an increase in @xmath102 for @xmath104 khz except that the 13.5 k anomaly becomes more pronounced with increasing frequency .
a weak temperature shift in both 13.5 k and 12.2 k anomalies towards the lower temperature is observed with the increase in @xmath102 for @xmath105 khz ( data not shown ) .
the magnitude of @xmath99 decreases with increasing frequency and becomes negative for @xmath106 hz ( data not shown ) . such a behavior of @xmath99 with increasing @xmath102 can be attributed to skin effect due to the confinement of ac magnetic flux in a surface layer which is possible because of the high conductivity of the sample at low temperature ( @xmath107 cm ) . for a metal
the skin depth @xmath108 , @xcite therefore at high frequency the ac flux is confined to a skin layer only , leading to the observed diamagnetic behavior in @xmath99 at higher frequencies . on the other hand , @xmath101 exhibits sharp anomaly near 12.2 k at 11 hz , however , there is no anomaly near 13.5 k corresponding to that observed in @xmath99 .
the hump near 7.5 k is very weak . at increased frequency
an anomaly appears at 13 k for @xmath109 hz .
the magnitude of @xmath100 increases very rapidly with increase in frequency . for @xmath100 too a weak temperature shift in both 12.2 k and 13 k anomalies towards the lower temperature is observed with the increase in @xmath102 for @xmath105 khz ( data not shown ) .
dependence of specific heat @xmath30 of prirsi@xmath0 measured in zero magnetic field in the temperature range 2300 k. inset : an expanded view of low-@xmath32 @xmath5 data below 16 k. the solid curve in the inset is the fit by @xmath110 below 7 k and the dashed line is an extrapolation .
( b ) the magnetic contribution to specific heat @xmath111 .
the solid curve represents the crystal electric field contribution to specific heat according to the crystal field level scheme deduced from the inelastic neutron scattering data .
inset : magnetic contribution to entropy @xmath112 .
the arrow marks the transition temperature.,title="fig:",width=288 ] dependence of specific heat @xmath30 of prirsi@xmath0 measured in zero magnetic field in the temperature range 2300 k. inset : an expanded view of low-@xmath32 @xmath5 data below 16 k. the solid curve in the inset is the fit by @xmath110 below 7 k and the dashed line is an extrapolation .
( b ) the magnetic contribution to specific heat @xmath111 .
the solid curve represents the crystal electric field contribution to specific heat according to the crystal field level scheme deduced from the inelastic neutron scattering data .
inset : magnetic contribution to entropy @xmath112 .
the arrow marks the transition temperature.,title="fig:",width=288 ]
figure [ fig : hc ] shows the @xmath32 dependence of the specific heat @xmath30 data of prirsi@xmath0 measured at constant pressure p. as shown in the inset of figure [ fig : hc](a ) , the @xmath5 data present a weak anomaly near 12 k. the low-@xmath32 @xmath5 data follow @xmath110 , where @xmath113 and @xmath114 are coefficients to the electronic and lattice contributions to specific heat , respectively . on fitting the @xmath5 data of prirsi@xmath0 by this relation in @xmath115 k we obtain the sommerfeld coefficient @xmath116 mj / molek@xmath117 and @xmath118 mj / molek@xmath119 .
the fit is shown in the inset of figure [ fig : hc](a ) .
the debye temperature @xmath120 estimated from @xmath114 using the relation @xmath121 , where @xmath122 is the molar gas constant and @xmath123 is the number of atoms per formula unit , is @xmath120 = 227(1 ) k. the magnetic contribution to specific heat @xmath124 is shown in figure [ fig : hc](b ) .
the @xmath124 was estimated by subtracting off the lattice contribution equal to the specific heat of isostructural laossi@xmath0 @xcite .
the effect of crystal electric field is reflected as a broad schottky - type anomaly centered around 40 k in @xmath124 .
the solid curve in figure [ fig : hc](b ) represents the crystal field contribution to specific heat according to the cef level scheme obtained from the analysis of inelastic neutron scattering data in sec .
[ sec : ins ] .
the magnetic contribution to entropy @xmath125 was obtained by integrating the @xmath126 versus @xmath32 plot and is shown in the inset of figure [ fig : hc](b ) .
an extremely small value of @xmath127 at @xmath68 which is only @xmath128% of @xmath129 suggests a cef - split singlet ground state in prirsi@xmath0 and is confirmed by the ins data ( section .
[ sec : ins ] ) .
dependence of electrical resistivity @xmath31 of prirsi@xmath0 measured in zero magnetic field .
inset : @xmath31 as a function of magnetic field @xmath33 at 2
k. the solid line is a guide to eye.,width=321 ] the @xmath32 dependence of the electrical resistivity @xmath31 of prirsi@xmath0 is presented in figure [ fig : rho ] .
the low value of residual resistivity @xmath130 cm at 2 k , and a high value of residual resistivity ratio @xmath131 reveal the good quality of our sample .
a metallic character is evident from the @xmath32 dependence of @xmath31 .
the effect of crystal electric field is reflected by a broad curvature in @xmath6 .
the @xmath33 dependence of @xmath31 at 2 k is shown in the inset of figure .
[ fig : rho ] .
though noisy , the basic trend of @xmath132 is quite clear from the data , revealing that the @xmath31 increases linearly with increasing @xmath33 .
the @xmath31 is found to increase from its value of @xmath133 cm at @xmath134 t to @xmath135 cm at @xmath136 t thus revealing a large positive magnetoresistance ( mr ) of about 200% at 2 k at 9 t.
( color online ) the zero field @xmath7sr spectra of prirsi@xmath0 for few representative temperatures .
the solid curves represent the fits to the data by the function described in the text.,width=288 ] ( color online ) the temperature @xmath32 dependence of ( a ) the initial asymmetries @xmath137 and @xmath138 , ( b ) the internal field @xmath139 , ( c ) the depolarization rate @xmath140 , and ( d ) the depolarization rate @xmath141 obtained from the analysis of zero field @xmath7sr spectra of prirsi@xmath0 collected at various temperatures .
the vertical dotted lines mark @xmath68 observed in the magnetic susceptibility in figure [ fig : chi - lowh].,width=316 ] figure [ fig : musr1 ] shows the @xmath7sr spectra of prirsi@xmath0 at few representative temperatures collected in zero field while warming the sample . at @xmath142
k the @xmath7sr spectra are well described by a simple exponential ( also called lorentzian ) decay function as the muons sense only paramagnetic fluctuations at these temperatures . however , for @xmath143 k the @xmath7sr spectra are best described by an oscillating function with a lorentzian envelope @xmath144 where @xmath145 and @xmath146 are the initial asymmetries , @xmath147 is the precession frequency ( @xmath148 is the gyromagnetic ratio of the muon and @xmath149 is the internal field at the muon site ) , @xmath150 is the phase , @xmath140 and @xmath141 are the depolarization rates , and bg is a constant background .
the phase angle accounts for the small misalignment of incident muon spins with respect to incident direction and small effect of zero time correction .
the fits of @xmath7sr data by this decay function are shown by the solid curves in figure [ fig : musr1 ] .
the parameters obtained from the fit of @xmath7sr data are plotted in figure [ fig : musr2 ] as a function of temperature .
the @xmath32 dependence of the asymmetries in figure [ fig : musr2](a ) show a clear gain in asymmetry @xmath145 and loss of asymmetry a@xmath24 at @xmath143 k which is accompanied by an increased frequency and hence the internal field at the muon site as shown in figure [ fig : musr2](b ) .
the temperature dependences of the muon depolarization rates @xmath140 and @xmath141 are shown in figures [ fig : musr2](c ) and ( d ) .
an increase is observed in @xmath140 at @xmath143 k again indicating that the muons sense a possible magnetic phase transition near 30 k. the appearance of internal field at @xmath143 k indicates that the muons sense a magnetically ordered state below 30 k. however , the observation of magnetically ordered state by muons at @xmath151 k is much higher than the @xmath152 k at which a bulk magnetic phase transition was inferred from the magnetic susceptibility and specific heat data above .
we suspect that the muons are sensing short range ordering at temperatures well above the long range magnetic ordering occurs in prirsi@xmath18 .
further , the @xmath32 dependences of the fitting parameters @xmath145 , @xmath146 , @xmath139 , @xmath140 and @xmath141 clearly show that all these are nearly temperature independent over @xmath153 distinguishing the possible long range ordered phase below @xmath68 and short range ordered phase in @xmath154 k.
( color online ) the color - coded contour map of inelastic neutron scattering intensity of prirsi@xmath0 measured with incident energy @xmath155 = 20 mev ( bottom ) and 40 mev ( top ) at 4.5 k and 20 k plotted as a function of energy transfer @xmath156 and wave vector transfer @xmath157 .
the intensity is in units of mb / srmevf.u.,width=316 ] ( color online ) the color - coded contour map of inelastic neutron scattering intensity of prirsi@xmath0 ( top ) and lairsi@xmath0 ( bottom ) measured with incident energy @xmath155 = 20 mev at 4.5 k plotted as a function of energy transfer @xmath156 and wave vector transfer @xmath157 .
the intensity is in units of mb / srmevf.u.,width=192 ] the ins spectra of prirsi@xmath0 collected with neutrons of incident energies @xmath158 and 40 mev for scattering angles between 3@xmath34 and 135@xmath34 at 4.5 k and 20 k are shown in figure [ fig : ins - prirsi3 ] as a color - coded map of the intensity , energy transfer versus momentum transfer .
the ins data have been corrected for the background signal and the data have been put into the absolute unit ( mb / sr / mev ) by using vanadium standard measured with the identical conditions .
the color - coded intensity map spectra of prirsi@xmath0 and lairsi@xmath0 obtained with @xmath159 mev at 4.5 k are presented in figure [ fig : ins - prlairsi3 ] .
two sharp inelastic excitations near 8.5 mev and 18.5 mev are clearly observed in the ins spectra of prirsi@xmath0 at 4.5 k which remain nearly invariant from 4.5 k to 50 k. the temperature invariance of inelastic excitations suggests that they result from the crystal field effects .
thus , ins spectra do not show any additional excitation arising from the spin wave contribution due to the magnetic ordering below the transition temperature .
further , no magnetic bragg peak is observed in @xmath160-cuts ( integrated over the elastic line from @xmath161 mev to 2 mev ) at 4.5 k. this may suggest that the ordered state moment of the pr ion is too weak to be detected in the present data .
( color online ) the estimated magnetic scattering from prirsi@xmath0 at different temperatures for momentum transfer @xmath162 @xmath163 measured with incident energy @xmath164 mev .
the solid lines are the fits based on the crystal electric field model and the dashed and dash - dotted lines are the components of the fit .
note : the temperature independent scattering on neutron energy loss side is the background from the closed cycle refrigerator.,width=288 ] ( color online ) the estimated magnetic scattering from prirsi@xmath0 at different temperatures for momentum transfer @xmath165 @xmath163 measured with incident energy @xmath166 mev .
the solid lines are the fits based on the crystal electric field model and the dashed and dash - dotted lines are the components of the fit.,width=288 ] ( color online ) the estimated magnetic scattering from prirsi@xmath0 at different temperatures for momentum transfer @xmath167 @xmath163 measured with incident energy @xmath168 mev .
the solid lines are the fits based on the crystal electric field model and the dashed and dash - dotted lines are the components of the fit.,width=288 ] ( color online ) the @xmath160 dependence of total intensity of prirsi@xmath0 integrated between 7 mev and 10 mev at 4.5 k for incident energy @xmath155 = 20 mev and 40 mev .
the solid line represents the square of the pr@xmath20 magnetic form factor , scaled to 102 at @xmath169.,width=302 ] the dynamic structure factor @xmath170 of neutron scattering of unpolarized neutrons is given by , @xcite @xmath171 where @xmath172 is the magnetic form factor of pr@xmath20 , @xmath173 is the static bulk susceptibility and @xmath174 is the normalized spectral function ( usually lorentzian ) .
the static susceptibility is given by @xmath175 with @xmath176 and @xmath177 here @xmath178 is the population of cef level with wave function @xmath179 and energy eigenvalue @xmath180 , @xmath181 is the partition function , @xmath182 is the @xmath39 , @xmath40 or @xmath41 component of angular momentum operator , and @xmath183 is the energy difference between the cef levels @xmath179 and @xmath184 . the magnetic scattering of prirsi@xmath0 for different incident energies estimated by subtracting off the phonon contributions using the scattering from lairsi@xmath0 , @xmath185 is shown in figures [ fig : ins - smag1 ] , [ fig : ins - smag2 ] and [ fig : ins - smag3 ] . the scaling factor
@xmath186 is the ratio of the scattering cross sections of prirsi@xmath0 and lairsi@xmath0 .
the @xmath187 shown in figure [ fig : ins - smag3 ] clearly demonstrates the observation of two inelastic excitations centered around 8.5 mev and 18.5 mev . the @xmath160-dependent integrated intensity between 7 mev and 10 mev at 4.5 k
is shown in figure [ fig : ins - form - factor ] .
the integrated intensity follows the square of pr@xmath20 magnetic form factor @xmath188 suggesting that the inelastic excitation results mainly from single - ion cef - transitions .
it is to be noted that due to very weak intensity of the 18.5 mev peak and also due to the presence of phonon peak at the same energy as seen in high-@xmath160 , it was not possibile to analyze the @xmath160-dependent intensity of this peak
. further support of magnetic origin of the 18.5 mev peak comes from an initial decrease ( at low-@xmath160 ) of the energy integrated ( between 17 mev and 21 mev ) @xmath160-dependence of the intensity .
.[tab : cef ] crystal field parameters @xmath189 and splitting energies @xmath190 of excited states ( with respect to ground state , @xmath191 ) obtained from the refinement of the ins data of prirsi@xmath0 .
[ cols="^,^ " , ] ( color online ) crystal electric field level scheme and @xmath192-electron wave functions of pr@xmath20 ions in prirsi@xmath0 deduced from the analysis of inelastic neutron scattering data .
the transitions from the ground state to the excited states that contribute to the observed excitations are shown by arrows.,width=288 ] the ins data were analyzed by a model based on crystal electric field effect which fits the ins data very well confirming that the two sharp inelastic excitations observed near 8.5 mev and 18.5 mev arise from the crystal field effect .
the crystal field hamiltonian for a rare earth ion in a tetragonal symmetry ( point group @xmath193 ) is given by @xmath194 where @xmath189 are the phenomenological crystal field parameters and @xmath195 are the stevens operators .
the crystal electric field removes the nine - fold degeneracy of @xmath9 ground multiplet of pr@xmath20 ( @xmath196 ) leading to five singlets ( @xmath197 , @xmath198 , @xmath199 , @xmath200 , @xmath201 ) and two doublets ( @xmath202 , @xmath203 ) in tetragonal symmetry , where the @xmath204 s are the irreducible representations of the point group .
the wave functions for @xmath204 are shown in figure [ fig : cef - level ] .
in cef analysis it is very important to find a good starting set of the cef parameters .
we have used monte - carlo method given in focus program @xcite , which allowed us to find the best starting value of the cef parameters .
then we simultaneously fitted all different @xmath155 and temperatures data sets .
our simultaneous fit of nine sets of ins data of prirsi@xmath0 for @xmath205 mev , 20 mev and 40 mev at 4.5 k , 20 k and 50 k are shown by solid lines in figures [ fig : ins - smag1 ] , [ fig : ins - smag2 ] and [ fig : ins - smag3 ] . a nice agreement between the ins data and the fit
is seen from figures [ fig : ins - smag1 ] , [ fig : ins - smag2 ] and [ fig : ins - smag3 ] .
the phenomenological crystal field parameters @xmath189 obtained from the simultaneous fit are listed in table [ tab : cef ] .
the cef wave functions and the cef level scheme corresponding to the @xmath189 parameters in table [ tab : cef ] are shown in figure [ fig : cef - level ] .
the transitions from the ground state to the excited states that contribute ( has nonzero matrix element ) to the observed excitations are shown by arrows in figure [ fig : cef - level ] .
the ground state is a singlet lying below an excited doublet at 92.3 k. the schottky contribution to specific heat due to cef can be estimated from @xcite @xmath206 ^ 2 \bigg\ } \times \bigg[\sum_{i}g_i e^{-\delta_i / t}\bigg]^{-2}\nonumber\end{aligned}\ ] ] where @xmath207 are the degeneracies of cef levels with energies @xmath190 .
the cef specific heat corresponding to the cef parameters in table [ tab : cef ] calculated according to eq .
( [ c_sch ] ) using the software amos @xcite is shown in figure [ fig : hc](b ) which shows a nice agreement with the @xmath124 data for @xmath208 , thus supporting the cef level scheme deduced from the ins data .
the single crystal @xmath209 calculated according to eq .
( [ eq : chi-0 ] ) using the software amos is shown in figure [ fig : ins - chi ] .
the magnitude of @xmath209 is much smaller than the measured @xmath1 in figure [ fig : chi ] ( can be attributed to the presence of ferromagnetic correlation ) , however @xmath209 reveals strong anisotropy , @xmath210 over the entire temperature range up to 300 k , revealing an easy plane behavior of magnetic moments .
( color online ) the calculated single crystal susceptibility @xmath1 of prirsi@xmath0 using cef parameters given in table [ tab : cef].,width=288 ]
the electronic and magnetic properties of prirsi@xmath0 indicate that it shows a magnetic phase transition with a singlet ground state .
the inelastic neutron scattering data , @xmath124 and @xmath125 consistently reveal a singlet ground state in prirsi@xmath0 . on the other hand @xmath1 and
@xmath5 reveal a magnetic phase transition at @xmath152 k. magnetic phase transition is further inferred from @xmath7sr results which show magnetic ordering at an elevated temperature of 30 k , possibly due to the presence of short range correlations well above the long range magnetic ordering inferred from the specific heat and magnetic measurements .
the nature of this magnetic ordering in prirsi@xmath0 is not clear .
however , considering the positive @xmath62 , irreversibility between the zfc and fc @xmath27 , and very slow relaxation of @xmath3 , dominance of ferromagnetic correlations is evident .
many singlet ground state systems are known to exhibit long range magnetic ordering due to the exchange - generated admixture of ground state singlet and the excited cef levels to meet the criterion of induced moment magnetism based on the relative strength of exchange interaction and splitting energy @xcite .
however , the crystal field splitting of 92 k between the ground state and the first excited state doublet of prirsi@xmath0 is very high therefore a very large exchange energy is required for an induced moment magnetic ordering in this compound .
the singlet ground state system prru@xmath19si@xmath19 crystallizing with body - centered tetragonal structure orders ferromagnetically below @xmath211 k with splitting energy of 26 k ( 2.25 mev ) between the ground state singlet and first excited state singlet @xcite .
prni@xmath19si@xmath19 is another singlet ground state system with body - centered tetragonal structure that exhibits amplitude modulated long range antiferromagnetic order below @xmath212 k and has first excited singlet at 38 k ( 3.3 mev ) @xcite .
both prru@xmath19si@xmath19 and prni@xmath19si@xmath19 exhibiting long range induced moment magnetic ordering with singlet ground state system have low cef splitting energies , therefore exchange - generated admixture of low - lying crystal field levels is achieved .
in contrast , the admixture of low - lying crystal field levels by exchange interaction in prirsi@xmath0 with large cef splitting energy of 92 k is unexpected and the observation of magnetic ordering in this singlet ground state system is not easy to understand
. induced moment magnetism has also been reported in singlet ground state systems with splitting energy larger than that of prru@xmath19si@xmath19 and prni@xmath19si@xmath19 but somewhat smaller than that of prirsi@xmath0 .
cubic structured singlet ground state compounds pr@xmath18tl and pr@xmath18 in exhibit long range induced moment magnetic ordering with @xmath213 k ( ferromagnetic ) and @xmath214 k ( antiferromagnetic ) , respectively @xcite .
they both have triplet as excited state with splitting energies of 79 k ( 6.8 mev ) and 73 k ( 6.3 mev ) , respectively @xcite .
face centered cubic ( fcc ) pr also shows an induced moment ferromagnetism ( @xmath215 k ) with larger splitting energy of 84 k ( 7.2 mev ) between singlet ground state and triplet excited state @xcite .
orthorhombic prcoal@xmath216 with singlet ground state is reported to exhibit an amplitude modulated antiferromagnetic ordering below @xmath217 k possibly with a large splitting energy of about 100 k between the ground state and the first excited state as deduced from the specific heat data , that remains to be verified by inelastic neutron scattering @xcite . for a singlet ground system like prirsi@xmath0 the mean - field critical temperature for self - induced moment ordering
is given by @xcite @xmath218 \right\}^{-1 } \label{eq : tc}\ ] ] where @xmath15 is the matrix element between the ground state singlet and the excited state doublet and @xmath219 is the degeneracy of the excited state which for a doublet is @xmath220 .
our analysis of ins data gives @xmath221 = 10.46 and @xmath14 = 7.95 mev , therefore according to eq .
[ eq : tc ] , in order to have an induced moment ordering at 12.2 k the critical exchange required is @xmath222 mev . on the other hand the exchange energy estimated from @xmath62 according to @xmath223 with @xmath224 , and @xmath225 is boltzmann s constant , is @xmath226 mev for @xmath57 k as determined in sec . [
sec : magnetization ] from @xmath1 .
this value of @xmath227 mev is much smaller than the @xmath222 mev required .
even though a more reliable estimate of @xmath13 is necessary , it is apparent from the @xmath13 values that there should be no spontaneous induced moment ordering in prirsi@xmath0 .
thus the observation of magnetic ordering in this singlet ground state prirsi@xmath0 is not understood on the basis of relative strength of @xmath13 and @xmath14 .
the physical properties of prirsi@xmath0 which forms in banisn@xmath0-type tetragonal crystal structure ( space group @xmath37 ) are reported .
a magnetic phase transition at @xmath152 k of ferromagnetic nature considering the irreversibility between the zfc and fc susceptibility and very slow relaxation of thermoremnant magnetization is inferred from the magnetic susceptibility and specific heat data .
similar irreversibility in @xmath1 , strong dependence of @xmath27 on applied @xmath33 and slow relaxation of @xmath3 have also been observed in spin - glass systems prrusi@xmath18 and prrhsn@xmath18 @xcite .
however , the ac magnetic susceptibility and @xmath7sr measurements do not show signatures of spin - glass behavior in prirsi@xmath0 .
presence of strong crystal field effect is revealed by a broad schottky - type anomaly in the magnetic part of specific heat and a broad curvature in electrical resistivity .
a cef - split singlet ground state is inferred from the extremely small value of magnetic entropy below @xmath68 .
the ins data confirms the cef - split singlet ground state in prirsi@xmath0 .
two sharp inelastic excitations are observed at 8.5 mev and 18.5 mev in ins spectra .
the ins data are well explained by a model based on crystal electric field effect .
the crystal field level scheme and wave functions have been evaluated .
a large cef splitting energy of 92 k between the ground state singlet and the first excited doublet is found from the ins data consistent with the magnetic specific heat .
the @xmath7sr data reveal possible magnetic ordering below 30 k which is much higher than the @xmath68 possibly due to the presence of short range correlations well above the long range magnetic ordering .
thus @xmath1 , @xmath5 and @xmath7sr provide conclusive evidence for magnetic ordering in prirsi@xmath0 .
however , considering the large splitting energy of 92 k the magnetic ordering in singlet ground state prirsi@xmath0 is unexpected .
very recently we also observed a similar magnetic behavior below 15 k in singlet ground state prrhsi@xmath18 which also has a large cef splitting energy of 81 k between the ground state singlet and the first excited doublet as inferred from the heat capacity analysis @xcite .
the difference between these two systems is that the specific heat of prrhsi@xmath18 does not show a corresponding anomaly like prirsi@xmath0 .
the observation of long range magnetic ordering in singlet ground state prirsi@xmath0 is enigmatic and needs further investigations .
it should be noted that the heat capacity , resistivity and magnetic susceptibility do not show any sign of ordering at 30 k , where @xmath7sr reveals collective behavior that is of putative magnetic origin .
this may also suggest that the implanted muons in prirsi@xmath0 are playing an important role in the 30 k phase transition .
this type of muon induced magnetic ordering has been seen in prni@xmath228 , prin@xmath18 and pr@xmath19rh@xmath18ge@xmath228 @xcite .
investigations on single crystals of prirsi@xmath0 are highly desirable to understand the observed complex magnetism .
authors vka , dta and adh would like to acknowledge financial assistance from cmpc - stfc grant number cmpc-09108 .
ab would like to thank uj and stfc for pdf funding .
ams thanks the urc of uj , and the sa - nrf ( 78832 ) for financial assistance .
fabi p 1995 focus : an interactive crystal electric field parameter fitting package using neutron scattering data technical report ral - tr-95 - 023 isis facility , rutherford appleton laboratory , chilton , uk
|
we report our comprehensive study of physical properties of a ternary intermetallic compound prirsi@xmath0 investigated by dc magnetic susceptibility @xmath1 , isothermal magnetization @xmath2 , thermo - remnant magnetization @xmath3 , ac magnetic susceptibility @xmath4 , specific heat @xmath5 , electrical resistivity @xmath6 , muon spin relaxation ( @xmath7sr ) and inelastic neutron scattering ( ins ) measurements . a magnetic phase transition is marked by a sharp anomaly at @xmath8 k in @xmath1 measured at low applied fields which is also reflected in the @xmath5 data through a weak anomaly at 12 k. an irreversibility between the zero field cooled and field cooled @xmath1 data below 12.2 k and a very large relaxation time of @xmath3 indicates the presence of ferromagnetic correlation .
the magnetic part of specific heat shows a broad schottky - type anomaly near 40 k due to the crystal electric field ( cef ) effect .
an extremely small value of magnetic entropy below 12 k suggests a cef - split singlet ground state which is confirmed from our analysis of ins data .
the ins spectra show two prominent inelastic excitations at 8.5 mev and 18.5 mev that could be well accounted by a cef model . the cef splitting energy between the ground state singlet and the first excited doublet
is found to be 92 k. our @xmath7sr data reveal a possible magnetic ordering below 30 k which is much higher than that found from the specific heat and magnetic susceptibility measurements .
this could be due to the presence of short range correlations well above the long range magnetic ordering or due to the electronic changes induced by muons .
the induced moment magnetism in the singlet ground state system prirsi@xmath0 with such a large splitting energy of 92 k is quite surprising .
|
although the standard model ( sm ) is a very successful theory , there are still some mysteries and problems . one of them is the flavor structure of fermions .
we currently know that quarks mix with each other through the cabibbo - kobayashi - maskawa ( ckm ) matrix @xcite and that there is a hierarchy among the three mixing angles of the ckm matrix : @xmath4 , @xmath5 and @xmath6 @xcite .
yet another hierarchy exists among quark masses and it can be expressed in terms of @xmath7 as @xmath8 and @xmath9 @xcite . in particular , @xmath10 and @xmath11 surprisingly coincide with each other .
however , the origin of the hierarchies is still unclear .
this is because the yukawa sector in the sm contains a huge number of unknown parameters .
consequently , fermion masses and mixing angles remain free parameters in the sm .
thus , it is interesting to extend the sm with a family symmetry . in ref .
@xcite , authors introduce a family @xmath12 symmetry and try to explain the quark mass hierarchies and small mixing angles via hierarchically suppressed nonrenormalizable yukawa interactions
. a finite group could also be the prime candidate of a family symmetry , which can reduce the number of parameters in the yukawa sector and make the theory predictive @xcite .
recently , friedberg and lee proposed a ( hidden ) translational family symmetry called friedberg - lee ( fl ) symmetry @xcite and tried to relate quark mixing angles with the symmetry .
a more detailed analysis is given in ref .
@xcite and an application to the fourth generation quark model is discussed in ref .
@xcite . furthermore
, many attempts for the lepton sector are performed in refs .
@xcite , some possible origins of the symmetry have been discussed in ref .
@xcite , and a new spontaneous cp violation mechanism with the symmetry is considered in ref .
moreover , one of the motivations to consider the fl symmetry is that the smallness of up- and down - quark masses as well as the electron mass can be naturally understood as it always makes one of three generation fermions massless @xcite .
however , the original approach @xcite with the fl symmetry can not explicitly reveal the hierarchies of quark masses and mixing angles , which are treated as input parameters . in this paper
, we extend the discussion to connect both quark masses and mixing angles with the fl symmetry .
in particular , as will be explained in the next section , we propose the fl symmetry which translates the three generation quarks hierarchically and also show that the hierarchical patterns can be responsible for the quark flavor structures mentioned above .
this paper is organized as follows . in sec .
ii , we present the hierarchical fl symmetry . in sec .
iii , we break both fl and cp symmetries in order to generate light quark masses and the cp violating dirac phase in the ckm matrix . in sec .
iv , we extend our model to the fourth generation case .
we summarize our results in sec .
we start our discussion with the lagrangian of the quark mass terms @xmath13 where @xmath14 are up- and down - type quark mass matrices , which are assumed to be symmetric .
the subscripts @xmath15 stand for the family indices . for the lagrangian , we impose the fl symmetry @xcite @xmath16 where @xmath17 and @xmath18 are c - numbers , @xmath19 is a global grassmann parameter and @xmath20 . because of the symmetry
, the quark mass matrices take the form @xmath21 where we have assumed that @xmath22 and @xmath18 are real .
hence , the theory is cp conserving .
also , the up and down quarks are massless because of the fl symmetry . in the next section
, we will insert phase factors into the mass matrices to generate the light quark masses and cp violation . in order to make our point more clear , here we redefine the parameters as follows : @xmath23 consequently
, the up- and down - type quark mass matrices become @xmath24 and @xmath25 respectively . in this basis , we impose @xmath26 we note that with eq .
( [ eq : para ] ) , the down - type quark mass matrix becomes similar to the hybrid texture discussed in ref . @xcite . on the other hand , the up - type one is almost diagonal .
since @xmath14 are real and symmetric matrices , they can be diagonalized by real orthogonal matrices : @xmath27 . with eq . (
[ eq : para ] ) , @xmath28 can approximately be written as @xmath29 with @xmath30 and @xmath31 for the up - quark sector , and @xmath32 with @xmath33 and @xmath34 for the down - quark sector , respectively .
note that the factor @xmath35 in eq .
( [ eq : vd ] ) arises from the @xmath36 element of eq .
( [ eq : md ] ) . from the mass ratios
, we can deduce that @xmath37 with @xmath38 .
therefore , in what follows , we consider the following hierarchical fl translation : @xmath39 the minus signs in the up sector come from @xmath40 which is negative to reproduce realistic ckm elements .
we note that although the up and down quarks are massless at this stage , in the above approximation , we remain @xmath41 and @xmath42 to be nonzero in order to determine the order of @xmath17 . the parameters @xmath43 and @xmath44 would be the origin of @xmath45 and @xmath46 , respectively , when we introduce the symmetry breaking terms .
on the other hand , the parameters @xmath17 and @xmath18 can also be the origin of the tiny mixing angles of the ckm matrix . for instance , three elements of the ckm matrix can be estimated as @xmath47 where we have used @xmath48 .
these results well coincide with the experimental values @xmath49 and @xmath50 mentioned in the introduction .
namely , in our model , the flavor structures in the quark sector are explained by the hierarchical patterns of the symmetry .
hence , it is also easy to establish the relations between the masses and mixing angles , such as @xmath51 via @xmath52 , and @xmath53 via @xmath54 , respectively .
in order to generate the light quark masses , the fl symmetry must be broken .
since we do not know the origin of the breaking , there may exist many possible ways to break the symmetry . here
, we aim at a minimal scheme and put phase factors into the mass matrices to break the fl and cp symmetries simultaneously as discussed in ref .
that is , we replace eqs .
( [ eq : mu ] ) and ( [ eq : md ] ) as @xmath55 @xmath56 respectively .
note that phase factors are added into @xmath57 and @xmath58 elements for each matrix .
the phases @xmath59 and @xmath60 are responsible for the up- and down - quark masses .
if we regard @xmath61 and @xmath62 as perturbations , we obtain @xmath63 as we expected .
the phases @xmath64 and @xmath65 are added to account for the cp violating dirac phase in the ckm matrix . in the following
, we will present our numerical analysis to illustrate our result . in our calculation , without loss of generality , we will normalize eqs .
( [ eq : mu2 ] ) and ( [ eq : md2 ] ) with @xmath66
. we will also ignore the terms associated with @xmath67 because they are suppressed by @xmath68 . as a result , the model has three real parameters : @xmath69 and @xmath70 , and three cp violating phases : @xmath71 and @xmath72 .
for these six parameters , we use the following experimental values from the particle data group ( pdg ) @xcite : @xmath73 as input parameters .
figure [ fig : vub - vtdts ] shows the allowed region in the @xmath74 plane with the bounds @xcite @xmath75 plane , where the experimental values from the pdg @xcite are also plotted as the cross lines .
, scaledwidth=80.0% ] one can easily see from fig .
[ fig : vub - vtdts ] that our results on @xmath76 and @xmath77 are consistent with the experimental data . here
, we would like to emphasize that the flavor structures of the quark sector are mainly originated from the hierarchical fl symmetry even though the parameters @xmath78 and @xmath70 are not completely fixed . at the best fit point , we have @xmath79 with @xmath66 , which fill the gap between the exact values and expected ones estimated in the previous section .
in our scheme , quark masses and mixing angles are closely related with each other , such as those in eqs .
( [ eq : masmix1 ] ) and ( [ eq : masmix2 ] ) .
so , it should be interesting to extend our scheme to the fourth generation quark model and see whether there exist relations between the fourth generation quark masses and their mixing angles .
we extend the fl symmetry as @xmath80 and the mass matrices eqs .
( [ eq : mu ] ) and ( [ eq : md ] ) as @xmath81 and @xmath82 respectively . here
, we again assume that @xmath83 note that we extend the model so that the mass matrices keep the features mentioned just behind eq .
( [ eq : para ] ) .
since a detailed analysis goes beyond the purpose of the paper , we would like to roughly study and try to figure out their implications . both eqs .
( [ eq : mu4 ] ) and ( [ eq : md4 ] ) are diagonalized by real orthogonal matrices .
one can easily find that @xmath84 determine the mass ratios of the fourth and third generation quarks : @xmath85 where @xmath86 and @xmath87 indicate the fourth generation quarks , and all coefficients are omitted . the fourth generation quark mixings with the other three generations
can also be estimated as @xmath88 where we have used @xmath89 and @xmath90 .
as one can see , the fourth generation quark mixing angles are directly related to the mass ration @xmath91 via @xmath92 .
we note that the contribution of @xmath93 to the mixing angles is negligibly small compared with that of @xmath92 . to illustrate our implications , we substitute the lower bound of @xmath94 , i.e. , @xmath3 @xcite , and the central value @xmath2 @xcite , into eq .
( [ eq : m4 ] )
. then we get @xmath95 and @xmath96 it is interesting to see that @xmath97 and @xmath98 can be large , which could be measurable at the upcoming experiments like the lhc .
finally , we briefly comment about cp violation .
although , in general , the discussion of cp violation in the fourth generation model is very complicated , in the chiral limit of @xmath99 , cp violation is described by only one simple quantity @xcite @xmath100\ .\end{aligned}\ ] ] as discussed in ref .
@xcite , the quantity may provide us useful information about the size of cp violation . by introducing the same phase factors discussed in sec .
iii , we find that @xmath101 .
unfortunately , the result is much smaller than the sm jarlskog invariant parameter @xmath102 .
hence , the fourth generation model , in our scheme , can not be the main source of the cp violating phenomena , such as the baryon asymmetry of the universe @xcite .
we have studied the quark masses and mixing angles with the hierarchical fl symmetry .
we have shown that the symmetry can explain the hierarchies in the quark masses and mixing angles at the same time . as a result
, the masses and mixing angles are closely related with each other in our model . to generate light quark masses and cp violation ,
we have introduced phase factors into the mass matrices , and then demonstrated that our model can reproduce all experimental values .
we have also extended our scheme to the fourth generation quark model and found the relation @xmath0 for @xmath2 and @xmath3 . we have speculated that @xmath97 and @xmath98 could be measurable at the upcoming experiments like the lhc in our extended fourth generation quark model . +
* acknowledgments * please see h. fritzsch and z. z. xing , _ prog .
* 45 * 1 ( 2000 ) for a review on the relation between the quark masses and mixing angles ; see also h. fritzsch , _ phys .
b _ * 70 * , 436 ( 1977 ) and references therein .
r. friedberg and t. d. lee , _ high energy phys .
phys . _ * 30 * , 591 ( 2006 ) ; z. z. xing , h , zhang , and s. zhou , _ phys .
b _ * 641 * , 189 ( 2006 ) ; s. luo and z. z. xing , _ phys . lett .
b _ * 646 * , 242 ( 2007 ) ; _ phys . rev
* 78 * , 117301 ( 2008 ) ; c. jarlskog , _ phys . rev .
d _ * 77 * , 073002 ( 2008 ) ; t. araki and r. takahashi , _ eur .
j. c _ * 63 * , 521 ( 2009 ) ; a. h. chan , h. b. low , and z. z. xing , _ phys .
d _ * 80 * , 073006 ( 2009 ) ; t. araki and c. q. geng , _ phys .
b _ * 680 * , 343 ( 2009 ) .
|
we consider the friedberg - lee symmetry for the quark sector and show that the symmetry closely relates to both quark masses and mixing angles .
we also extend our scheme to the fourth generation quark model and find the relation @xmath0 with @xmath1 for @xmath2 and @xmath3 .
|
a triangle packing in a graph @xmath2 is a collection of vertex - disjoint triangles .
the triangle packing problem asks for a triangle packing of maximal cardinality .
the triangle partition problem asks whether the vertices of a graph can be partitioned into triangles .
we refer to appendix [ prel tp ] for an overview of known results on triangle packing problems .
our objective is the study of the triangle partition problem on permutation graphs .
we establish polynomial - time algorithms for several classes of graphs that are related to cographs and permutation graphs .
we show that there exist polynomial - time algorithm for triangle packings of colored permutations , complete multipartite graphs , @xmath0-modular permutation graphs , distance - hereditary graphs and complements of @xmath0-partite graphs .
we show that the @xmath1-packing problem can be solved on bipartite permutation graphs and that this problem becomes np - complete on the class of bipartite graphs .
we also characterize the cobipartite graphs that admit a triangle partition . since a lot of research on triangle packing centers on interval graphs , we start our discussion with this class of graphs .
a graph is an interval graph if it is the intersection graph of a collection of intervals on the real line @xcite . a consecutive clique arrangement of a graph @xmath2 is a linear arrangement of its maximal cliques such that for each vertex , the cliques that contain it are consecutive .
let @xmath2 be an interval graph with @xmath3 vertices .
then it has at most @xmath3 maximal cliques .
the following theorem was proved in ( * ? ? ?
* theorem 7.1 ) .
a graph @xmath2 is an interval graph if and only if it has a consecutive clique arrangement .
consider the following problem , called the ` partition into bounded cliques ' problem in @xcite .
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ let @xmath4 be a graph and let @xmath5 and @xmath6 be integers . can @xmath7 be partitioned into @xmath6 cliques each of cardinality at most @xmath5 ? _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ this problem can be solved in linear time on interval graphs @xcite . in the following we denote by @xmath8 the maximal number of vertex disjoint triangles in @xmath2 .
the following lemma is easy to check .
[ at least three vertices ] let @xmath2 be an interval graph and let @xmath9 $ ] be a consecutive clique arrangement . 1 .
if there is a maximal clique with only one vertex @xmath10 then @xmath10 is an isolated vertex . in that case @xmath11 .
if there is a maximal clique with exactly two vertices @xmath10 and @xmath12 then @xmath13 is a bridge .
let @xmath14 be the graph obtained from @xmath2 by deleting the edge @xmath13 but not its endvertices .
then @xmath14 is an interval graph and @xmath15 .
henceforth we may assume that every maximal clique in the consecutive clique arrangement has at least three vertices .
[ greedy ] let @xmath2 be an interval graph and let @xmath9 $ ] be a consecutive arrangement of its maximal cliques .
consider an ordering of the vertices in @xmath16 by increasing degree or , equivalently , by increasing right endpoints of the corresponding intervals .
there is a triangle packing of @xmath2 with maximal cardinality such that all vertices of @xmath16 are covered except , possibly the smallest or , the two smallest vertices .
if there are at least three vertices in @xmath16 not covered by a triangle then we can add a triangle to the packing .
let @xmath17 and @xmath18 be two vertices in @xmath16 and assume that @xmath19 in the right endpoint ordering .
assume that @xmath17 is in a triangle of a triangle packing @xmath20 and that @xmath18 is not .
then we can switch @xmath17 and @xmath18 in the triangle and obtain an alternative packing @xmath21 with the same number of triangles such that @xmath18 is covered .
the claim follows by induction by recursively switching any vertex that is not covered with the smallest vertex in @xmath16 that is covered .
[ all three ] let @xmath2 be an interval graph and let @xmath9 $ ] be a consecutive arrangement of its maximal cliques .
let @xmath17 , @xmath18 and @xmath22 be the three smallest vertices of @xmath16 in the ordering by right endpoints . if @xmath17 is covered by a triangle in a triangle packing @xmath20 then there exists a triangle packing @xmath21 of the same cardinality as @xmath20 such that @xmath23 is a triangle of @xmath21 . since @xmath16 is a maximal clique it has a vertex that is not contained in @xmath24
consequently , @xmath17 is contained only in @xmath16 .
consider a triangle @xmath25 in @xmath20 .
then @xmath26 and @xmath27 are in @xmath16 .
let @xmath26 be the smallest of the two .
assume that @xmath28 .
if @xmath18 is not covered by any triangle of @xmath20 then we can replace @xmath26 with @xmath18 .
assume that @xmath18 is in a triangle @xmath29 of @xmath20 .
let @xmath30 be the first clique that contains all three vertices @xmath18 , @xmath5 and @xmath6 .
then @xmath30 also contains @xmath26 , since @xmath26 is larger than @xmath18 .
replace the two triangles @xmath25 and @xmath29 with @xmath31 and @xmath32 .
a similar argument shows that , if @xmath33 then we can replace @xmath27 with @xmath22 and obtain an alternative packing @xmath21 with the same number of triangles .
the triangle partition problem can be solved in linear time on interval graphs .
the triangle packing problem can be solved by an exponential algorithm which runs in @xmath34 time .
the first claim follows from lemma [ all three ] .
the second claim follows from the recurrence @xmath35 to see that this recurrence holds , observe that the minimal element of @xmath16 is either not in any triangle or it is in a triangle together with the next two smallest elements of @xmath16 .
an @xmath36 algorithm for maximum matching in interval graphs is presented in @xcite . for the class of strongly chordal graphs , which includes the class of interval graphs
, there exists a linear - time algorithm for maximum matching when a strong elimination ordering of the graph is part of the input @xcite .
et al . _ , extend the greedy algorithm for a @xmath37-partition for general @xmath5 on interval graphs to the class of strongly chordal graphs @xcite .
concerning packings of vertex - disjoint maximal cliques we have the following theorem .
let @xmath2 be an interval graph .
there exists a linear - time algorithm that finds the maximal number of vertex - disjoint maximal cliques .
this can be seen as follows .
first recall that the maximal number of vertex - disjoint maximal cliques in an interval graph is equal to the minimal number of vertices that represent all maximal cliques @xcite .
a vertex @xmath10 represents a clique @xmath38 if @xmath39 . a set of vertices that together represent all maximal cliques
is called a clique - transversal .
let @xmath9 $ ] be a consecutive clique ordering of @xmath2 .
we use the following trick that we learned from @xcite to reduce the problem to a domination problem . add one vertex to each maximal clique @xmath30 .
then the new graph @xmath40 has @xmath41 vertices and @xmath40 is an interval graph .
it is easy to check that the minimal cardinality of a clique transversal in @xmath2 is equal to the minimal cardinality of a dominating set in @xmath40 . here ,
a dominating set in a graph is a set @xmath42 of vertices such that every vertex not in @xmath42 has a neighbor in @xmath42 .
there exists a linear - time algorithm that finds a dominating set of minimal cardinality in an interval graph @xcite .
this proves the claim .
we refer to appendix [ prel perm ] for a brief overview on permutation graphs .
notice that the triangle partition problem for permutation graphs is equivalent to the following problem .
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ let @xmath43 be a permutation of @xmath44 .
decide if there exist a partition @xmath20 of @xmath45 into triples such that for each element @xmath46 with @xmath47 , @xmath48 _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ as far as we know , both the triangle partition - and the triangle packing problem for permutation graphs are open . in this section
we consider a variation of the partition problem .
let @xmath49 and let @xmath50 there exists a polynomial - time algorithm that decides whether there exists a partition @xmath20 of @xmath45 into triples such that for each @xmath46 with @xmath51 : @xmath52 construct two bipartite graphs as follows .
the first bipartite graph has vertices that are the elements with colors 1 and 2 .
a vertex @xmath53 with color 1 is adjacent to a vertex @xmath54 with color 2 if @xmath55 the second bipartite graph has vertices with colors 2 and colors 3 .
a vertex @xmath26 with color 2 is adjacent to a vertex @xmath27 with color 3 if @xmath56 it is easy to check that there exists a partition into triangles if and only if both bipartite graphs have a perfect matching .
one can find a perfect matching in @xmath57 time @xcite , where @xmath58 is the exponent of a matrix multiplication algorithm .
it is easy to check that this result generalizes to the @xmath37-partitioning problem when an @xmath5-coloring is a part of the input .
complete @xmath59-partite graphs form a subclass of the permutation graphs . in this section
we show that the triangle partition problem can be solved in polynomial time for complete @xmath59-partite graphs .
let @xmath60 be positive natural numbers .
we use @xmath61 to denote the complete @xmath59-partite graph with color classes @xmath62 such that @xmath63 for all @xmath64 .
[ large classes ] let @xmath2 be a @xmath59-partite graph . if there exists a triangle partition @xmath20 of @xmath2 then there exists a triangle partition @xmath21 which contains a triangle with vertices in three of the largest color classes .
let @xmath62 be the color classes of @xmath4 and let @xmath65 such that @xmath66 we prove the claim by induction on the number of vertices .
if there are only three vertices then the claim is obviously true .
consider a triangle @xmath67 and consider the subgraph @xmath14 of @xmath2 induced by @xmath68 .
then @xmath14 has a triangle partition @xmath69 . by induction
there exists a triangle partition @xmath70 of @xmath14 with a triangle @xmath71 contained in three of the maximal color classes of @xmath14 .
note that color classes with the same cardinality are interchangeable .
therefore , we may assume that the three maximal classes of @xmath14 that contain the vertices of @xmath71 are also maximal classes of @xmath2 .
there exists a linear - time algorithm that solves the triangle partition problem on complete @xmath59-partite graphs . by lemma [ large classes ] a greedy algorithm which chooses recursively triangles from three of the largest color classes produces a triangle partition if it exists .
it is easy to see that this algorithm can be implemented to run in linear time .
we have not found an easy condition on the numbers @xmath60 which characterizes the complete @xmath59-partite graphs that have a triangle partition .
a similar greedy algorithm solves the triangle packing problem on complete @xmath59-partite graphs .
let @xmath72 be a bipartite permutation graph and consider the left - to - right orderings of the vertices of @xmath73 and @xmath74 on the topline of a permutation diagram .
it is easy to check that this is a strong ordering , which is defined as follows .
let @xmath72 be a bipartite graph .
strong ordering _ is a pair of linear orderings @xmath75 and @xmath76 on @xmath73 and @xmath74 such that for all @xmath77 and @xmath78 with @xmath79 and @xmath80 @xmath81 spinrad , _ et al .
_ , obtained the following characterization @xcite .
let @xmath72 be a bipartite graph .
then @xmath2 is a bipartite permutation graph if and only if there is a strong ordering on @xmath73 and @xmath74 . in this section
we show that there is a greedy algorithm that computes a @xmath1-packing on bipartite permutation graphs .
consider a diagram for a bipartite permutation graph @xmath72 and let @xmath82 be the left - to - right ordering of the points on the topline .
denote by @xmath75 and @xmath76 the sub - orderings of @xmath82 induced by the vertices of @xmath73 and @xmath74 .
[ right ] let @xmath20 be a @xmath1-packing on @xmath2
. assume that @xmath83 with @xmath84 and assume that @xmath85 and @xmath86 are in a square @xmath87 .
let @xmath88 and assume that @xmath89 is the smallest element @xmath90 .
then there is a packing @xmath21 of the same cardinality as @xmath20 such that @xmath85 and @xmath89 are in a square @xmath91 .
let @xmath92 .
first notice that for each vertex @xmath93 , its neighborhood @xmath94 forms an interval in @xmath95 @xcite .
then @xmath96 thus @xmath97 is a square .
if @xmath89 is not in any square of @xmath20 then we can replace @xmath38 with @xmath98 .
assume that @xmath89 is in a square @xmath99 .
we consider the following cases .
first assume that @xmath100 .
then @xmath26 and @xmath27 are adjacent to @xmath85 , @xmath89 , @xmath86 and @xmath101 .
since @xmath5 and @xmath6 intersect the linesegments of @xmath89 and @xmath101 but not the linesegments of @xmath26 and @xmath27 each of @xmath5 and @xmath6 intersects at least one of @xmath85 and @xmath86 .
if @xmath5 and @xmath6 both intersect @xmath85 then we can replace @xmath38 and @xmath24 with @xmath102 and @xmath103 .
similarly , of both @xmath5 and @xmath6 intersect @xmath86 we can replace @xmath38 and @xmath24 with @xmath97 and @xmath104 .
assume that @xmath5 intersects @xmath85 and that @xmath6 intersects @xmath86 .
then replace @xmath38 and @xmath24 with @xmath105 and @xmath106 .
now assume that @xmath107 .
then @xmath108 thus @xmath104 is a square .
replace @xmath38 and @xmath24 with @xmath97 and @xmath104 .
this proves the lemma . by lemma
[ right ] , there exists a maximum @xmath1-packing @xmath20 of @xmath2 such that the pairs of vertices of @xmath73 that are contained in a square of @xmath20 are consecutive in @xmath75 and the pairs of vertices of @xmath74 that are contained in a square of @xmath20 are consecutive in @xmath76 .
[ left ] consider two squares @xmath109 and @xmath110 in a @xmath1-packing @xmath111 .
assume that @xmath112 and that @xmath113 .
there exists a packing @xmath21 of the same cardinality as @xmath20 such that @xmath97 and @xmath104 are squares in @xmath21 .
assume that @xmath114 . since the linesegments of @xmath26 and @xmath27 intersect the linesegments of @xmath86 and @xmath101 , and @xmath5 and @xmath6 intersect @xmath85 and @xmath89 , and since the linesegments of @xmath26 , @xmath27 , @xmath5 and @xmath6 are parallel , by the ordering the linesegments of @xmath26 , @xmath27 , @xmath5 and @xmath6 intersect all linesegments of @xmath85 , @xmath89 , @xmath86 and @xmath101 .
thus @xmath104 and @xmath97 are squares .
the only other possible case is where @xmath115 .
this case is similar .
this proves the lemma .
there exists a linear - time algorithm which computes a maximum @xmath1-packing in a bipartite permutation graph .
let @xmath72 be a bipartite permutation graph and let @xmath82 the the ordering of the vertices on the topline of a diagram for @xmath2 .
we prove that there exists a maximum packing @xmath20 such that the first four vertices in the ordering @xmath82 that form a square are in @xmath20 .
consider the first four vertices @xmath116 that form a square .
we may assume that @xmath117 .
assume that @xmath118 .
consider the square @xmath119 in @xmath20 with the smallest vertices in @xmath73 . if @xmath120 then @xmath38 is disjoint from all squares in @xmath20 which contradicts the maximality of @xmath20 . in all other cases we can replace @xmath16 with @xmath38 .
this proves the correctness of the following algorithm .
remove vertices that are smallest in the @xmath82-ordering that are not in any square of @xmath2 .
assume next that the smallest element is @xmath121 .
thus @xmath122 is in a square .
take the first element @xmath123 that is in a square with @xmath122 .
then take the first two elements @xmath78 such that @xmath116 induces a square in @xmath2
. put @xmath38 in @xmath20 . remove the vertices @xmath38 from @xmath2 and recurse .
it is easy to see that , with some care this algorithm can be implemented to run in linear time .
in the previous section we proved that the @xmath1-packing problem on bipartite permutation graphs can be solved in linear time . in this section
we show that the @xmath1-packing problem is np - complete for general bipartite graphs .
[ np - c ] @xmath1-packing on bipartite graphs is np - complete .
it is easy to see that the problem is in np . to show the np - hardness for our problem we use a reduction from the 3-dimensional matching problem ( 3 dm ) which is described as follows .
suppose we are given three sets @xmath124 and @xmath125 such that @xmath126 , and a set @xmath127 of triples @xmath128 .
the 3 dm problem asks for a subset @xmath129 of @xmath130 such that each element of @xmath124 and @xmath125 is contained in exactly one triple in @xmath129 .
we apply local replacements to the input instance of 3 dm . for each pair @xmath10 and
@xmath12 that appear in some triple @xmath131 , we create a vertex @xmath132 and a path of length two @xmath133 $ ] as shown in figure [ fig : replace ] . in @xmath130 .
] for each triple @xmath134 , we create four local vertices @xmath135 \quad
1\leq i \leq 4\ ] ] and eight edges as shown in figure [ fig : replace ] .
this finishes the description of the construction of the bipartite graph @xmath2 .
the bipartite graph @xmath2 is obtained in linear time .
suppose that @xmath129 is a solution of the 3 dm problem .
we obtain a @xmath1-packing of size @xmath136 as follows .
the packing is constructed by taking 4-cycles @xmath137\ } \;\;\text{and}\;\ ; \{a_{\tau}[2],a_{\tau}[3 ] , a_{\tau}[4],z\ } & \quad\text{if}\quad \tau=(x , y , z ) \in
m^{\prime}\\ \{a_{\tau}[1 ] , a_{\tau}[2 ] , a_{\tau}[3 ] , a_{\tau}[4]\ } & \quad\text{if}\quad \tau \not\in m^{\prime}. \end{cases}\ ] ] this ensures that each element in @xmath138 is included in exactly one 4-cycle in the packing .
now assume that there is a @xmath1-packing of size @xmath26 , where @xmath136 . in the constructed figure for any triple @xmath134
, there are only two possible ways to pack the 4-cycles : one way contains the 4-cycle induced by @xmath139 , a_{\tau}[2 ] , a_{\tau}[3 ] , a_{\tau}[4 ] \}\ ] ] and the other way contains the two 4-cycles induced by @xmath140 \ } \quad\text{and}\quad \{a_{\tau}[2 ] , a_{\tau}[3 ] , a_{\tau}[4 ] , z \}.\ ] ] the first choice contains none of vertices from @xmath138 and the second contains exactly one vertex from each of @xmath124 and @xmath125 .
our packing is of size @xmath141 so there must be at least @xmath27 triples that use the second kind of @xmath1 s in the packing . since none of these @xmath27 triples have a common element , they cover all @xmath142 elements in @xmath138 .
thus these @xmath27 triples forms a 3-dimensional matching .
a cobipartite graph is the complement of a bipartite graph .
we denote a bipartite graph @xmath2 with color classes @xmath143 and @xmath144 by @xmath145 .
we use the same notation for a cobipartite graph where @xmath143 and @xmath144 are the color classes of the complement .
a star is a bipartite graph @xmath145 with @xmath146 .
the single vertex in @xmath143 is called the center of the star . in this section
we show that there is a good characterization of the cobipartite graphs that have a triangle partition .
[ cobipartite ] let @xmath145 be cobipartite
. then @xmath2 can be partitioned into triangles if and only if one of the following holds true . 1 .
@xmath147 , or 2 .
@xmath148 and @xmath149 and @xmath2 has a triangle with one vertex in @xmath143 and two in @xmath144 , or 3 .
similar as above with the role of @xmath143 and @xmath144 interchanged .
if @xmath150 then @xmath2 can be partitioned into triangles .
assume that @xmath148 and that @xmath149 .
if there exists a triangle with one vertex in @xmath143 and two in @xmath144 then @xmath2 can be partitioned into triangles . for the converse , assume that there is no such triangle .
let @xmath14 be the bipartite graph obtained from @xmath2 by deleting edges between vertices that are contained in the same color class .
then @xmath14 does not contain a @xmath151 , _
i.e. _ , a path with three vertices , with its midpoint in @xmath143 .
then @xmath14 has no @xmath152 and no @xmath1 and so @xmath14 is trivially perfect @xcite .
it follows that @xmath14 is a disjoint collection of isolated vertices in @xmath143 and stars with their centers in @xmath144 .
then each triangle of @xmath2 that is not contained in @xmath143 nor in @xmath144 has one vertex in @xmath144 and two in @xmath143 . if @xmath2 has a triangle partition then @xmath153 , which is a contradiction .
there exists a linear - time algorithm which check if a cobipartite graph has a triangle partition .
assume that @xmath154 and that @xmath155 .
it is easy to check in linear time whether @xmath14 is disjoint collection of isolated vertices and stars with their midpoints in @xmath144 . by theorem [ cobipartite ]
the graph @xmath2 has a partition into triangles if and only if @xmath14 is not a disjoint collection of isolated vertices in @xmath143 and stars with midpoints in @xmath144 .
a @xmath0-partite graph is a graph @xmath4 of which the vertices can be partitioned into @xmath0 independent sets .
notice that the recognition of @xmath156-partite graphs is np - complete since it is an instance of the @xmath156-coloring problem .
henceforth , we assume that the partition into color classes of a @xmath0-partite graph is a part of the input . in this section
we show that the triangle partition problem on the complements of @xmath0-partite graphs can be solved in polynomial time .
we start with the case where @xmath157 .
[ 3-partite ] let @xmath2 be the complement of a @xmath156-partite graph with color classes @xmath158 , @xmath159 and @xmath160 .
there exists a collection of colored graphs @xmath161 , each with at most @xmath162 vertices , such that @xmath2 has a partition into triangles if and only if one of the graphs @xmath163 is an induced subgraph of @xmath2 .
consider a partition of @xmath2 into triangles .
each triangle is either contained in one of the color classes or , it has one vertex in each color class or , it has one vertex in one color class and two vertices in another color class .
assume that 3 vertices in @xmath158 , @xmath159 and @xmath160 are mutually connected by triangles .
then we can change the triangle partition such that it contains the triangles on the three vertices in each @xmath164 instead .
assume that there are 3 vertices in @xmath158 that are in triangles with 3 edges in @xmath159 .
then we may replace those triangles by one triangle in @xmath158 and two in @xmath159 . in this way we obtain a partition into triangles such that there are at most 14 vertices in each color class @xmath164 that are in triangles that are not completely contained in @xmath164 .
consider all possible colored graphs @xmath163 with at most @xmath162 vertices .
then there is a partition of @xmath2 into triangles if and only if there is a colored induced subgraph @xmath163 , which can be partitioned into triangles , such that the number of remaining vertices in each class @xmath164 is @xmath165 .
there exists a polynomial - time algorithm that checks whether the vertices of the complement of a @xmath156-partite graph can be partitioned into triangles .
it is folklore that the triangle partition problem is np - complete on @xmath156-partite graphs , see , _
e.g. _ , @xcite .
[ k - partite ] let @xmath0 be a natural number .
there exists a polynomial - time algorithm that solves the triangle partition problem on complements of @xmath0-partite graphs .
let @xmath4 be the complement of a @xmath0-partite graph with color classes @xmath166 .
assume that @xmath7 can be partitioned into triangles .
first we show that there exists a triangle partition of @xmath2 with only a constant number of triangles of which the vertices are not monochromatic .
consider three color classes @xmath158 , @xmath159 and @xmath160 . by lemma [ 3-partite ]
we may assume that at most @xmath167 triangles that are not monochromatic are contained in @xmath168 .
since this holds for any three color classes , we find that there is a partition with at most @xmath169 non - monochromatic triangles .
a graph @xmath2 is distance hereditary if for every component in every induced subgraph the distance between two vertices is the same as their distance in @xmath2 @xcite . in this section
we show that the triangle packing problem can be solved in polynomial time on distance - hereditary graphs .
a decomposition tree for a graph @xmath4 is a pair @xmath170 where @xmath171 is a ternary tree and where @xmath172 is a 1 - 1 map from the vertices in @xmath2 to the leaves of @xmath171 .
a line in @xmath171 induces a partition of @xmath7 into two sets , say @xmath143 and @xmath144 .
the twinset of @xmath143 is the subset of vertices in @xmath143 that have neighbors in @xmath144 .
the graph @xmath2 is distance hereditary if and only if it has a decomposition tree @xmath170 such that for every partition @xmath173 induced by a line in @xmath171 every pair of vertices in the twinset of @xmath143 have the same neighbors in @xmath144 @xcite .
if @xmath2 is distance hereditary , such a decomposition tree for @xmath2 can be found in linear time @xcite .
[ dh ] there exists a polynomial - time algorithm that solves the triangle packing problem on distance - hereditary graphs .
our method resembles the one used in @xcite used to solve the triangle packing problem on cographs .
let @xmath4 be distance hereditary and let @xmath170 be a decomposition tree for @xmath2 which satisfies the properties mentioned above .
the algorithm performs dynamic programming on branches of @xmath171 of increasing size . consider a branch @xmath144 rooted at some line @xmath174 of @xmath171 .
suppose that @xmath144 decomposes into two smaller branches @xmath175 and @xmath176 .
let @xmath177 and @xmath178 be the twinsets of the vertices mapped to leaves of @xmath175 and @xmath176 .
then every vertex of @xmath177 is adjacent to every vertex of @xmath178 or no vertex of @xmath177 is adjacent to a vertex of @xmath178 .
moreover , the twinset for @xmath144 is either @xmath179 or , it is one of the two or it is empty . the dynamic programming keeps track of the maximum cardinality of a triangle packing in a branch @xmath144 that avoids a specified number of ` free ' vertices and a specified number of ` free ' edges in the twinset . as an invariant , the free edges do not contain any of the free vertices and they form a matching .
free vertices and edges can be used to form triangles with vertices outside the branch .
it is easy to update the table for a branch @xmath144 from the tables of @xmath175 and @xmath176 .
details for the updating procedure can be found in @xcite . since the number of entries of each table is bounded by @xmath180
, it follows that the algorithm can be implemented to run in polynomial time @xcite .
it is fairly easy to see that the algorithm above can be extended so that it works for graphs of bounded rankwidth @xcite .
we have not been able to formulate the triangle partition problem in monadic second - order logic .
guruswami , _ et al .
_ , describe an algorithm for the @xmath37-packing problem on cographs that runs in polynomial time for each fixed @xmath5 . the problem whether this problem can be solved by a fixed - parameter algorithm with respect to @xmath5 remains an open problem .
let @xmath43 be a permutation of @xmath45 .
a module in @xmath43 is a consecutive subsequence of @xmath43 that is a permutation of a consecutive subsequence of @xmath181 $ ] .
let @xmath2 be a permutation graph .
consider a diagram for @xmath2 with the labels @xmath181 $ ] in order on the topline and with the labels @xmath182 $ ] in order on the bottom line .
then a module in the permutation corresponds with a subset @xmath130 of vertices such that every vertex outside @xmath130 is adjacent to all vertices of @xmath130 or to no vertex of @xmath130 .
a subset @xmath130 of vertices in a graph @xmath2 with this property is called a module of @xmath2 .
a module @xmath130 is trivial if it contains zero , one or all the vertices of the graph .
a module is strong if it does not overlap with other modules .
a graph is prime if it contains only trivial modules .
if @xmath183 is a strong module then there exists a unique strong module @xmath129 of minimal size that properly contains @xmath130 .
this defines a parent relation in a modular decomposition tree for @xmath2 .
a modular decomposition tree is a rooted tree @xmath171 with a 1 - 1 map from the leaves to a set @xmath7 of vertices .
each internal node of @xmath171 is labeled as a join node , as a union node , or as a prime node .
a modular decomposition tree @xmath171 defines a graph @xmath2 with vertex set @xmath7 as follows .
a join node @xmath54 stands for the operation which adds an edge between every pair of vertices that are mapped to leaves in different subtrees of of @xmath54 .
a union node @xmath184 stands for the operation that unions the subgraphs represented by the children of @xmath184 .
each prime node @xmath26 is labeled with a graph @xmath185 .
each vertex in @xmath185 corresponds with one child of @xmath26 .
if two vertices in @xmath185 are connected by an edge then every pair of vertices in the two graphs represented by the two corresponding children is connected by an edge .
given a graph @xmath2 , a tree that decomposes @xmath2 recursively into strong modules can be constructed in linear time @xcite .
a graph is _
@xmath0-modular _ if it has a modular decomposition tree such that the graph @xmath185 of every prime node @xmath26 has at most @xmath0 vertices .
if the modular decomposition tree has no prime nodes then the graph is decomposable by unions and joins .
we call these graphs @xmath186-modular .
the class of @xmath186-modular graphs is exactly the class of cographs .
the graphs that are @xmath0-modular are characterized by a finite collection of forbidden induced subgraphs .
this follows from kruskal s theorem @xcite .
consider the class of graphs obtained from paths by replacing the endvertices by false twins , _
i.e. _ , modules consisting of two nonadjacent vertices .
this class is contained in the class of permutation graphs and it is not well - quasi - ordered by the induced subgraph relation .
this proves the following corollary .
another way to see that is by showing that paths are prime .
there exists a function @xmath187 such that @xmath0-modular permutation graphs have no induced paths of length more than @xmath187 .
we omit the easy proof of the following theorem . for each natural number @xmath0
there exists a polynomial - time algorithm which computes a triangle packing in @xmath0-modular graphs .
the main question that we leave open in this paper is whether there exists a polynomial - time algorithm that checks if a permutation of @xmath188 can be partitioned into increasing subsequences of length three .
we thank klaas zwartenkot for doing some calculations on the number of permutations that can be partitioned into triangles .
an obvious lowerbound for the number of permutations of @xmath188 that can be partitioned into triangles is @xmath189 .
we have not been able to determine an exact formula for the number of permutations that can be partitioned into triangles nor have we been able to determine the asymptotics for those permutations .
99 alon , n. , j. pach and j. solymosi , ramsey - type theorems with forbidden subgraphs , _ combinatorica _ * 21 * ( 2001 ) , pp .
155170 .
baker , b. and e. coffman , mutual exclusion scheduling , _ theoretical computer science _ * 162 * ( 1996 ) , pp .
225243 .
baker , k. , p. fishburn , f. roberts , partial orders of dimension 2 , _ networks _ * 2 * ( 1971 ) , pp .
bodlaender , h. , on disjoint cycles , _ international journal on foundations of computer science _
* 5 * ( 1994 ) , pp .
bodlaender , h. and k. jansen , restrictions of graph partition problems .
part i , _ theoretical computer science _
* 148 * ( 1995 ) , pp .
93109 .
chudnovsky , m. and s. safra , the erds - hajnal conjecture for bull - free graphs , _ journal of combinatorial theory , series b _ * 98 * ( 2008 ) , pp .
13011310 .
chudnovsky , m. , n. robertson , p. seymour and r. thomas , the strong perfect graph theorem , _ annals of mathematics _ * 164 * ( 2006 ) , pp .
51229 .
cornelsen , s. and g. di stefano , treelike comparability graphs , _ discrete applied mathematics _ * 157 * ( 2009 ) , pp .
17111722 .
cornujols , g. , d. hartvigsen and w. pulleyblank , packing subgraphs in a graph , _ operations research letters _ * 1 * ( 1982 ) , pp . 139143 .
dahlhaus , e. , efficient parallel and linear time sequential split decomposition ( extended abstract ) , _ proceedings of the @xmath190 conference on foundations of software technology and theoretical computer science _
, springer lncs 880 ( 1994 ) , pp .
171180 .
dahlhaus , e. and m. karpinsky , matching and multidimensional matching in chordal and strongly chordal graphs , _
discrete applied mathematics _ * 84 * ( 1998 ) , pp .
dilworth , r. , a decomposition theorem for partially ordered sets , _ annals of mathematics _ * 51 * ( 1950 ) , pp . 161166
. dushnik , b. and e. miller , partially ordered sets , _ american journal of mathematics _ * 63 * ( 1941 ) , pp .
600610 .
erds , p. and a. hajnal , ramsey - type theorems ,
_ discrete applied mathematics _ * 25 * ( 1989 ) , pp .
erds , p. and l. psa , on independent circuits contained in a graph , _
canadian journal of mathematics _ * 17 * ( 1965 ) , pp . 347352
. erds , p. and g. szekeres , a combinatorial problem in geometry , _ compositio mathematica _ * 2 * ( 1935 ) , pp .
463470 .
felsner , s. and l. wernisch , maximum @xmath0-chains in planar point sets , _ siam journal on computing _ * 28 * ( 1993 ) , pp .
192209 . fox , j. and b. sudakov , induced ramsey - type theorems , _ advances in mathematics _ * 219 * ( 2008 ) , pp . 17711800 .
frank , a. , on chain and antichain families of a partially ordered set , _ journal of combinatorial theory , series b _ * 29 * ( 1980 ) , pp . 176184
. fredman , m. , on computing the length of longest increasing subsequences , _ discrete mathematics _
* 11 * ( 1975 ) , pp .
fulkerson , d. and o. gross , incidence matrices and interval graphs , _ pacific journal of mathematics _ * 15 * ( 1965 ) , pp .
835855 .
golumbic , m. , _ algorithmic graph theory and perfect graphs _ , elsevier series _ annals of discrete mathematics _ * 57 * , 2004 .
graham , r. , m. grtschel and l. lovsz ( eds . ) , _ handbook of combinatorics , volume 1 _ , north - holland , elsevier science b. v. , 1995 . greene ,
c. , some partitions associated with a partially ordered set , _ journal of combinatorial theory , series a _ * 20 * ( 1976 ) , pp .
greene , c. and d. kleitman , the structure of sperner @xmath0-families , _ journal of combinatorial theory , series a _ , * 20 * ( 1976 ) , pp .
grtschel , m. , l. lovsz and a. schrijver , polynomial algorithms for perfect graphs , _ annals of discrete mathematics _
* 21 * ( 1984 ) , pp .
325356 .
guruswami , v. and c. rangan , algorithmic aspects of clique - transversal and clique - independent sets , _ discrete applied mathematics _
* 100 * ( 2000 ) , pp .
183202 .
guruswami , v. , c. rangan , m. chang , g. chang and c. wong , the @xmath37-packing problem , _ computing _ * 66 * , ( 2001 ) , pp .
hajs , g. , ber eine art von graphen , _ internationale mathematische nachrichten _ * 11 * ( 1957 ) , problem 65 .
hansen , p. , a. hertz and j. kuplinsky , bounded vertex colorings of graphs , _ discrete mathematics _ * 111 * ( 1993 ) , pp .
305312 . hell , p. , graph packings , _ electronic notes in discrete mathematics _
* 5 * ( 2000 ) , pp .
170173 .
hell , p. and d. kirkpatrick , packing by cliques and by finite families of graphs , _ discrete mathematics _ * 49 * ( 1984 ) , pp . 4559 .
howorka , e. , a characterization of distance - hereditary graphs , _ the quarterly journal of mathematics _ * 28 * ( 1977 ) , pp .
417420 .
jansen , k. , the mutual exclusion scheduling problem for permutation and comparability graphs , _ information and computation _ * 180 * ( 2003 ) , pp . 7181 .
kaller , d. , a. gupta and t. shermer , linear - time algorithms for partial @xmath0-tree complements , _ algorithmica _ * 27 * ( 2000 ) , pp .
254274 .
kellog , s. and j. johnson , dominating sets in chordal graphs , _ siam journal on computing _ * 11 * ( 1982 ) , pp .
191199 .
kzdy , a. , h. snevily and c. wang , partitioning permutations into increasing and decreasing subsequences , _ journal of combinatorial theory , series a _ * 73 * ( 1996 ) , pp .
353359 .
kruskal , j. , well - quasi - ordering , the tree theorem , and vazsonyi s conjecture , _ transactions of the american mathematical society _ * 95 * ( 1960 ) , pp .
210225 .
liang , y. and c. rhee , finding a maximum matching in a circular - arc graph , _ information processing letters _ * 45 * ( 1993 ) , pp .
185190 .
loebl , m. and s. poljak , efficient subgraph packing , _ journal of combinatorial theory , series b _ * 59 * ( 1993 ) , pp . 106121
. lonc , z. , on complexity of chain and antichain partition problem , _
proceedings wg92 _ , springer lncs 509 ( 1992 ) , pp .
97104 .
lonc , z. , chain partitions of ordered sets , _ order _ * 11 * ( 1994 ) , pp .
343351 .
lovsz , l. , on graphs not containing independent circuits ( hungarian ) , _ mat . lapok _ * 16 * ( 1965 ) , pp .
289299 .
mader , w. , wohlquasigeordnete klassen endlicher graphen , _ journal of combinatorial theory , series b _ * 12 * ( 1972 ) , pp . 105122 .
mani , g. and y. wakabayashi , packing triangles in low degree graphs and indifference graphs , _ discrete mathematics _ * 308 * ( 2008 ) , pp . 14551471 .
mcconnell , r. and j. spinrad , modular decomposition and transitive orientation , _ discrete mathematics _
* 201 * ( 1999 ) , pp .
189241 .
mckee , t. a. and f. r. mcmorris , _ topics in intersection graph theory _
, siam monographs on discrete mathematics and applications , 1999 .
moitra , a. and r. johnson , a parallel algorithm for maximum matching in interval graphs , _ proceedings of the 1989 international conference on parallel processing _ vol .
iii : algorithms and applications ( 1989 ) , pp .
114120 .
moonen , l. and f. spieksma , partitioning a permutation graph : algorithms and an application .
research report or 0358 , katholieke universiteit leuven , belgium , 2003 .
moonen , l. and f. spieksma , partitioning a weighted partial order , _ journal of combinatorial optimization _ * 15 * ( 2008 ) , pp .
342356 .
morandini , m. , np - complete problem : partition into triangles .
report 112360 , universit di udini , 2004 .
mucha , m. and p. sankowski , maximum matchings via gaussian elimination , _ proceedings of the @xmath191 annual ieee symposium on foundations of computer science _ ( 2004 ) , pp . 248255 .
oum , s. , _ graphs of bounded rank - width _ , phd thesis , princeton university , 2005 .
papadimitriou , c. and m. yannakakis , scheduling interval - ordered tasks , _ siam journal on computing _ * 8 * ( 1979 ) , pp .
405409 .
pnueli , a. , a. lempel and s. even , transitive orientation of graphs and identification of permutation graphs , _
canadian journal of mathematics _ * 23 * ( 1971 ) , pp . 160175
. rhee , c. and y. liang , finding a maximum matching in a permutation graph , _ acta informatica _ * 32 * ( 1995 ) , pp .
779792 .
robertson , n. and p. seymour , graph minors . v. excluding a planar graph , _ journal of combinatorial theory , series b _ * 41 * ( 1986 ) , pp .
92114 .
robertson , n. and p. seymour , graph minors .
the disjoint path problem , _ journal of combinatorial theory , series b _ * 63 * ( 1995 ) , pp .
65110 .
spinrad , j. p. , _ efficient graph representations _ , ams fields institute monographs * 19 * , 2003 .
spinrad , j. , a. brandstdt and l. stewart , bipartite permutation graphs , _ discrete applied mathematics _
* 18 * ( 1987 ) , pp . 279292 .
tedder , m. , d. corneil , m. habib and c. paul , simpler linear - time modular decomposition via recursive factorizing permutations , _
proceedings icalp08 _ , springer lncs 5125 ( 2008 ) , pp .
634645 .
wagner , k. , monotonic coverings of finite sets , _ elektronische informationsverarbeitung und kybernetik _ * 20 * ( 1984 ) , pp .
633639 .
wolk , e. , a note on `` the comparability graph of a tree , '' _ proceedings of the american mathematical society _ * 16 * ( 1965 ) , pp .
despite great interest in the cycle packing and cycle cover problem there is relatively little theoretical progress on the first of the two problems .
the two kinds of problems are related by the erds and psa theorem which states that there is a function @xmath192 such that any graph contains either @xmath0 vertex - disjoint cycles or a set with @xmath187 vertices which intersect every cycle @xcite . a graph @xmath40 is topologically contained in a graph @xmath2 if @xmath2 has a subgraph @xmath193 which is a subdivision of @xmath40 .
it is well - known that the graphs that have no @xmath0 disjoint cycles are well - quasi ordered by topological containment @xcite .
see also ( * ? ? ?
* theorem 5.6 ) .
it follows that there exists a finite set @xmath194 of graphs , each containing a maximal number of @xmath0 disjoint cycles , such that a graph @xmath2 has @xmath0 disjoint cycles if and only if some element of @xmath194 is topologically contained in it .
this can also be seen as follows .
notice that a graph @xmath2 does not have @xmath0 vertex - disjoint cycles if and only if @xmath2 does not contain the graph @xmath40 that consists of @xmath0 disjoint triangles as a minor .
this implies the previous observation ( * ? ? ?
* proposition 5.16 ) .
it follows also that the class of graphs without @xmath0 disjoint cycles is minor closed .
furthermore , this class does not contain all planar graphs , so the class has a uniform bound on the treewidth @xcite .
bodlaender subsequently showed that the elements of this class can be recognized in @xmath195 time @xcite .
likewise , for any natural number @xmath0 one can check in @xmath195 time whether a graph has a cycle cover with at most @xmath0 vertices .
another research area that is related to the topic of this paper is that of finding vertex colorings of graphs that are restricted in some way .
there are too many variations to cover in any limited survey , even when restricted to permutation graphs .
we mention some of the results that seem closely related to our research . on partially ordered sets ,
one of the major contributions is the result of greene and kleitman and of frank @xcite .
greene and kleitman generalize dilworth s theorem @xcite and frank describes an efficient algorithm that finds an optimal solution . if @xmath196 is a partially ordered set then one can find in polynomial time a collection of @xmath59 antichains @xmath197 that maximizes @xmath198 .
it follows that one can find in polynomial time an induced subgraph of a permutation graph with a maximal number of vertices that has chromatic number at most @xmath59 .
likewise , one can find a collection of @xmath59 cliques @xmath199 that maximizes @xmath200 .
when one bounds the number of vertices in the color classes the picture changes drastically . the following problem has been investigated in great detail due to its applications in various scheduling problems .
suppose we wish to find a vertex coloring with a minimal number of colors , such that each color class contains at most @xmath27 vertices @xcite .
this problem is np - complete on permutation graphs for each @xmath201 @xcite .
suppose one wishes to color the vertices of a graph with a minimal number of colors such that each color class induces a clique or an independent set .
we call this a homogeneous coloring .
the erds - hajnal conjecture states that for every graph @xmath40 there exists a @xmath202 such that every graph @xmath2 that does not contain @xmath40 as an induced subgraph can be homogeneously colored with at most @xmath203 colors @xcite .
considerable progress towards proving this conjecture is reported in @xcite .
the conjecture is known to be true for perfect graphs with @xmath204 and recently it was proved for bull - free graphs with @xmath205 @xcite .
one of the smallest graphs @xmath40 for which the conjecture is still open is @xmath206 .
wagner showed that finding the minimum number of colors in a coloring of this type is np - complete , even for permutation graphs @xcite . on the other hand , for any pair of nonnegative numbers @xmath5 and
@xmath6 , the class of permutation graphs that have a homogeneous coloring with @xmath5 cliques and @xmath6 independent sets ( possibly empty ) is characterized by a finite collection of forbidden induced subgraphs @xcite .
lonc mentions the following open problem in @xcite .
given a sequence of @xmath207 distinct positive integers .
find a partition of the sequence into @xmath3 increasing subsequences , each of 3 terms .
this is equivalent to finding a partition of the vertices of a permutation graph into triangles .
packing triangles in a graph is np - complete @xcite , even when restricted to chordal graphs , planar graphs and linegraphs @xcite .
interestingly , the question whether the vertices of a chordal graph can be partitioned into triangles can be solved in polynomial time @xcite .
packing triangles in splitgraphs , unit interval graphs and cographs can be solved in polynomial time @xcite .
for @xmath208 the @xmath37-packing problem is np - complete for splitgraphs @xcite .
when one allows besides triangles also edges in the packing and one wishes to maximize the number of vertices that are covered , then this packing problem becomes polynomial @xcite .
a permutation diagram is obtained as follows .
let @xmath209 and @xmath210 be two horizontal lines in the plane , one above the other .
label @xmath3 points on the topline and on the bottom line by @xmath211 .
connect each point on the topline by a straight linesegment with the point with the identical label on the bottom line .
a graph is a permutation graph if it is the intersection graph of the linesegments of a permutation diagram @xcite .
if @xmath2 is a permutation graph then its complement @xmath212 is also a permutation graph .
this is easy to see ; simply reverse the ordering of the points on one of the two horizontal lines . also notice this : for any independent set
the corresponding linesegments are noncrossing .
so they can be ordered left to right , which is of course a transitive ordering .
this shows that @xmath212 is a comparability graph and so also @xmath2 is a comparability graph .
the converse holds as well since transitive orderings of the vertices of @xmath2 and of @xmath212 provides the ordering of the points on the top and bottom line .
this can be seen as follows .
let @xmath213 and @xmath214 be transitive orientations of @xmath2 and @xmath212 .
we claim that @xmath215 is an acyclic orientation of the complete graph .
otherwise there is a directed triangle , and so two edges in the triangle are directed according to one of @xmath213 and @xmath214 and the third is directed according to the other one of @xmath213 and @xmath214 . but
this contradicts the transitivity of @xmath213 or the transitivity of @xmath214 .
likewise , @xmath216 is acyclic .
order the vertices on the topline according to @xmath215 and the vertices on the bottom line according to @xmath216 .
it is easy to check that this yields the permutation diagram .
the following characterization of permutation graphs illustrates the relation of this class of graphs to the class of interval graphs . consider a collection of intervals on the real line .
construct a graph of which the vertices are the intervals and make two vertices adjacent if one of the two intervals contains the other .
such a graph is called an interval containment graph .
consider a diagram of a permutation graph .
when one moves the bottom line to the right of the topline then the linesegments in the diagram transform into intervals .
it is easy to check that two linesegments intersect if and only if one of the intervals is contained in the other one .
this proves that permutation graphs are interval containment graphs .
now consider an interval containment graph .
construct a permutation diagram as follows . put the left endpoints of the intervals in order on the topline and the right endpoints in order on the bottom line of the diagram .
then one interval is contained in another interval if and only if the two linesegments intersect .
this proves that every interval containment graph is a permutation graph .
permutation graphs can be recognized in linear time @xcite .
notice that permutation graphs are perfect since they have no induced cycles of length more than four @xcite .
a graph is perfect if for every induced subgraph the clique number is the same as the chromatic number . if the clique and chromatic number of a graph are the same then these numbers can be computed in polynomial time @xcite .
however , in a permutation graph computing a largest clique corresponds to finding a longest increasing subsequence and this can be computed very efficiently @xcite .
( notice that a permutation graph has a clique or an independent set with at least @xmath217 vertices @xcite . )
|
we show that there exist efficient algorithms for the triangle packing problem in colored permutation graphs , complete multipartite graphs , distance - hereditary graphs , @xmath0-modular permutation graphs and complements of @xmath0-partite graphs ( when @xmath0 is fixed ) .
we show that there is an efficient algorithm for @xmath1-packing on bipartite permutation graphs and we show that @xmath1-packing on bipartite graphs is np - complete .
we characterize the cobipartite graphs that have a triangle partition .
|
markov processes are used to model a wide variety of physical , chemical and biological phenomena @xcite . while calculating the mean of a random
variable is often relatively straightforward , obtaining its dispersion can be much harder .
a prominent example is the expression for the dispersion of the number of steps of a particle hopping in a one - dimensional lattice @xcite .
fluctuating currents are important random variables in stochastic thermodynamics @xcite .
if their mean is nonzero in the steady state , the system is out equilibrium . a standard method to calculate
the dispersion of this random variable is to obtain the scaled cumulant generating function as the maximum eigenvalue of the so - called modified generator @xcite .
however , determining this eigenvalue as a function of the transition rates is typically difficult , even for systems with a small number of states .
the problem of calculating the dispersion of fluctuating currents can be overcome with an elegant method introduced by koza @xcite . with koza
s method , instead of calculating the maximum eigenvalue , the dispersion is obtained from coefficients of the characteristic polynomial associated with the modified generator . for instance , this method is useful to prove a bound on the minimal number of states of a one - dimensional lattice in terms of this dispersion @xcite ( see @xcite for discussions on the importance of this bound ) .
recently , we have used this method to derive a bound on the minimal energetic cost of small uncertainty in biomolecular processes @xcite .
other references that use this method are @xcite ( see also @xcite for related methods that use perturbation theory ) .
as we point out in the present paper , koza s method can be used to calculate the dispersion of any random variable that changes whenever a jump occurs during a stochastic trajectory .
besides fluctuating currents , another random variable pertaining to this class is the dynamical activity , for which the total number of transitions is counted .
this activity appears in fluctuation - dissipation relations out of equilibrium @xcite and is studied in so - called dynamical phase transitions @xcite .
we show that a similar method can be used to calculate the dispersion of a second class of random variables .
this class corresponds to variables counting the fraction of time a stochastic trajectory spends in a cluster of states .
our calculations lead to an expression for the dispersion of this variable in terms of the coefficients of the characteristic polynomial associated with a discrete - time modified generator .
the time spent in a state is a key random variable in the problem of a cell estimating an external ligand concentration by monitoring the time a receptor is in the bound state , i.e. , occupied by an external ligand . the uncertainty on the estimated concentration sets a fundamental limit on the cell s ability to sense the external environment , as established in the seminal work of berg and purcell @xcite ( see also @xcite ) .
we analyze a four - state model for a two - component network that estimates an external concentration @xcite .
the two - component network is composed by the receptor and an internal protein that can be phosphorylated due to atp consumption .
this model provides an example for which we can calculate the dispersion for three different random variables : current and activity from the first class , and time spent in the phosphorylated state from the second .
specifically , we compare the uncertainty of the estimated external concentration that is obtained from these three different random variables and discuss its relation with energy dissipation .
the paper is organized as follows .
the expression for the dispersion of the first class of random variables is provided in sec .
[ sec2 ] . in sec .
[ sec3 ] , we introduce a method to calculate the dispersion for the second class . sec .
[ sec4 ] contains the application to the problem of sensing an external concentration .
we conclude in sec .
[ sec5 ] . in appendices [ app1 ] and [ app2 ] ,
we perform detailed derivations for the first and second classes of random variables , respectively .
the calculations of the four - state markov process considered in sec .
[ sec4 ] are explained in appendix [ app3 ] .
the transition rate from state @xmath0 to state @xmath1 of a generic continuous - time markov process with a finite number of states @xmath2 is denoted by @xmath3 .
the time evolution of the occupation probability of state @xmath0 at time @xmath4 is determined by the master equation @xmath5.\ ] ] for the discussion below it is convenient to write this equation in the from @xmath6 where @xmath7 is the probability vector with @xmath2 components and @xmath8 is the stochastic matrix with elements @xmath9_{ij}\equiv\left\{\begin{array}{ll } w_{ji } & \quad \textrm{if } i\neq j\\ -\sum_kw_{ik } & \quad \textrm{if } i = j \end{array}\right.\ , .
\label{generator}\ ] ] furthermore , the stationary probability of state @xmath0 is written as @xmath10 .
two key quantities in this paper are the ( stationary ) average probability current from state @xmath0 to @xmath1 @xmath11 and the average activity @xmath12 this activity is the average number of transitions per time between the pair of states @xmath13 . quite generally , we consider a random variable @xmath14 that is a functional of the stochastic trajectory .
this trajectory is a sequence of jumps during a time interval @xmath15 , which is assumed to be long , i.e. , the expressions below are valid in the formal limit @xmath16 .
the random variable @xmath14 is then characterized by the increments @xmath17 : it increases by @xmath17 whenever a jump from @xmath0 to @xmath1 occurs . if we choose the increments @xmath18 and zero for any other pair of states , we obtain @xmath19 , where here and in the following the brackets denotes an average over stochastic trajectories in the stationary state . on the other hand , the choice @xmath20 results in @xmath21 .
the `` velocity '' associated with the generic variable @xmath14 is @xmath22 furthermore , the dispersion is defined as @xmath23 an expression for these quantities in terms of the transition rates is obtained in the following way .
the scaled cumulant generating function associated with @xmath14 is given by the maximum eigenvalue of the modified generator @xcite @xmath24_{ij}\equiv\left\{\begin{array}{ll } w_{ji}\exp(z \theta_{ji } ) & \quad \textrm{if } i\neq j\\ -\sum_kw_{ik } & \quad \textrm{if } i = j \end{array}\right.\ , .
\label{modgenerator}\ ] ] the characteristic polynomial of this matrix reads @xmath25 where @xmath26 are coefficients of the polynomial and @xmath27 is the identity matrix .
it turns out that the velocity and dispersion can be written in terms of these coefficients and its derivatives at @xmath28 .
this procedure is very convenient since we can calculate the dispersion without knowing the full scaled cumulant generating function , which is a root of this polynomial .
the specific expressions are @xmath29 and @xmath30 where primes denote derivatives with respect to @xmath31 and the lack of explicit @xmath31 dependence in the coefficients denotes evaluation at @xmath28 .
these expressions are derived in appendix [ app1 ] .
they have been obtained by koza @xcite for the random variable @xmath14 being the current .
we are simply pointing out here that they are also valid for any variable @xmath14 with generic increments @xmath17 , i.e. , in particular also for the activity .
the second class of random variables @xmath32 corresponds to the time the stochastic trajectory spends in a state or in a cluster of states .
the first moment of this random variable is easy to determine .
for example , the average time spent in state @xmath0 is the stationary probability @xmath10 multiplied by the total time , i.e. , @xmath33 .
generally , we denote a cluster of states by @xmath34 , with @xmath35 if state @xmath36 and @xmath37 otherwise .
the `` velocity '' associated with @xmath32 is defined as @xmath38 furthermore , the dispersion reads @xmath39 we discretize time in order to calculate this dispersion .
the discrete - time stochastic matrix is @xmath40 where @xmath41 is the size of the time - interval .
since we are interested in the continuous - time limit , at the end of the calculation the limit @xmath42 is taken .
we define the modified matrix @xmath43_{ij}\equiv\left\{\begin{array}{ll } \epsilon w_{ji}\exp(z \delta_{\gamma , i } ) & \quad \textrm{if } i\neq j\\ \left(1-\epsilon\sum_kw_{ik}\right)\exp(z \delta_{\gamma , i } ) & \quad \textrm{if } i = j \end{array}\right.\ , .
\label{modgeneratordisc}\ ] ] the maximum eigenvalue of this modified matrix generates the moments of a random variable that is the number of discrete - steps the system is in states pertaining to @xmath34 during the time interval @xmath15 @xcite .
defining the coefficients @xmath44 we obtain @xmath45 and @xmath46 where @xmath47 these expressions for velocity and dispersion are derived in appendix [ app2 ] , where we also show that eqs .
( [ dispersion2 ] ) and ( [ dispersion3 ] ) are in perfect agreement with numerical simulations .
the expression for the dispersion of this second class of random variables is the main technical result of this paper .
given a continuous - time markov process with a small number of states , small in the sense that the matrix in eq .
( [ modgeneratordisc ] ) can be handled analytically , one can readily obtain the dispersion @xmath48 as a function of the transition rates using eqs .
( [ dispersion2 ] ) and ( [ dispersion3 ] ) . in the next section
we calculate the dispersion from different random variables and analyze their relation with energy dissipation .
related studies that analyze the relation between the relative uncertainty associated with another random variable , the life - time of cluster of states , with energy dissipation are @xcite for a biological timer , and @xcite for a receptor that dissipates chemical energy . in general
, this lifetime of a cluster of states is different from the time an stochastic trajectory spends in a cluster of states .
depend on the occupancy of the receptor .
, width=170 ] as an application , we consider a two - component network of a cell estimating an external concentration @xmath49 represented in fig .
a review on two - component signaling networks is given in @xcite . in our model ,
transition rates can depend on @xmath49 , and , therefore , the moments of a random variable @xmath50 , pertaining to any of the two classes , will depend on @xmath49 .
the relative uncertainty in estimating @xmath49 by monitoring @xmath50 for a time @xmath15 is written as @xmath51 characteristically , the uncertainty decreases with the integration time @xmath15 .
the prefactor @xmath52 is the integration time required to get a relative uncertainty smaller than 100%
. we shall refer to it as the inference time .
below we will calculate this inference time obtained from different random variables in a two - component network .
the inference time can also be expressed as the inverse of the signal - to - noise ratio normalized by the integration - time time multiplied by the relative change in the external concentration @xcite .
the receptors sitting on the cell membrane constitute the first layer of the two - component sensory network .
in the two - state model for a single receptor shown in fig . [ fig2 ] , the receptor can be either bound by a ligand or free .
the parameters determining the transition rates in fig .
[ fig2 ] are the external concentration @xmath49 , the concentration for which both states are equally probable @xmath53 , and the average lifetime of the bound state @xmath54 .
the key feature of these transition rates is that the binding rate is proportional to the external ligand concentration @xmath49 . and from bound to unbound @xmath55.,width=170 ] if the concentration is estimated by monitoring the time spent in the bound state , the inference time is @xmath56 which is the result obtained by berg and purcell @xcite .
the subscript in @xmath57 indicates that the random variable is from the second class .
if instead of monitoring the time spent in the bound state the activity , i.e. , the total number of transitions is monitored , a different uncertainty is obtained . in @xcite
, it has been shown that the inference time obtained from the activity is @xmath58 for low concentrations @xmath59 , the uncertainty obtained by monitoring the activity is half of the berg and purcell uncertainty , i.e. , @xmath60 . this same limit is achieved by estimating the concentration with a maximum likelihood method applied to the probability of a trajectory of the two - state model @xcite .
it is straightforward to obtain eq .
( [ actlim ] ) with the formalism from sec .
[ sec2 ] and eq .
( [ bplim ] ) with the formalism from sec .
[ sec3 ] , in both cases one has to calculate the characteristic polynomial of a 2@xmath612 matrix .
the second layer of the two - component network is an internal protein which can be in states @xmath62 and @xmath63 , where @xmath63 is the phosphorylated form of the protein .
the full two - component network then has four states @xcite , as shown in fig .
similar models for two - component networks consider a large number of internal proteins @xcite .
if the receptor is bound by a ligand the following phosphorylation reaction can happen @xmath64{\kappa_+ } y^*+adp , \label{eqpho}\ ] ] where @xmath65 are transition rates .
if the receptor is empty , dephosphorylation may take place , i.e. , @xmath66{\omega_+ } y+p_i .
\label{eqdepho}\ ] ] for thermodynamic consistency these transition rates must fulfill the generalized detailed balance relation @xcite @xmath67 where @xmath68 is the free energy liberated in one atp hydrolysis and we are setting @xmath69 throughout .
this model is in equilibrium only if @xmath70 , which means that no energy is dissipated .
otherwise , chemical free energy is dissipated , with the rate of dissipated heat being characterized by the entropy production @xcite . .
the transition rates for the internal protein are defined in eqs .
( [ eqpho ] ) and ( [ eqdepho ] ) . , width=170 ]
if we are interested in the occupation of the receptor , we obtain the two - state model from fig .
[ fig2 ] by integrating @xmath62 out since the receptor is independent of the internal protein .
hence , the inference time associated with the time the receptor is bound in this four - state model is as given by eq .
( [ bplim ] ) and the one associated with the activity of binding / unbinding transitions is as given by eq .
( [ actlim ] ) .
the external concentration can also be estimated by monitoring random variables related to the internal protein .
we consider three random variables : the time the internal protein is in the phosphorylated form , the probability current and the activity of the transitions involving the @xmath62 protein .
note that the present four - state model has only one probability current , which , for @xmath71 , goes in the clockwise direction in fig .
[ fig3 ] . the inference time is denoted by @xmath72 for the time spent in the phosphorylated state . for random variables of the first class we denote @xmath73 the inference time related to the current and @xmath74 the one related to the activity .
the calculation of these quantities using the expressions from sec .
[ sec2 ] and sec .
[ sec3 ] is explained in appendix [ app3 ] .
we first compare @xmath73 with @xmath72 . in equilibrium ,
the current is zero .
hence , the inference time @xmath73 diverges for @xmath70 .
it turns out that also for monitoring the time spent in the phosphorylated state @xmath63 the inference time @xmath72 diverges in equilibrium .
this divergence comes from the fact that the sensitivity @xmath75 goes to zero in equilibrium , since the probability of being in the cluster of states for which the protein is phosphorylated becomes independent of @xmath49 for @xmath76 . for the current variable
we can obtain the following relation between heat dissipation and inference time .
in @xcite we have obtained a general relation between uncertainty and dissipated free energy that for the present model reads @xmath77\ge 2 , \label{neweq1}\ ] ] where @xmath78 is the thermodynamic entropy production that quantifies the rate of dissipation due to @xmath79 consumption .
this entropy production is simply given by @xmath80 , where @xmath81 is the average current . from the definition of inference time in eq .
( [ uncertaintydef ] ) and from eq .
( [ neweq1 ] ) we obtain @xmath82 where the term @xmath83 is the sensitivity of the average current to changes in @xmath49 .
prominently , this lower bound on inference time decreases with increasing dissipation .
this expression provides a tradeoff between inference time , sensitivity and dissipation .
since relation ( [ neweq1 ] ) is general , the tradeoff relation ( [ tradeoff ] ) is valid beyond the specific model considered in this section . for the following calculations we set the transition rates to @xmath84 where @xmath85 is the time - sale of the reversed transitions in eqs .
( [ eqpho ] ) and ( [ eqdepho ] ) .
we define the dimensionless parameter @xmath86 first we take @xmath87 , which corresponds to the internal protein reactions being fast .
the inference time related to the current in this limit is @xmath88 for large @xmath68 , this inference time reaches its minimal value given by the uncertainty obtained with the activity of the binding / unbinding transitions @xmath89 in eq .
( [ actlim ] ) . in this case , the phosphorylated form of the protein @xmath63 is strongly correlated with the bound state , while @xmath62 is correlated with the unbound state .
the binding of ligand is immediately followed by the transition @xmath90 , whereas the unbinding of a ligand is followed by @xmath91 .
therefore , monitoring the current becomes equivalent to monitoring the activity of the receptor .
( black solid line ) and the time spent in the phosphorylated state @xmath72 ( red dashed line ) as a function of @xmath68 for @xmath92 , @xmath93 , and @xmath94 .
, width=207 ] in the limit @xmath87 with any finite @xmath95 , the time spent in the phosphorylated state leads to @xmath96 , with @xmath97 given in eq .
( [ bplim ] ) .
while for a large @xmath95 this result can be explained by the fact that @xmath63 ( @xmath62 ) and bound ( unbound ) receptor become strongly correlated it is surprising that for any finite @xmath95 monitoring the internal protein is equivalent to monitoring the receptor . for both random variables , current and time - spent in the phosphorylated state , the best uncertainty that a variable related to the internal protein can reach
is bounded by the uncertainty obtained by monitoring a corresponding random variable related to the receptor directly .
the main difference between these two random variables in the present limit is that the precision related to the time spent in the phosphorylated state does not improve by increasing @xmath98 .
hence , there is a region for which increasing energy dissipation leads to an exponential decrease in @xmath73 , while @xmath72 remains approximately constant , as shown in fig .
[ fig4 ] .
another important limit corresponds to the case where the reversed transitions of the internal protein are much slower than unbinding of the ligand , i.e. , @xmath99 . in this limit , where the relevant time - scale for the averaging time required for a small uncertainty is @xmath100 , the inference times related to current and time spent in the phosphorylated state are @xmath101}{c ( c - c_0)^2 c_0 ( \textrm{e}^{\delta\mu/2}-1)^2 ( \textrm{e}^{\delta\mu/2}+1)}\tau_y+\textrm{o}(\tau_b ) \label{sy1jinf}\ ] ] and @xmath102 } { c^2 c_0 ^ 2 ( \textrm{e}^{\delta\mu/2}-1)^2 ( \textrm{e}^{\delta\mu/2}+1)}\tau_y + \textrm{o}(\tau_b ) , \label{sy2inf}\ ] ] respectively . in these expressions
we are assuming that the prefactors are large compared to @xmath103 .
particularly , for large @xmath68 both the prefactor in eq .
( [ sy1jinf ] ) and the prefactor in eq .
( [ sy2inf ] ) decay with @xmath104 . in this case
, the expressions remain valid for @xmath105 .
comparing eq .
( [ sy1jinf ] ) with eq .
( [ sy2inf ] ) we see that for @xmath49 close to @xmath53 @xmath106 , while for @xmath107 or @xmath59 , @xmath108 .
hence for intermediate values of the concentration close to @xmath53 the inference time related to the current is larger .
this feature is quantified in fig .
[ fig5 ] , showing that the ratio @xmath109 is smaller than @xmath110 in an intermediate concentration range .
the precise range for which monitoring the time spent in the phosphorylated state leads to a smaller uncertainty than monitoring the current depends on @xmath98 . in the @xmath111 plane for @xmath112 and @xmath93 .
the region in blue corresponds to @xmath113 , where monitoring the current leads to a smaller inference time compared to monitoring the time spent in the phosphorylated state .
, width=207 ] finally , we analyze the inference time arising from monitoring the activity of the transitions involving changes in the internal protein @xmath62 . for the transition rates set in eq .
( [ choicerates ] ) we observe that @xmath114 . furthermore , in the limit of @xmath87 and large @xmath98 , for which @xmath63 and bound receptor
are strongly correlated , the inference time @xmath115 , with @xmath116 given by eq .
( [ actlim ] ) .
the main feature of the inference time @xmath74 is that it can be finite even in equilibrium .
therefore , while the other two variables related to @xmath62 , namely , the current and the time spent in the phosphorylated state , require dissipation to infer an external concentration , by monitoring the activity it is possible to infer the concentration without any dissipation . in order to make this feature transparent we set the transition rates as @xmath117 which from eq .
( [ eqmu ] ) imply @xmath70 .
the factor @xmath118 quantifies how much faster the transitions that change @xmath62 are when the receptor is bound .
if @xmath119 internal transitions are faster with the receptor bound and if @xmath120 they are faster for a free receptor .
the inference time with the rates in eq .
( [ choicerates2 ] ) becomes @xmath121 ) } { c^2 c_0 ^ 2 ( a-1)^2}\tau_b,\ ] ] where @xmath122 .
for large @xmath123 , @xmath124 , with @xmath57 given by eq .
( [ bplim ] ) .
this result can be explained as follows . for @xmath125 , whenever the receptor is bound a large number of internal transitions take place while for a unbound receptor this number is much smaller .
hence , counting the number of transitions of the internal protein , i.e. , the activity of the internal protein , becomes equivalent to monitoring the time spent in the bound state .
the surprising feature here is that depending on the random variable , it is possible to reach the berg and purcell limit by monitoring the internal protein with no energy dissipation .
let us compare our results with recent work on two - component signaling networks that estimate an external concentration @xcite . in these works
the random variable is the fluctuating number of internal proteins that are in the phosphorylated ( or active ) state , which can be viewed as memory of the receptor occupancy . as demonstrated in @xcite , the uncertainty related to this random variable diverges in equilibrium , similar to current and time spent in the phosphorylated state .
a main point of the present paper is that the relation between dissipation and uncertainty depends on which random variable we choose to estimate the external concentration , as manifested by the fact that activity allows for dissipationless inference .
we note that for a different model with internal proteins directly binding to free receptors the uncertainty estimated from the number of internal proteins can also be finite in equilibrium @xcite .
the rate of mutual information in bipartite systems , for which the two - component network is an example , has been studied in @xcite . in these references
it has been shown that the rate of mutual information between an external process , which corresponds to the receptor in our model , and an internal process , corresponding to the protein @xmath62 , can be non - zero even in equilibrium .
a posteriori , this result indicates that even in equilibrium the inference of an external concentration with finite uncertainty should be possible if an appropriate random variable is monitored .
we have shown that the activity is a random variable that achieves this dissipationless inference .
we have developed a general formalism that allows for the calculation of the dispersion of the time a stochastic trajectory spends in a cluster of states that can be applied to arbitrary markov processes with a finite number of states .
this method is potentially useful for any application where the dispersion of this random variable is of interest .
furthermore , our formalism is similar to the formalism to calculate the dispersion for another class of random variables comprising current and activity , which has been pioneered by koza @xcite .
the expressions in sec .
[ sec2 ] and sec .
[ sec3 ] thus provide a unified recipe to calculate the dispersion of these two classes of random variables .
these expressions have been used in the four - state model for a two - component network estimating an external ligand concentration in sec .
[ sec4 ] . with this application ,
we have compared the uncertainties of the external concentration that are obtained by monitoring three different random variables : current , activity of the internal protein , and the time the internal protein is in the phosphorylated state .
the main results obtained with our case study are as follows . the best performance ( smallest uncertainty ) that a random variable related to the internal protein can achieve
is limited by random variables associated with the receptor , which is the first layer of the network .
this best performance is achieved when the internal protein is much faster than the receptor . in general ,
determining which random variable leads to the smallest uncertainty depends on parameters like the external concentration and the chemical potential driving the phosphorylation cycle .
if the internal protein is much slower than the receptors , the integration time to obtain a reliable estimate from the internal protein must be much larger than the typical time of a transition of the protein , which is much larger than the lifetime of the bound state of the receptor . finally , while current and time spent in the phosphorylated state require energy dissipation for a finite uncertainty , it is possible to estimate the external concentration from the activity of the internal protein even in equilibrium , which we call dissipationless inference .
recently , we have obtained a universal bound on the energetic cost of the relative uncertainty of a generic probability current @xcite . as shown here , this bound implies a general bound on the inference time based on a current variable .
an intriguing question is whether it is possible to find such universal relations between energy dissipation and uncertainty for other random variables like the activity or the time a trajectory spends in a cluster of states .
the formalism developed in this paper provides a tool to investigate the existence of such thermodynamic uncertainty relations that connect energy dissipation with uncertainty .
the maximum eigenvalue of the modified generator in eq .
( [ modgenerator ] ) is denoted @xmath126 .
it can be shown that this maximum eigenvalue is the scaled cumulant generating function associated with the random variable @xmath14 @xcite .
the velocity and dispersion are related to @xmath126 as @xmath127 and @xmath128 respectively .
the lack of explicit @xmath31 dependence denotes evaluation at @xmath28 . since @xmath126 is a root of the polynomial in eq .
( [ eqdet ] ) , it follows that @xmath129^n=0 . \label{poly2}\ ] ] using the fact that @xmath130 , we obtain eqs .
( [ eqjc ] ) and ( [ eqdc ] ) from eqs .
( [ jcum ] ) and ( [ dcum ] ) by taking the first and second derivatives with respect to @xmath31 in eq .
( [ poly2 ] ) , respectively .
these calculations can be found in more detail in @xcite , for example .
the time interval @xmath15 is discretized in @xmath131 intervals of size @xmath41 , with @xmath132 .
the time evolution of the corresponding markov chain is determined by the discrete - time stochastic matrix in eq .
( [ discsto ] ) .
we consider a random variable @xmath133 which counts in how many of the @xmath131 segments the system was in the cluster of states @xmath34 .
we denote @xmath134 the maximum eigenvalue of the matrix in eq .
( [ modgeneratordisc ] ) .
hence , @xmath134 is a root of the polynomial in eq .
( [ eqdet2 ] ) , i.e. , @xmath135^n=0 . \label{poly3}\ ] ] for @xmath28 , the matrix in eq .
( [ modgeneratordisc ] ) becomes the discrete - time stochastic matrix in eq .
( [ discsto ] ) , which implies @xmath136 . by taking first and second derivatives in eq .
( [ poly3 ] ) and setting @xmath28 we obtain @xmath137 and @xmath138 furthermore this maximum eigenvalue is the generating function associated with @xmath133 @xcite , i.e. , @xmath139 hence , in the large @xmath131 limit , @xmath140 and @xmath141 the first cumulant of the discrete and continuous variables fulfill the relation @xmath142 this relation comes from the fact that the stationary distribution of the continuous - time markov process is the the same as the discrete - time markov chain . the second cumulant of the continuous variable is related to the second cumulant of @xmath133 by @xmath143 the factor @xmath41 on the right hand side accounts for the correct dimension .
this relation is equivalent to @xmath144 .
note that in eqs .
( [ eq1app3 ] ) and ( [ eq2app3 ] ) the average in the left hand side is over continuous - time trajectories , whereas the average in the right hand side is over discrete - time trajectories . from eqs .
( [ psip1 ] ) , ( [ psip2 ] ) and ( [ eq1app3 ] ) we obtain eq .
( [ velocity2 ] ) . from eqs .
( [ psipp1 ] ) , ( [ psipp2 ] ) and ( [ eq2app3 ] ) we obtain eq .
( [ dispersion2 ] ) , with @xmath145 given by eq .
( [ dispersion3 ] ) . even though we do not have a rigorous proof for eq .
( [ eq2app3 ] ) we have verified numerically its validity for several different networks of states by comparing the dispersion obtained from continuous - time numerical simulations with the dispersion obtained from our analytical expressions in eqs .
( [ dispersion2 ] ) and ( [ dispersion3 ] ) .
for example , we have performed this test for the networks of states in fig .
[ fig6 ] with the results shown in fig .
the four states of the model in fig . [ fig3 ] are identified as @xmath146 , @xmath147 , @xmath148 , and @xmath149 , where @xmath150 ( @xmath151 ) denotes bound ( unbound ) receptor .
the transition rates are given by @xmath152 , @xmath153 , @xmath154 , @xmath155 , @xmath156 , and @xmath157 .
this four - state model has only one probability current , which is the same in any of the four links in fig .
[ fig3 ] . if we take the current from @xmath110 to @xmath158 , the respective modified generator in eq .
[ modgenerator ] is @xmath159 where @xmath160 , @xmath161 , @xmath162 , and @xmath163 .
the subscript @xmath164 indicates the modified generator for the current . in the case of the activity of the transitions changing the internal portein state , which are transitions involving the pairs @xmath165 and @xmath166 , the modified generator reads @xmath167 the inference times @xmath73 and @xmath74 are calculated from the coefficients of the characteristic polynomial in eq .
( [ eqdet ] ) with expressions ( [ eqjc ] ) , ( [ eqdc ] ) , and ( [ uncertaintydef ] ) , where for @xmath73 the modified generator is in eq .
( [ matrix1 ] ) and for @xmath74 the modified generator is in eq .
( [ matrix2 ] ) . in both cases
, we can get full analytical expressions for arbitrary transition rates , however they are too long to be displayed here . for the time spent in the phosphorylated state of the protein ,
the corresponding cluster is @xmath168 .
the matrix in eq .
( [ modgeneratordisc ] ) then becomes @xmath169 with the coefficients of the characteristic polynomial associated with this matrix , as given by eq .
( [ eqdet2 ] ) , the inference time @xmath72 is obtained from eqs .
( [ velocity2 ] ) , ( [ dispersion2 ] ) , ( [ dispersion3 ] ) and ( [ uncertaintydef ] )
. we also get a long analytical expression for @xmath72 in terms of the transition rates .
10 [ 2]#2 [ 1]#1 [ 1]#1 , ( , amsterdam , ) . c. w. gardiner , _ handbook of stochastic methods for physics , chemistry and the natural sciences _ ,
springer series in synergetics vol . 13 , 3rd ed .
( springer - verlag , berlin , 2004 ) . ,
( , ) . , , ( ) . , , ( )
. and , , ( ) .
, , ( ) . and , , ( ) . and , , ( )
. and , , ( ) . , , and , , ( )
, , and , ( ) . , , and , ( ) . , , and , , ( ) . and , , ( ) . , , and , , ( ) . , , , , , and , , ( ) . and , , ( ) . and , , ( ) . and , , ( ) . and , , ( ) . and , , ( ) . and , , ( ) . and , , ( ) . , , , , , and , , ( ) . and , , ( ) . , , and , , ( ) . , , and , , ( ) . and , , ( ) . and , , ( ) . , , ( ) . and , , ( ) . , , ( ) .
, , , and , , ( ) . , , and , , ( ) . , , and , , ( )
, , , and , , ( ) . , , and , , ( )
|
we derive expressions for the dispersion for two classes of random variables in markov processes .
random variables like current and activity pertain to the first class , which is composed by random variables that change whenever a jump in the stochastic trajectory occurs .
the second class corresponds to the time the trajectory spends in a state ( or cluster of states ) .
while the expression for the first class follows straightforwardly from known results in the literature , we show that a similar formalism can be used to derive an expression for the second class . as an application
, we use this formalism to analyze a cellular two - component network estimating an external ligand concentration . the uncertainty related to this external concentration
is calculated by monitoring different random variables related to an internal protein .
we show that , _ inter alia _ , monitoring the time spent in the phosphorylated state of the protein leads to a finite uncertainty only if there is dissipation , whereas the uncertainty obtained from the activity of the transitions of the internal protein can reach the berg and purcell limit even in equilibrium .
|
the general philosophy for bootstrap or permutation methods for testing hypotheses is to simulate the variation of the test statistic by generating the sampling distribution which assumes both that the null hypothesis is true , and that the data in the sample is somehow representative of the population .
this philosophy works well for testing hypotheses regarding the correlation coefficient , but is inapplicable for testing hypotheses for a single parameter like the population mean , since the two assumptions are contradictory .
for example , how can we assume both that the mean of the population is @xmath0 and that the individuals in the sample with a mean @xmath1 are representative of the population ? one naive way that has been used is the shift method @xcite , where each individual in the sample is shifted by @xmath3 , which is essentially equivalent to testing the hypotheses with a confidence interval .
the shift method hypothesizes that the variance between the sampled individuals is representative of the population variance , yet it loses any semblance to the assumption that the sampled individuals themselves are representative of the population .
the bootstrap method provides a more elegant resolution .
the philosophy of the mirror bootstrap method for testing hypotheses regarding one population parameter is to assume both that the null hypothesis is true , and that the individuals in our sample are as representative as they could be without assuming more extreme cases than observed .
for example , the mirror bootstrap method for testing hypotheses of one mean uses a generated symmetric distribution constructed by reflecting the original sample around the hypothesized population mean @xmath2 .
the rationale for the mirror bootstrap is that , on the one hand , to carry out a test against the null hypothesis one assumes that it s true and computes the rarity of the evidence , and on the other hand bootstrap methods try to assume that the sample is in some ways representational of the population . obviously for
a test of one mean , one ca nt assume both that the true population mean is @xmath2 and that the sampled individuals are representational of the population ( except in the rare cases where the sample mean equals @xmath2 , and those cases are not interesting ) .
but we can do the next best thing : assume that the individuals sampled are as representational as they can be with the null hypothesis being true . if we also do nt want to assume individuals more extreme than those observed , the easiest way to do that is to assume that for every observed individual there is a partner somewhere in the population who is on the opposite side of the hypothesized population mean @xmath2 .
thus , suppose we have a random sample of size @xmath4 and sample mean @xmath5 of a random variable @xmath6 whose population distribution is unknown , and we wish to test against the null hypothesis @xmath7 with some alpha level .
the mirror bootstrap method goes as follows : reflect the sample around @xmath8 , getting a symmetric sample of size @xmath9 ; then repeatedly sample with replacement samples of size @xmath4 from this symmetric constructed population , counting how many times the means of these bootstrap samples are further from @xmath2 than the original sample mean @xmath5 ; if this proportion is less than alpha , reject the null hypothesis .
how well does the bootstrap method work in practice ?
simulations of the performance of the mirror bootstrap for testing hypotheses of one mean show that , while the method is slightly on the conservative side for very small samples , its validity and power quickly approach that of the widely used t - test . the author used maple 15 to implement the simulations , and
would gladly share the scripts upon demand .
for a quick check of the power and validity of the mirror bootstrap method , we compare its performance to that of the widely used two - tailed t - test and to the shift bootstrap test , in the cases where the population is normally distributed , very skewed , and bimodal .
the results indicate that the shift bootstrap test has serious problems with validity for small samples ( @xmath10 ; the mirror bootstrap method is more conservative than the t - test for very small samples ( @xmath11 ) but matches the validity and power of the t - test for samples size @xmath12 . to represent these three cases , we use the following distributions : standard normal distribution @xmath13 , highly skewed distribution gamma(2,2 ) , and a bimodal distribution @xmath14 xor @xmath15 , all normalized so that the population means are at 0 , and the population standard deviations are 1 .
we simulated drawing a sample of specified size 10,000 times in each case , bootstrapping 1000 samples , and doing two - tailed hypothesis tests with alpha level 0.05 .
the results are as follows . for normal and skewed populations ,
mirror bootstrap is conservative for very small samples , after which it performs comparably or better than the t - test .
it should nt come as a surprise that the performance of the t - test is unbeaten for the normal distribution , since the t - test was specifically designed with the assumption of normality . for a bimodal population ,
mirror bootstrap is nt as conservative , and slightly overshoots the target alpha of 0.05 for small samples of size 5 .
it performs comparably or better than the t - test .
we simulated drawing a sample of specified size 1000 times in each case , bootstrapping 1000 samples , and doing two - tailed hypothesis tests with alpha level of 0.05 .
the results are as follows . for the distribution skewed to the left
, we used a mirror version of the normalized gamma(2,2 ) distribution .
for all three distributions , the power curve for mirror bootstrap is slower for small samples of size five , but are seen to converge to the power curve of the t - test for samples of size 10 .
an interesting result happens when looking at the case where the population is heavily skewed to the right , as is the case for the normalized gamma(2,2 ) : in some situations , the mirror bootstrap outperforms the t - test .
for a more systematic assessment of the mirror bootstrap method s validity and power , we turn to the family of g - and - h distributions : @xmath16 which , in the case of @xmath17 is @xmath18 .
depending on the parameters @xmath19 and @xmath20 , the distributions vary in skewness and kurtosis .
the family includes the standard normal distribution , the lognormal distribution with its long skinny tail , a symmetric distribution with heavy tails , a skewed symmetric distribution with heavy tails , and everything in - between @xcite .
skewness is measured by @xmath21 and kurtosis is measured by @xmath22 , where the @xmath23 are the moments around the mean , defined as usual : @xmath24 $ ] , and @xmath25 $ ] .
following @xcite , the skewness and kurtosis of various g - and - h distributions are as follows , with skewness of 0 whenever @xmath26 and undefined for @xmath27 , and kurtosis undefined for @xmath28 : the g - and - h distribution with @xmath29 is an example of a very skewed distribution with heavy tail .
the skewness and kurtosis are undefined for the population with such distribution ; the mean is approximately 0.8 .
wilcox @xcite notes that a sample drawn from such population has statistics that do not correspond to the population parameters , showing in particular that a sample of 100,000 observations had skewness of 120 and kurtosis of about 18,400 . for this distribution
, power is nt the issue for the mirror bootstrap or the t - test .
power curves are comparable , even for small samples .
validity is a real problem for both methods : mirror bootstrap performs a little better than the t - test for small samples , though still performing much higher than the assumed alpha level of 0.05 . for
large samples , both methods plateau with approximately 20% of type i errors for samples of size less than 100 , and very slowly descend as the sample size grows . in practice
, one may ask if the mean is even an appropriate measure of central tendency for such a distribution . to test the validity of the mirror bootstrap method in testing hypotheses of one mean , we simulated drawing a sample of specified size from a specified g - and - h distribution with @xmath30 , doing so 10,000 times in each case to get an accurate estimate of proportion of type i errors in a two - tailed hypothesis test with alpha level 0.05 . in each test , we bootstrapped 1000 samples .
the results are as follows : as the skewness increases , the mirror bootstrap loses validity for small samples , though the performance increases with sample size .
the validity is good for @xmath31 , and still decent for @xmath32 , but after that it becomes unacceptably high .
compared to the validity of the t - test , mirror bootstrap performs as well as or better , especially for small samples : to test the validity of the mirror bootstrap method in testing hypotheses of one mean , we simulated drawing a sample of specified size from a specified g - and - h distribution with @xmath26 , doing so 10,000 times in each case to get an accurate estimate of proportion of type i errors in a two - tailed hypothesis test with alpha level 0.05 . in each test , we bootstrapped 1000 samples .
the results are as follows : mirror bootstrap method is conservative for very small samples ( @xmath33 ) , and stays somewhat conservative for distributions with heavy tails ( @xmath34 ) , but for lighter tails the validity holds just below the ideal level of 0.05 , so the method is slightly conservative . comparing the bootstrap method s performance to that of the t - distribution , the mirror bootstrap is consistently more conservative than the t - test , though only slightly so for @xmath12 : overall , the proportion of type i errors appears reasonable for @xmath35 as long as the tails are not too heavy , and this result seems fairly consistent for various sample sizes :
the author would like to thank dr .
wilcox for his helpful feedback , and to the lock5 team ( drs .
robin lock , patti lock , kari lock , and masters eric lock and dennis lock , of www.lock5stat.com ) for their work in making bootstrapping methods readily accessible to teachers of introductory statistics .
hoaglin , d.c .
( 1985 ) . summarizing shape numerically : the g - and - h distribution . in d. hoaglin , f. mosteller , & j. tukey ( eds . ) , exploring data tables trends and shapes ( pp .
461 - 515 ) .
new york , ny : wiley .
|
the general philosophy for bootstrap or permutation methods for testing hypotheses is to simulate the variation of the test statistic by generating the sampling distribution which assumes both that the null hypothesis is true , and that the data in the sample is somehow representative of the population .
this philosophy is inapplicable for testing hypotheses for a single parameter like the population mean , since the two assumptions are contradictory ( e.g. , how can we assume both that the mean of the population is @xmath0 and that the individuals in the sample with a mean @xmath1 are representative of the population ? ) .
the mirror bootstrap resolves that conundrum .
the philosophy of the mirror bootstrap method for testing hypotheses regarding one population parameter is that we assume both that the null hypothesis is true , and that the individuals in our sample are as representative as they could be without assuming more extreme cases than observed .
for example , the mirror bootstrap method for testing hypotheses of one mean uses a generated symmetric distribution constructed by reflecting the original sample around the hypothesized population mean @xmath2 .
simulations of the performance of the mirror bootstrap for testing hypotheses of one mean show that , while the method is slightly on the conservative side for very small samples , its validity and power quickly approach that of the widely used t - test .
the philosophy of the mirror bootstrap is sufficiently general to be adapted for testing hypotheses about other parameters ; this exploration is left for future research .
|
bl lac objects are a subclass of active galactic nuclei ( agn ) showing a strong , non thermal variable emission from radio to tev energies .
these properties are usually ascribed to the relativistic jet emission that is closely aligned with the line of sight @xcite .
four decades of studies of the host galaxies and of the bl lacs close environment have lead to a general consensus that they are mainly hosted by luminous elliptical galaxies embedded in small clusters or group of galaxies ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* see also @xcite for a recent review ) .
most of these studies are , however , based on photometric data only , and evidence for a rich environment is inferred from the increase number counts of sources in the proximity of the bl lacs . only for a handful of objects the physical association of nearby sources has been further confirmed via dedicated spectroscopy ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) . in this paper
we investigate the close galactic environment of 2155 , often considered an archetypal of x ray selected bl lac objects @xcite .
it is one of the most luminous , non transient , extragalactic source known and it shows a rapid and strong variability in the whole electromagnetic spectrum ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
the host of 2155 is a luminous elliptical galaxy with m@xmath9=-24.4 @xcite at redshift z=0.116 ( @xcite ; @xcite ) .
optical imaging of the field of 2155 revealed the presence of a moderated overdensity of galaxies @xcite .
this finding is corroborated by the detection in high resolution nir images of 5 galaxies located within @xmath520 ( @xmath540kpc at z=0.116 ) from 2155 @xcite and by spectroscopic observations of some of the sources @xcite .
we here present a comprehensive study of the field of 2155 using new spectra gathered with the magellan telescope and archival data from @xcite .
this unique dataset , including @xmath1034% of all the galaxies brighter than r=20mag located within 4 from the bl lac ( see sec.[sec : res ] ) , will allow us to probe the galactic environment with unprecedent details . throughout this paper
we assume a concordance cosmology with h@xmath11=70kms@xmath2mpc@xmath2 , @xmath12=0.3 , and . in this cosmology , at z=0.116 an angular scale of @xmath13@xmath141 corresponds to a proper transverse separation of @[email protected] .
all the quoted magnitudes are expressed in the ab standard photometric system @xcite .
in this section we detail the selection of the targets and we present the multi object and long slit spectroscopic observations performed with the inamori
magellan areal camera & spectrograph ( imacs , * ? ? ?
* ) mounted on the 6.5 m magellan telescope baade ( las campanas observatory , chile ) . the data reduction process and the analysis of the spectra is also described . eventually , we present a summary of the archival photometric and spectroscopic data from @xcite used in this work . targets for the multi object spectroscopy were designated from the analysis of the broad v band image collected by @xcite with the eso multi
mode instrument ( emmi , * ? ? ?
* ) on the eso new technology telescope ( ntt , fig.[fig : ntt ] ) .
the size of this image is 75@xmath675 .
this roughly corresponds to 950@xmath6950kpc@xmath16 at z=0.116 , and well fits inside the f/4 camera field of view of imacs ( 154@xmath6154 ) .
data were recalibrated by cross matching the sources present in the field with the nomad catalogue @xcite , reaching an uncertainty on the photometric zero point of @xmath50.1mag . in order to achieve astrometric accuracy better than 05 on the whole frame ( required for mask alignment )
, the astrometric solution was refined with the ` astrometry.net ` software @xcite . during the imaging observations the seeing was 16 and the reached 5@xmath17 detection limit ( estimated from the rms of the sky counts integrated within a seeing radius ) was v@[email protected] .
we used ` sextractor ` @xcite to identify and classify sources .
an object was considered as possible target for spectroscopy if its ` sextractor ` star / galaxy classifier ( the ` class_star ` parameter ) was smaller than 0.5 ( i.e. , more likely to be an extended source ) and it was brighter than v@xmath1421mag . in figure
[ fig : cs ] , we plot the ` class_star ` classifier as a function of the v band magnitude .
the adopted cut in magnitude allows us to avoid faint sources that ` sextractor ` may fail to classify due their low signal to noise in the image .
the conservative limit for the ` class_star ` parameter was chosen to remove the most obvious stars , yet to not miss possible compact galaxies . however , we observe that the majority of the selected sources are in the ` class_star`@xmath50 regime .
the imacs mask generation software ( ` maskgen ` ) requires a priority flag to solve conflict between possible overlapping spectra and to maximise the number of sources in the mask .
we thus arbitrarily assigned priority 1 ( high probability to be included in the mask ) to objects with [email protected] and ellipticity @[email protected] , priority 2 to those sources with either [email protected] or @[email protected] but not both , and priority 3 to the remaining sources ( see fig.[fig : cs ] ) .
this corresponds to prioritise for bright elliptical galaxies , that are expected to be the dominant population of a putative low redshift cluster .
in addition to this list of targets we considered also the four galaxies g1 , g2 , g5 , and g6 ( following the labels assigned by @xcite and @xcite ) observed in high resolution nir images by @xcite .
these sources are positioned at less than @xmath520 from 2155 and we assigned them a priority of 1 . eventually , sources from the sample of @xcite located outside the ntt image field of view were included with priority of 3 . to avoid the gaps between the 8 chips of the ccd mosaic ( 77@xmath638 each )
, we centred our frame 15 north and 15 west from the position of 2155 and the mask position angle was set to 270 .
sixteen apertures of 1@xmath612were cut into the mask , corresponding to the targets listed in table [ tab : sample ] and showed in figure[fig : ntt ] .
in addition , nine reference stars were used for the mask alignment ( see fig.[fig : ntt ] ) . for details ) .
filled circles are objects spectroscopically observed with imacs . ]
multi object spectroscopy data were gathered on october 5th , 2013 using the imacs grating 300-l . by selecting a central wavelength of 6000
we were able to almost continuously cover the spectral region from @xmath53500 to @xmath59000 ( with small variations due to the different position of the sources ) .
three contiguous exposures of 1200s each were collected in clear sky conditions and with an average seeing of @xmath507 .
the ` cosmos ` package was employed for the data reduction .
flux calibration was performed using a spectrum of the standard star feige 110 acquired during the same night and rescaling the spectra to the v band magnitude of the targets ( see * ? ? ?
* for further details ) .
galactic extinction was accounted for according to the map of dust reddening from ( * ? ? ?
* i.e. , e(b - v)@xmath140.048 ) .
the reduced spectra typically have a signal to noise ratio s / n@xmath223 per pixel at @xmath23=6000 and are showed in figure [ fig : mos ] in the online edition of the journal . to determine the source redshifts we followed a three step procedure .
first we find the redshift ( in step of @xmath24z=0.1 ) that best match each spectrum with the @xcite galaxy templates rescaled to the magnitude of the target .
then the redshift was refined measuring the position of prominent absorption and emission features and matching them with the line list from the sdss data set ( e.g. * ? ? ?
finally , spectra were independently visually inspected by the authors to verify their redshifts and a quality flag was assigned : @xmath25=1 means reliable redshift from apparent emission and/or absorption features ; and @xmath25=2 means probable redshift . with the exception of pks 2155 - 304_22 , that seems to show the presence of rest
frame balmer lines ) , with [email protected] and ` class_star`@xmath140.5 . ] , a redshift was established for all the sources from the measure of two or more features ( see tab.[tab : sample ] ) .
the uncertainties on the redshifts are typically @[email protected] .
lccccccccl i d & ra & dec & b & v & r & @xmath27 & z & @xmath25 & alt .
i d + & ( j2000 ) & ( j2000 ) & ( mag ) & ( mag ) & ( mag ) & ( arcsec ) & & & + pks 2155 - 304_09 & 21:58:52.38 & -30:13:30.5 & & & & 4.3 & [email protected] & 1 & g1@xmath29 + pks 2155 - 304_10 & 21:58:53.33 & -30:13:44.6 & [email protected] & [email protected] & [email protected] & 21.1 & [email protected] & 1 & g2@xmath29 , 1790@xmath30 + pks 2155 - 304_11 & 21:58:46.16 & -30:10:43.6 & [email protected] & [email protected] & [email protected] & 185.1 & [email protected] & 2 & 2085@xmath30 + pks 2155 - 304_12 & 21:58:54.56 & -30:11:34.6 & [email protected] & [email protected] & [email protected] & 122.0 & [email protected] & 2 & 1760@xmath30 + pks 2155 - 304_13 & 21:58:50.13 & -30:11:47.1 & [email protected] & [email protected] & [email protected] & 108.0 & [email protected] & 1 & 1928@xmath30 + pks 2155 - 304_14 & 21:59:03.20 & -30:12:05.0 & [email protected] & [email protected] & [email protected] & 168.5 & [email protected] & 1 & 1389@xmath30 + pks
2155 - 304_15 & 21:58:48.05 & -30:12:18.3 & [email protected] & [email protected] & [email protected] & 90.3 & [email protected] & 1 & 2158@xmath30 + pks 2155 - 304_17 & 21:58:56.36 & -30:13:07.8 & [email protected] & [email protected] & [email protected] & 60.5 & [email protected] & 1 & 1689@xmath30 + pks 2155 - 304_18 & 21:58:41.59 & -30:15:05.8 & [email protected] & [email protected] & [email protected] & 165.1 & [email protected] & 2 & 2210@xmath30 + pks 2155 - 304_19 & 21:58:37.91 & -30:15:43.5 & [email protected] & [email protected] & [email protected] & 225.9 & [email protected] & 1 & 2364@xmath30 + pks 2155 - 304_20 & 21:59:00.21 & -30:15:56.2 & [email protected] & [email protected] & [email protected] & 178.5 & [email protected] & 1 & 1496@xmath30 + pks 2155 - 304_21 & 21:58:34.53 & -30:16:08.1 & [email protected] & [email protected] & [email protected] & 275.8 & & 2 & 2499@xmath30 + pks 2155 - 304_22 & 21:58:44.87 & -30:16:34.5 & & [email protected] & & 205.1 & [email protected] & 2 & + pks 2155 - 304_26 & 21:59:31.12 & -30:16:15.0 & [email protected] & & [email protected] & 531.3 & [email protected] & 1 & 351@xmath30 + pks 2155 - 304_28 & 21:59:12.70 & -30:11:32.0 & [email protected] & & [email protected] & 292.9 & [email protected] & 1 & 1027@xmath30 + pks 2155 - 304_29 & 21:58:27.60 & -30:14:16.0 & [email protected] & & [email protected] & 320.3 & [email protected] & 2 & 2762@xmath30 + + + = 5500 .
spatial profile of the two spectra ( green line ) .
the contributions of pks 2155 - 304_09 and of the 2155 s nebulosity were deblended performing a simultaneous fit with a gaussian and a lorentzian function , respectively ( blue and purple shaded areas ) .
the result of the fit is also shown ( orange line ) .
residual spectrum . ] in addition to multi
object spectroscopy , we acquired a long slit spectrum of the two closest galaxies to 2155 ( i.e. , g1 and g5 ; * ? ? ?
* ; * ? ? ?
data were gathered during the same night of the multi
object spectroscopy with a seeing of @xmath505 .
two subsequent exposures of 1500s each were acquired with the 07 long slit and the 86 blaze angle of the imacs grating , allowing us to nominally cover the wavelength range between 3650 and 6750 .
the position angle was set to 104 in order to simultaneously collect spectra of both the galaxies and of the bl lac nebulosity .
standard ` iraf ` procedures for optical spectroscopy were adopted in the data reduction .
first , each of the 8 detectors of mosaic ccd camera was bias subtracted and flat fielded independently .
single exposures were then aligned , combined , and cleaned for cosmic rays using the laplacian edge detection algorithm presented by @xcite .
the wavelength calibration was applied taking as reference the spectrum of he+ne+ar arc lamps .
the small angular separation of pks 2155 - 304_09 from 2155 causes its spectrum to be embedded within the 2155 nebulosity ( see fig.[fig : deb ] ) . in order to disentangle the two contributions we make use of our own ` python ` routine .
namely , first the sky emission was removed considering regions free from emission by astrophysical sources . then , at each pixel of the dispersion axis we sliced the spectra along the spatial direction and we fitted the profile with a combination of a gaussian and a lorentzian functions to model the contributions of pks 2155 - 304_09 and of the nebulosity , respectively ( see fig.[fig : deb ] ) .
the area subtended by each of the curves was considered as the flux of the sources at that wavelength .
the resulting spectra were corrected for galactic extinction and the achieved signal to noise ratios per pixel are typically s / n@xmath224 at @xmath23=5500 ( see figures [ fig : lon ] and [ fig : neb ] in the online version of mnras ) .
the redshifts of the two galaxies were inferred using the method described above ( see tab.[tab : samplels ] ) , while for the bl lac object we have first normalised to spectrum with a power
law , and than searched for the presence of significant lines ( i.e. , with an equivalent width greater than ew@xmath31=1.4 , as estimated using the method of * ? ? ?
* ) . given this limit
, the spectrum does not show the presence of any spectral features .
we observe , however , that the redshifted caii @xmath323934,3968 and nad @xmath235892 lines may be tentatively ( @xmath52@xmath17 ) detected if the redshift z=0.116 @xcite is assumed ( see fig.[fig : neb ] in the electronic version of the journal ) .
higher signal to noise ratio spectra are mandatory to confirm the presence of these features .
we point out that these absorptions could also arise from the cool gas in the circum
galactic medium of the galaxy group ( e.g. * ? ? ?
* ; * ? ? ?
* ) . in this case
, the estimates of @xcite and @xcite have to be considered as a lower limit for the redshift of 2155 .
this latter scenario is discouraged by the identification of the bright elliptical host galaxy @xcite that could be used as a _ standard candle _ to determine the redshift ( e.g. * ? ? ?
* ) , and by the observation a strong emission of 2155 in the tev regime @xcite that push the redshift at z@xmath331 due to the interaction of @xmath34ray photons with the lower frequency photons of the extragalactic background light ( see e.g. * ? ? ?
lcccccc i d & ra & dec & @xmath27 & z & @xmath25 & alt .
i d + & ( j2000 ) & ( j2000 ) & ( arcsec ) & & & + pks 2155 - 304_08 & 21:58:52.63 & -30:13:31.4 & 7.3 & [email protected] & 2 & g5@xmath29 + pks 2155 - 304_09 & 21:58:52.38 & -30:13:30.5 & 4.3 & [email protected] & 1 & g1@xmath30 + pks 2155 - 304_neb & 21:58:52.13 & -30:13:29.6 & 2.6 & & &
+ + + to investigate the environment of 2155 we also take advantage of the study of @xcite aimed to identify galaxies within @xmath51mpc from ultraviolet bright quasars . in summary : a field of 22@xmath622 ( i.e. , [email protected]@xmath16 at z=0.116 ) centred on 2155 was imaged in b and r band with the swope 40 telescope in photometric conditions reaching magnitude limits of b@[email protected] and r@[email protected] .
in addition , multi
object spectroscopy of 160 spatially extended sources brighter than r=20mag was collected using the wfccd instrument on the dupont 100 telescope ( see * ? ? ?
* for further details ) .
we observe that , in the overlapping region between the ntt v band image and the b and r band images from @xcite , there is a discrepancy on the photometric classification of the sources : employing their cut on the ` sextractor ` star / galaxy classifier ( i.e. , ` class_star`@xmath200.98 ) , @xmath530% of the sources we photometrically selected for our multi
object spectroscopic observations ( see sec.[sec : mos ] ) are not recognised as `` galaxy '' by @xcite .
this is most probably due to the different atmospheric conditions and to subtle difference in the parameters ingested to ` sextractor ` to detect sources in the images .
we obtained spectra of four sources that in @xcite have ` class_star`@xmath350.98 ( i.e. , 2155_12 , 2155_13 , 2155_17 , and 2155_21 ) , and three of those turn out to be spectroscopically confirmed galaxies . in order to estimate the overdensity of sources around 2155 ( see sec.[sec : over ] ) we thus decided to slightly relax the constrain on ` class_star ` in @xcite .
in the following we will consider as _ galaxies _ ( based only on the photometric information ) objects present in the @xcite sample that have ` class_star`@xmath201 .
this will allow us to recover the missing fraction of galaxies , and , at the same time , to take advantage of the wider sky region covered by the b and r band images .
using imacs multi object and long slit spectroscopy we obtained the redshifts of 17 targets in the field of 2155 . with the exception of 2155_22 , that possibly presents rest frame balmer lines , all the sources are located in the 0.106@[email protected] redshift range ( see tables [ tab : sample ] and [ tab : samplels ] ) . combining these new data with the sample of @xcite
we reach a completeness in spectroscopy of @xmath534% ( @xmath525% ) for galaxies brighter than r=20mag lying within 240 ( 600 ) from the bl lac ( see fig.[fig : cm ] ) .
we stress that , among the sources observed with multi
object spectroscopy , only 2155_09 was chosen with _ a priori _ knowledge of the redshift .
these two samples seem thus well suited for an unbiased study of the galactic environment of 2155 . in the following
, we will first investigate the environment of 2155 using the photometric information only ( sec.[sec : over ] ) .
afterward , we will include the spectroscopic information in our analysis ( sec.[sec : overspec ] ) . from the bl lac object .
the best fit of coma cluster red sequence from @xcite shifted at the redshift of 2155 is also showed ( see text for details ) .
the cross in the bottom right represents typical error bars .
number of galaxies per magnitude bin located within 600 ( i.e. , @xmath51.3mpc at z=0.116 , green histogram ) and 240 ( 500kpc , orange shaded histogram ) from 2155 .
fraction of the sources per magnitude bin for which a redshift is determined spectroscopically .
the colour code is the same of the _ middle panel_. , title="fig : " ] from the bl lac object .
the best fit of coma cluster red sequence from @xcite shifted at the redshift of 2155 is also showed ( see text for details ) .
the cross in the bottom right represents typical error bars .
number of galaxies per magnitude bin located within 600 ( i.e. , @xmath51.3mpc at z=0.116 , green histogram ) and 240 ( 500kpc , orange shaded histogram ) from 2155 .
fraction of the sources per magnitude bin for which a redshift is determined spectroscopically .
the colour code is the same of the _ middle panel_. , title="fig : " ] to verify early suggestions for an overdensity of sources around 2155 @xcite , we first measured the density of galaxies brighter than r=21.08mag in different boxes of 5@xmath65 spread over the field . at the redshift of the bl lac ,
this correspond to cut the luminosity function at m@xmath37 + 4 ( where m@xmath37 is the characteristic luminosity of galaxies derived from * ? ? ? * ) .
we obtain an average number density of sources of : n@xmath38=([email protected])arcmin@xmath39 that is lower than what observed in the box centred on the bl lac object : n@xmath40=([email protected])arcmin@xmath39 , corresponding to a 3.3@xmath17 overdensity . in this estimate , galaxies not present in the bvr images ( i.e. , objects detected only in high resolution nir images by * ? ? ?
* ) were excluded barring the host galaxy of 2155 .
since the bvr images considered in this work cover the @xmath41 break at [email protected] , we can refine the position and the extent of the overdensity using the cluster red sequence method ( crs , * ? ? ?
* ) . in summary
, we took as reference the red sequence of the coma cluster ( data from the ` goldmine ` archive , * ? ? ?
* ) shifted at z=0.116 using the @xcite galaxy templates for k and filter corrections ( see fig.[fig : cm ] ) . to each galaxy
brighter than m@xmath37 + 4 we then assigned a probability to be consistent with the coma red sequence on the basis of their b - r colours and r band magnitudes .
eventually , a luminosity weighted probability density was computed using a fixed kernel smoothing with radius @xmath42=60 . in this map 2155
is embedded in a @xmath104@xmath17 overdensity ( see fig.[fig : ntt ] ) , confirming the evidence that a group of galaxies is associated with the bl lac . in order to compare these results with previous works
, we measured the galaxy bl lac angular cross correlation function ( @xmath43 ) and spatial covariance function ( @xmath44 , * ? ? ? * ) of galaxies brighter that m@xmath37 + 2 .
this cut in the luminosity function was performed to minimise the uncertainties in @xmath44 ( e.g. * ? ? ?
* ) and for consistency with @xcite .
we obtain @xmath43=([email protected])rad@xmath45 and @xmath44=(80@xmath2879)mpc@xmath46 . after correcting for the different cosmologies , these values are smaller but consistent with the clustering result of @xcite .
this suggests that 2155 reside in a small cluster / group of galaxy with an abell richness class @xmath200 . in figure
[ fig : clus ] we present the distribution of objects in the plane redshift angular distance from the luminosity centre of the group member galaxies , that is located @xmath570 south east from 2155 . considering only sources
brighter than m@xmath37 + 4 the decrease of the galaxy density at increasing angular distance is apparent .
the redshift distribution of the sources , in bins of @xmath24z=0.001 , shows a clear peak at z@[email protected]@xmath280.00006 , consistent with previous estimates of the redshift of 2155 , with a nearly gaussian shape . assuming that only the 12 galaxies with a spectroscopic redshift within @xmath281000kms@xmath2 from z@xmath47 ( not including the bl lac host galaxy ) and lying within a radius of 240 ( i.e. 500kpc at [email protected] ) from the luminosity centre are part of the overdensity , the radial component of the velocity dispersion is @xmath48=250@xmath1kms@xmath2 ( where the quoted uncertainties are the 68% confidence level , * ? ? ?
* ) . under the hypothesis that the galaxy distribution traces the underlying mass distribution
, we used the relations in @xcite to estimated the virial radius ( r@[email protected] ) and the virial mass ( m@[email protected]@xmath610@xmath7m@xmath8 ) of the system .
we are cautious to take these estimates as face values .
indeed , for small groups of galaxies , the traditional estimators of the virial radius and mass are not tightly correlated with their real values .
the presence of a second peak at [email protected] indicates that a second overdensity of galaxies may be located @xmath540 north from 2155 . ) .
the position of 2155 is marked with grey dashed lines .
green , purple , and blue squares are targets with spectra collected by @xcite , and by our imacs multi object and long slit spectroscopic campaign , respectively .
galaxies with redshift consistent within 1000kms@xmath2 from 2155 are marked with filled symbols .
redshift distribution of the sources ( green histogram ) .
the inset shows a zoom in of galaxies within @xmath281400kms@xmath2 from 2155 .
objects located closer than 240 ( i.e. @xmath5500kpc at z=0.116 ) from the peak of the overdensity are highlighted in orange .
the best gaussian fit of the distribution is also showed .
angular distance distribution of the sources .
the blue histogram are galaxies brighter that r@xmath521 ( i.e. m@xmath37 + 4 at [email protected] ) .
the green shaded histogram are sources with redshift determined spectroscopically , and the orange filled one show the distribution of the sources within @xmath281000kms@xmath2 from the bl lac . ]
we investigated the properties of the environment of 2155 using broad band images , new multi
object spectroscopy collected with the imacs instrument of the magellan baade telescope , and archival spectra gathered with the dupont 100 telescope by @xcite .
our measurements confirm that 2155 is harboured by a moderate overdensity of galaxies located at [email protected]@xmath280.00006 , with a virial mass of m@[email protected]@xmath610@xmath7m@xmath8 . to our knowledge
, a detailed study of the environment of bl lac objects with multi object spectroscopy was performed for only two other targets : rgb 1745 + 398 ( z=0.267 , * ? ? ?
* ) ; and pks 0447 - 439 ( with a possible redshift z=0.343 , * ? ? ?
both the sources are found to be associated with large galaxy overdensities with virial masses of few times 10@xmath49m@xmath8 .
in addition , the detection with chandra of a diffuse x ray emission , on scales of @xmath10100kpc , around pks 0548322 and pks 2005489 suggests that these two bl lacs are embedded comparably massive galaxy clusters @xcite . on the contrary ,
the potentially presence of gravitational arcs in the hst images of h 1517 + 656 ( z=0.702 , * ? ? ?
* ) allowed @xcite to estimate a virial mass more similar to what we observed for 2155 : m@[email protected]@xmath610@xmath7m@xmath8 .
this ( small ) sample for which the virial mass could be estimated , indicates that bl lac objects are typically hosted by haloes spanning a range of masses , from @xmath510@xmath7m@xmath8 to few times 10@xmath49m@xmath8 .
this suggests that radio loud sources , in particular bl lac objects , may be found in a variety ( although still massive ) galactic environments ( see also * ? ? ?
* ; * ? ? ?
although based on a small and heterogeneous sample , it is interesting to put this result in the contest of the current merger driven paradigm of the nuclear activity ( e.g. * ? ? ?
* ; * ? ? ?
in small groups , major mergers between gas rich galaxies are expected to be frequent and particularly effective in funnelling huge amount of gas in the nuclear regions , eventually triggering the central black holes activity ( e.g. * ? ? ?
conversely , in rich clusters , the high velocity dispersion decreases the effective cross section of galaxy interactions , in spite of the higher galaxy densities ( e.g. * ? ? ?
galaxy harassment due to high speed interactions ( but not merger ) of galaxies could drive dynamical instabilities that efficiently channel gas onto the super massive black holes in rich environment @xcite .
this requires the host galaxy to contains huge gas reservoir
. however , ram
pressure stripping is likely to remove the cold gas in cluster galaxies , and the bl lac hosts are found to be deficient in molecular gas when compared with quasars ( e.g. * ? ? ?
the trigger mechanism(s ) for the nuclear activity of the bl lac host galaxies is thus likely to be complex and not universal . a larger sample of bl lac with a detailed characterisation of the large scale environment is thus needed to understand this process in a statistical manner .
epf acknowledges funding through the erc grant ` cosmic dawn ' .
mf acknowledges support by the science and technology facilities council , grant number st / l00075x/1 .
this paper includes data gathered with the 6.5meter magellan telescopes located at las campanas observatory ( chile ) .
we thank j. omeara for assisting during observations .
epf is grateful to t. schmidt , e. baados , a. obreja , m. fouesneau , and c. ferkinhoff for providing support in the use of ` python ` for the data reduction .
the authors thank a. treves and r. falomo for their important contributions to data interpretation and to the shaping of the article .
the authors acknowledge c. mazzucchelli and m. landoni for useful comments and suggestions .
this research made use of ` astropy ` , a community developed core python package for astronomy @xcite and of ` aplpy ` , an open - source plotting package for ` python ` .
99 aarseth s. j. , fall s. m. , 1980 , apj , 236 , 43 aharonian f. , et al . , 2005 , a&a , 430 , 865 aharonian f. , et al . , 2007 , apj , 664 , l71 aharonian f. , essey w. , kusenko a. , prosekin a. , 2013 , phrvd , 87 , 063002 astropy collaboration , et al . , 2013 , a&a , 558 , a33 bahcall n. a. , 1981 , apj , 247 , 787 barnes j. e. , hernquist l. e. , 1991 , apj , 370 , l65 beckmann v. , bade n. , wucknitz o. , 1999 , a&a , 352 , 395 bertin e. , arnouts s. , 1996 , a&as , 117 , 393 blades j. c. , hunstead r. w. , murdoch h. s. , 1981 , mnras , 194 , 669 blandford r. d. , rees m. j. , 1978 , bllo.conf , 328 blanton m. r. , et al .
, 2001 , aj , 121 , 2358 boksenberg a. , sargent w. l. w. , 1978 , apj , 220 , 42 chadwick p. m. , et al . , 1999 , aph , 11 , 145 chadwick p. m. , et al . , 1999 , apj , 513 , 161 decarli r. , labita m. , treves a. , falomo r. , 2008 , mnras , 387 , 1237 danese l. , de zotti g. , di tullio g. , 1980 , a&a , 82 , 322 dekker h. , delabre b. , dodorico s. , 1986 , spie , 627 , 339 di matteo t. , springel v. , hernquist l. , 2005 , natur , 433 , 604 donato d. , gliozzi m. , sambruna r. m. , pesce j. e. , 2003 , a&a , 407 , 503 dressler a. , et al .
, 2011 , pasp , 123 , 288 falomo r. , melnick j. , tanzi e. g. , 1990 , natur , 345 , 692 falomo r. , giraud e. , melnick j. , maraschi l. , tanzi e. g. , treves a. , 1991 , apj , 380 , l67 falomo r. , pesce j. e. , treves a. , 1993 , apj , 411 , l63 falomo r. , pesce j. e. , treves a. , 1995 , apj , 438 , l9 falomo r. , 1996 , mnras , 283 , 241 falomo r. , kotilainen j. k. , 1999 , a&a , 352 , 85 falomo r. , pian e. , treves a. , 2014 , a&arv , 22 , 73 fan j. h. , lin r. g. , 2000 , a&a , 355 , 880 fumagalli m. , dessauges - zavadsky m. , furniss a. , prochaska j. x. , williams d. a. , kaplan k. , hogan m. , 2012 , mnras , 424 , 2276 gavazzi g. , boselli a. , donati a. , franzetti p. , scodeggio m. , 2003 , a&a , 400 , 451 girardi m. , giuricin g. , mardirossian f. , mezzetti m. , boschin w. , 1998 , apj , 505 , 74 gladders m. d. , yee h. k. c. , 2000 , aj , 120 , 2148 griffiths r. e. , briel u. , chaisson l. , tapia s. , 1979 , apj , 234 , 810 hopkins p. f. , hernquist l. , cox t. j. , kere d. , 2008 , apjs , 175 , 356 kelson d. d. , 2003 , pasp , 115 , 688 kotilainen j. k. , falomo r. , scarpa r. , 1998 , a&a , 336 , 479 kotilainen j. k. , hyvnen t. , falomo r. , treves a. , uslenghi m. , 2011 , a&a , 534 , l2 kinney a. l. , calzetti d. , bohlin r. c. , mcquade k. , storchi - bergmann t. , schmitt h. r. , 1996 , apj , 467 , 38 lang d. , hogg d. w. , mierle k. , blanton m. , roweis s. , 2010 , aj , 139 , 1782 lietzen h. , nilsson k. , takalo l. o. , heinmki p. , nurmi p. , keinnen p. , wagner s. , 2008 , a&a , 482 , 771 lietzen h. , heinmki p. , nurmi p. , liivamgi l. j. , saar e. , tago e. , takalo l. o. , einasto m. , 2011 , a&a , 535 , a21 liuzzo e. , et al . , 2013 , aj , 145 , 73 longair m. s. , seldner m. , 1979 , mnras , 189 , 433 lpez - cruz o. , barkhouse w. a. , yee h. k. c. , 2004 , apj , 614 , 679 mandelbaum r. , li c. , kauffmann g. , white s. d. m. , 2009 , mnras , 393 , 377 moore b. , katz n. , lake g. , dressler a. , oemler a. , 1996 , natur , 379 , 613 muriel h. , donzelli c. , rovero a. c. , pichel a. , 2015 , a&a , 574 , a101 nilsson k. , pursimo t. , sillanp a. , takalo l. o. , lindfors e. , 2008 , a&a , 487 , l29 oh k. , sarzi m. , schawinski k. , yi s. k. , 2011 , apjs , 195 , 13 oke j. b. , 1974 , apjs , 27 , 21 oke j. b. , gunn j. e. , 1983 , apj , 266 , 713 pesce j. e. , falomo r. , treves a. , 1994 , aj , 107 , 494 pesce j. e. , falomo r. , treves a. , 1995 , aj , 110 , 1554 prochaska j. x. , weiner b. , chen h .- w . , cooksey k. l. , mulchaey j. s. , 2011 , apjs , 193 , 28 richter p. , krause f. , fechner c. , charlton j. c. , murphy m. t. , 2011 , a&a , 528 , a12 rieger f. m. , volpe f. , 2010 , a&a , 520 , a23 sandrinelli a. , covino s. , treves a. , 2014a , a&a , 562 , aa79 sandrinelli a. , covino s. , treves a. , 2014b , apj , 793 , ll1 sbarufatti b. , treves a. , falomo r. , heidt j. , kotilainen j. , scarpa r. , 2005 , aj , 129 , 559 sbarufatti b. , treves a. , falomo r. , 2005 , apj , 635 , 173 sbarufatti b. , falomo r. , treves a. , kotilainen j. , 2006 , a&a , 457 , 35 scarpa r. , urry c. m. , falomo r. , pesce j. e. , webster r. , odowd m. , treves a. , 1999 , apj , 521 , 134 scarpa r. , urry c. m. , falomo r. , pesce j. e. , treves a. , 2000 , apj , 532 , 740 schlafly e. f. , et al . , 2014 , apj , 789 , 15 schwartz d. a. , griffiths r. e. , schwarz j. , doxsey r. e. , johnston m. d. , 1979 , apj , 229 , l53 sembay s. , warwick r. s. , urry c. m. , sokoloski j. , george i. m. , makino f. , ohashi t. , tashiro m. , 1993 , apj , 404 , 112 smith p. s. , hall p. b. , allen r. g. , sitko m. l. , 1992 , apj , 400 , 115 smith e. p. , odea c. p. , baum s. a. , 1995 , apj , 441 , 113 tody d. , 1986 , spie , 627 , 733 tody d. , 1993 , aspc , 52 , 173 urry c. m. , et al . , 1993 , apj , 411 , 614 urry c. m. , scarpa r. , odowd m. , falomo r. , pesce j. e. , treves a. , 2000 , apj , 532 , 816 van dokkum p. g. , 2001 , pasp , 113 , 1420 weistrop d. , smith b. a. , reitsema h. j. , 1979 , apj , 233 , 504 worpel h. , brown m. j. i. , jones d. h. , floyd d. j. e. , beutler f. , 2013 , apj , 772 , 64 wurtz r. , ellingson e. , stocke j. t. , yee h. k. c. , 1993 , aj , 106 , 869 wurtz r. , stocke j. t. , ellingson e. , yee h. k. c. , 1997 , apj , 480 , 547 yee h. k. c. , lpez - cruz o. , 1999 , aj , 117 , 1985 zacharias n. , monet d. g. , levine s. e. , urban s. e. , gaume r. , wycoff g. l. , 2004 , aas , 36 , 1418 zacharias n. , monet d. g. , levine s. e. , urban s. e. , gaume r. , wycoff g. l. , 2005 , ycat , 1297 , zhang b .- k . ,
zhao x .- y . , wang c .- x . ,
dai b .- z . , 2014 , raa , 14 , 933
spectra of the sources collected through multy object and long slit spectroscopy are available in the online version of this article .
|
2155 is one of the brightest extragalactic source in the x ray and euv bands , and is a prototype for the bl lac class of objects . in this paper
we investigate the large scale environment of this source using new multi object as well as long slit spectroscopy , together with archival spectra and optical images .
we find clear evidence of a modest overdensity of galaxies at @xmath0 , consistent with previous determinations of the bl lac redshift .
the galaxy group has a radial velocity dispersion of 250@xmath1kms@xmath2 and a virial radius of @xmath3mpc , yielding a role of thumb estimate of the virial mass of m@[email protected]@xmath610@xmath7m@xmath8 , i.e. , one order of magnitude less than what observed in other similar objects .
this result hints toward a relatively wide diversity in the environmental properties of bl lac objects .
[ firstpage ] ;
|
among many cosmological surveys , the submillimetre ( submm ) survey is very unique , in the sense that the expected flux density of sources is almost insensitive to redshift for @xmath11 1 8 , owing to the strong negative @xmath1-correction ( e.g. * ? ? ?
although the current sensitivity allows us to detect only the brightest infrared galaxies in the universe , it is possible to detect massive starbursts and gas rich qsos at extreme redshifts @xmath12 if exist ( e.g. * ? ? ?
* ) . however , most of the submm galaxies ( smgs ) currently identified lie at @xmath13 .
this is not because of the detection limit as noted above , but because of the ` identification limit ' , owing to a large beam size of current ( sub)mm telescopes used for surveys .
the radio emission provides a high - resolution substitute for the infrared emission observed in the submm ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) , and hence the detection limit in the radio has set the upper limits on redshifts of identified smgs @xcite . the best way to identify optical counterparts of smgs is to have high resolution submm images with interferometers @xcite .
although time consuming , it allows us to perform direct optical / near - infrared ( nir ) identifications from the submm position .
here we adopt another method of direct optical identification using optical / nir colours of galaxies around the submm source .
the expected number of random optical / nir associations within the error circle of the submm position is not negligible ; e.g. there would be @xmath14 @xmath1-band sources ( @xmath15 on average within a @xmath16 radius ( see below ) .
however , if smgs have rather confined optical - nir colours and the surface density of objects with similar colours is reasonably low , it will be possible to directly identify smgs without using costly high - resolution submm images .
as in the radio identification method @xcite , we can calculate the formal significance of each optical / nir association , given the number counts of colour - selected galaxies . to do this
we need a firm basis for the colours of smgs . since the discovery of smgs , extremely red objects ( eros ) usually defined with @xmath17 or @xmath18 and distant red galaxies with @xmath19
have been paid special attention as candidates of smg counterparts ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
it turns out that the fraction of eros or drgs in smgs is not very high and their surface densities are not low enough to reject random associations . among the nir - selected galaxy populations in the literature , bzk - selected star - forming galaxies ( bzks *
* ) may be the most promising counterparts of smgs @xcite , which lie at @xmath20 and include heavily obscured galaxies , such as smgs , as an extreme subset . in this paper , we indeed find that almost all of the @xmath1-faint ( indicating higher redshifts ) radio - detected smgs are bzks .
we then use bzks as a key galaxy population to identify radio - undetected smgs at @xmath20 . since the bzk - selection is based on observed wavelengths covering a spectral break at 4000
, we can naturally extend this selection technique to the higher redshift range , e.g. @xmath21 , in the future by using a combination of wavebands at longer wavelengths .
we also emphasize that the understanding of the physical relation between smgs and bzks would be important to reveal the evolution of galaxies at @xmath0 , given that they have similar stellar masses of @xmath22m@xmath23 , and spatial correlation lengths of @xmath10@xmath24mpc @xcite .
we discuss the hypothesis of smgs being merging bzks .
we adopt the sample of smgs from the scuba half degree extragalactic survey ( shades * ? ? ?
* ; * ? ? ?
* ) as described in section 2 . in section 3 ,
we investigate typical optical / nir colours of the radio - detected smgs . we apply our direct identification method for radio - undetected smgs in section 4 .
we then discuss a possible evolutionary link between smgs and bzks in section 5 .
finally we give our summary in section 6 . throughout this paper ,
we assume a cosmology with @xmath25 , @xmath26 and @xmath27@xmath28km@xmath28sec@xmath29@xmath28mpc@xmath29 .
all magnitudes in this paper use the ab system unless otherwise noted . [ cols="<,<,<,^,^,^,^,^,^,^,^,^,^ " , ] note bold figures indicate the adopted @xmath30 .
@xmath31 angular distance from shades centroid .
@xmath32 @xmath6 values for bzk galaxies .
no entry indicates that the galaxy does not satisfy bzk colour criteria .
@xmath33 same as @xmath32 but for the bzk - ero overlapping population .
@xmath34 same as @xmath32 but for the bzk - ero - drg overlapping population .
@xmath35 two robust identifications for one smg .
@xmath36 tentative identification , because of @xmath37 ( sxdf850.91 ) or the presence of another @xmath1-bright association ( sxdf850.93 ) .
figure [ pval ] shows the histogram of @xmath30 values for bzks around the radio - undetected smgs in the sxdf and also for the radio identifications .
the radio identifications have a systematically lower @xmath30 than bzks around the radio - undetected smgs , owing to the pre - selection by radio with @xmath38 .
the dashed lines in figure [ pval ] indicate the expected number of chance associations as a function of @xmath6 .
the normalization of these curves is given by the number of searched regions , i.e. 36 for the shades sources with radio counterparts and 24 for those without radio counterparts . in table 3
, we tabulate the derived @xmath6 for bzks around the radio - undetected shades sources with @xmath39 .
the robust identifications are indicated by solid squares in the colour - colour diagrams ( figures [ bzk ] and [ rjk ] ) . from the @xmath6-statistics of nir galaxies
, we find 6 robust ( including one double identification ) and 2 tentative identifications .
thus , the success rate of identification is not very high , only @xmath40%
. this might be reasonable if we consider the possible bias in radio - undetected smgs toward optically faint counterparts at high redshifts .
all of the robust nir identifications but one are @xmath41/bzks or @xmath42/bzks , owing to lower surface densities than that of bare bzks .
this indicates that the current positional uncertainty is too large to use bare bzks for a counterpart population .
therefore , it is practical to use @xmath41/bzks and/or @xmath42/bzks , although the completeness of identification would decrease by a factor of 2 .
this is still an encouraging performance for future exploitation .
we can extend this technique to higher redshifts , using a combination of photometric bands at longer wavelengths , such as @xmath43 ( see * ? ? ?
* ) , @xmath44 and so on
. this technique will be useful to push the current ` identification limit ' of smgs to @xmath21 .
bzk - selected radio - undetected smgs would have a lower dust temperature than radio detected ones .
although the sample suffers from various selection effects , due to optical - nir colour cuts , we could put some constraints on the number of smgs with cool dust .
we identified 5 radio - undetected smgs as bzks . on the other hand
, we found that 13 radio - detected smgs ( singly associated ones ) are bzks .
thus , we estimate that the fraction of radio - undetected smgs at @xmath45 would be @xmath46% .
this is a lower limit , since we can identify only smgs with extremely red colours such as @xmath41/bzks and @xmath42/bzks .
we also caution that our sample is limited to those with @xmath1-band magnitudes of @xmath47 , and missing extremely faint smgs .
we summarise the fraction of smgs in each nir galaxy population by using the observed surface density .
the effective area of shades in the sxdf is @xmath48arcmin@xmath49 @xcite .
considering both the radio and nir identifications , we found 18(@xmath50 ) , 12(@xmath51 ) , and 12(@xmath52 ) singly associated smgs ( plus possible candidates from multiple identifications ) are bzks , eros , and drgs , respectively / bzk . ] .
these numbers give surface densities of smgs with bzk , ero , drg colours of 0.07@xmath53 , @xmath54 , and @xmath54arcmin@xmath55 , respectively . using the surface density of each galaxy population in table 2 ,
we derive the fraction of smgs in bzks , eros , and drgs , to be about 1 per cent .
these are the current best estimates of the fraction of smgs in these galaxy populations .
nine ( 6 radio and 3 nir identified ) singly associated smgs are found to be @xmath42/bzks , giving a surface density of @xmath560.04arcmin@xmath55 .
the surface density of @xmath42/bzks would be around 0.25 0.7arcmin@xmath55 , considering the results by takagi et al .
( 2007 ) and lane et al .
thus , @xmath5610% of @xmath42/bzks are likely to be smgs , and therefore @xmath42/bzks contain a much higher fraction of smgs than other optical / nir - selected galaxy populations .
this confirms the results by takagi et al .
( 2007 ) .
given the assumed cosmology , a cosmic ( co - moving ) volume we consider is 9@xmath57mpc@xmath58 for the redshift range of @xmath59 and the survey area of @xmath48arcmin@xmath49 .
we find 18 bzk - selected smgs in this volume , corresponding to a space density of @xmath60mpc@xmath61 .
we discuss a possible evolutionary link between smgs and bzks and its implications for physical properties of smgs , by using the star formation rate ( sfr)-stellar mass ( @xmath62 ) relation of bzks . in the following ,
we discuss only smgs satisfying the bzk selection criterion and @xmath63 ( vega ) , in order to securely derive stellar masses .
we find seven such smgs ( singly associated ones ) from our sample . in order to increase the sample
, we also use the sample of radio - detected smgs in @xcite , which have spectroscopic redshifts and @xmath1-band photometry @xcite .
we find seven smgs in their sample , which have a starburst - dominated optical spectrum , a redshift at @xmath64 ( i.e. the redshift range of bzks ) , and @xmath63 ( vega ) .
we refer to this sample and the sample from the sxdf as the chapman sample and the sxdf sample , respectively .
the starburst activity of smgs seems to be induced by galaxy interactions / mergers as indicated by their morphology and dynamical properties @xcite .
immediate progenitors of smgs would be gaseous star - forming galaxies as massive as smgs .
most plausible candidates for this parent population would be star - forming bzks , given a similar stellar mass and coevality with smgs .
although the sfrs of bzks are sometimes as high as those of local ulirgs @xcite , the kinematical properties of bzks are more similar to quiescent disc galaxies @xcite .
this quiescent nature of star formation in bzks is also supported by the measurement of the star formation efficiency from the co luminosity , which is an order of magnitude lower than that of smgs and similar to local spirals @xcite . in the following discussion , we assume that some dynamical perturbations occurring in bzks induce the vigorous star formation of smgs .
we derive the stellar masses of smgs , using the empirical formulae by daddi et al .
( 2004 , their eq . 6 and 7 for the sed - fitting derived mass ) : @xmath65 \end{aligned}\ ] ] where @xmath66 ( vega ) is the @xmath1-band magnitude corresponding on average to a mass of @xmath67m@xmath23 .
since these formulae are calibrated for @xmath63 ( vega ) , we restrict our sample of smgs to those in this magnitude range as noted above .
the correction term with @xmath68 colour is applied only for the sxdf sample , since @xmath69-band photometry is not available for the chapman sample in the literature .
the stellar masses derived from these formulae are based on the salpeter initial mass function extending between 0.1 and 100m@xmath23 . in deriving the stellar masses ,
we adopt petrosian magnitudes from the ukidss catalogue for estimation of total @xmath1-band magnitudes . for the chapman sample ,
we adopt the total @xmath1-band magnitudes in smail et al .
( 2004 ) , which are estimated with the aperture photometry of a large ( 4@xmath70 ) diameter .
we find that the mean stellar mass of 14 smgs ( 7 from sxdf and the other 7 from the chapman sample ) is 1.2@xmath71m@xmath23 .
if we included smgs with @xmath72 ( vega ) as well , the mean stellar mass would reduce about 20% .
the uncertainty in the stellar mass on single objects is about 60% ( daddi et al .
however , we caution that the uncertainty may be much larger than this , since smgs would have a large fraction of mass in substantially obscured stellar populations . for example , @xcite derive on average 5 times higher stellar masses from rest - frame @xmath1-band magnitudes than those reported from modeling the uv / optical seds of smgs in smail et al .
( 2004 ) .
the sfrs are derived from radio 1.4ghz fluxes and redshifts .
we estimate the fir luminosities from the radio - infrared luminosity relation , @xmath73 found for smgs @xcite , where @xmath74 and @xmath75thz .
the radio luminosities at the rest - frame 1.4ghz are calculated with @xmath76 where @xmath77 and @xmath78 are the luminosity distance and the 1.4ghz flux density , and @xmath79 is the spectral index ( @xmath80 ) .
we adopt @xmath81 . for the sxdf sample ,
we assume a redshift of @xmath82 , an average redshift of bzks @xcite .
for the chapman sample , we use the spectroscopic redshifts from chapman et al .
we convert the derived fir luminosity to the sfr using a relation sfr = @xmath83l@xmath23 @xcite . for the combined sample of smgs
, we derive an average sfr of 680m@xmath23yr@xmath29 .
this becomes 600m@xmath23yr@xmath29 if we include the smgs with @xmath72 ( vega ) . if we adopt the widely - used local radio - fir relation by , instead of that for smgs by @xcite , the derived sfrs increases by a factor of @xmath8 .
local ulirgs are also considered for comparison .
we adopt the average of the sample in @xcite .
we use the dynamical mass , instead of the stellar mass , which gives a solid upper limit on the stellar mass .
the averages of the dynamical masses and the sfrs are 1.3@xmath71m@xmath23 and 340m@xmath23yr@xmath29 , respectively .
we now consider how much sfrs are enhanced in smgs , owing to dynamical perturbations .
such an enhancement may be estimated from a comparison of the average sfr of smgs with that of bzks , a suspected progenitor population of smgs , with a similar stellar mass .
we note that the increase in the stellar mass would be less than a factor of @xmath84 in merging of gas - righ galaxies to produce still gaseous smgs , in which we could neglect the increase in stellar mass after merging .
the average sfr of progenitor galaxies may be estimated from the correlation between sfrs and stellar masses found for bzks or local star - forming galaxies , once we derive the stellar mass of smgs or local ulirgs . in figure
[ mass ] , we plot the sfrs and stellar masses of smgs and local ulirgs . @xcite present the correlation between the sfr and the stellar mass of bzks , sfr = 200@xmath85 where @xmath86m@xmath23 , which is depicted as the solid line in the figure . also , the same correlation but for local blue galaxies found in the sloan digital sky survey
sfr = 8.7@xmath87 is shown as the dashed line . comparing the average sfr and @xmath62 of smgs to the sfr-@xmath62 relation of bzks
, we estimate that a typical enhancement of sfr experienced by smgs is around a factor of 3 indicated as the solid arrow in figure [ mass ] .
this enhancement is rather moderate as if smgs follow the same sfr-@xmath62 relation . in the local universe ,
such a moderate enhancement is , for example , found for m82 ( figure 18 in elbaz et al .
2007 ) , whose starburst activity is likely to be induced by interactions with m81 . compared to local ulirgs , this enhancement is less significant by an order of magnitude .
we estimate the sfr enhancement of local ulirgs to be a factor of @xmath88 ( the dotted arrow ) , comparing the mean sfr and @xmath62 of local ulirgs with the sfr-@xmath62 relation of local star - forming galaxies .
this is a lower limit , since we use the dynamical mass for local ulirgs .
if progenitors of smgs are bzks with a stellar mass comparable to smgs , then the sfr enhancement could _ not _ be as high as local ulirgs .
as @xcite indicated , smgs are probably experiencing maximum starbursts @xcite , determined roughly by the dynamical time scale of galaxies and negative feedback due to super novae .
the maximum sfr expected for smgs would be around 600m@xmath23yr@xmath29 @xcite , which is already comparable to the observed sfrs .
what are implications of a moderate ( @xmath89 ) sfr enhancement of smgs ? in the local universe , such a moderate enhancement is found for typical interacting pairs of galaxies ( not necessarily in the most luminous merging phase ) with a projected distance of @xmath90 kpc @xcite .
the numerical simulation by also indicates that a typical enhancement of sfr during the major merger phase is only a factor of @xmath91 .
if this moderate enhancement is a typical case at @xmath0 as well , the fraction of smgs in star - forming bzks could be as high as the merger fraction .
although the merger fraction at @xmath92 is quite uncertain , it would be at least similar to or higher than the value at @xmath9 of @xmath10% ( e.g. * ? ? ?
* ; * ? ? ?
* ) , because of the hierarchical clustering properties of the galaxy formation process .
this argument leaves us an open question why the fraction of smgs in bzks is not @xmath93 per cent , but only @xmath94 per cent . a bias of the submm surveys , which tend to miss galaxies as luminous as smgs but with higher dust temperatures , should play a role here , but the effect would be only a factor of @xmath95 @xcite . also we are missing some fraction of radio - undetected , and hence cool - dust smgs , which are not extremely red ( see section 4.2 ) . even if we consider luminous dusty galaxies at @xmath0 with both higher and lower dust temperatures , a correction for the incompleteness would be less than an order of magnitude , and therefore not enough to reconcile the apparent inconsistency .
we could explain this apparent inconsistency between the rareness of smgs and its moderate sfr enhancement as follows .
at @xmath0 , galaxy interactions may be too common , which means that high specific sfrs ( mass - normalized sfrs ) of bzks are a result of galaxy interactions to some extent . in this case
, there would be almost no net enhancement of sfr in typical interacting galaxies over the sfr-@xmath62 relation .
the sfr enhancement of smgs looks moderate , but might be indeed significant as galaxies at @xmath0 , possibly caused only by rare merging events , such as those with some particular orbit parameters , very equal mass ratios , or multiple mergers .
for example , according to the numerical simulation by , high gas concentrations and hence high sfr enhancement can be realized not in direct mergers , but in retrograde mergers .
we should bare in mind that smgs are not the only galaxy population experiencing major mergers .
smgs will merely be a tip of iceberg .
@xcite suggest that uv morphologies of bzks and uv - selected ( bx / bm ) galaxies are not distinguishable from that of smgs .
this indicates either that not only smgs but also bzks and bx / bm galaxies are experiencing merger - induced starbursts , or simply that the uv morphology is not a reliable indicator of a major merger .
on the other hand , @xcite show that smgs are a dynamically distinctive galaxy population with large mass concentration , i.e. dynamically hotter than bzks and bx / bm galaxies .
therefore , the uv morphology may not be very useful to identify dynamically hot galaxies at @xmath96 , as suggested by no obvious correlations between the uv morphology and other galactic properties , such as sfr , outflow and stellar mass @xcite .
we need more direct measures on dynamical properties of galaxies at @xmath0 , in order to identify kin of smgs . for this purpose ,
atacama large millimeter array ( alma * ? ? ?
* ) will play an important role , which is capable of measuring gas dynamics of @xmath7-mjy sources , such as bzks @xcite .
we identify optical - nir counterparts of 28 radio - detected smgs , including multiple associations , in the shades / sxdf , and have compiled optical and near - infrared photometry of these sources .
we then investigate optical / near - infrared colours ( @xmath2 , @xmath97 and @xmath98 ) of radio - detected smgs , along with the additional sample from the hdf , to find that almost all of @xmath1-faint ( @xmath99 ) smgs have @xmath100 , i.e. bzks .
this forms a strong basis of identifying radio - undetected smgs using bzks around submm sources .
we calculate the formal significance ( @xmath6 value ) for individual bzks around radio - undetected smgs , and found 6 robust ( @xmath101 ) identifications , including one double identification .
all of these new identifications but one are extremely red , i.e. @xmath41/bzks or @xmath42/bzks .
it turns out that the current positional uncertainty is too large to use bare bzks for the direct optical identification .
however , this result is encouraging for future exploitation of smgs at @xmath21 using a combination of longer wavebands , such as @xmath43 and @xmath44 . from a comparison of observed surface densities , we find that only @xmath94 per cent of bzks , eros and drgs are smgs , while this fraction is as high as @xmath102% for the bzk - ero - drg overlapping population , @xmath103bzks .
the average sfr of bzk - selected smgs is @xmath56680m@xmath23yr@xmath29 .
if smgs are major mergers of typical disk - like star - forming galaxies at @xmath0 , i.e. bzks , the enhancement of sfr occurring in smgs would be only a factor of @xmath563 .
this enhancement is an order of magnitude lower than that of local ulirgs .
such a moderate enhancement of sfr is found in ordinary galaxy pairs with a separation of @xmath104kpc in the local universe , and not significant at all .
thus , the rareness of smgs at @xmath0 is rather puzzling .
this may be because the sfr of star - forming bzks is already enhanced by galaxy interactions to some extent , and hence the moderate sfr enhancement of smgs over bzks could indeed be significant as a galaxy at @xmath0 . a quantitative explanation on statistical properties of ulirgs near and far would be a challenging task for the theory of galaxy formation , requiring clear understanding of galaxy evolution and the physics of galaxy interactions .
we thank all the members of the shades consortium for their persistent efforts to publish the source catalogue .
we are grateful to k. motohara for providing us the number counts of nir - selected galaxies .
also we are thankful to j. furusawa for providing us photometric redshifts from the sxds .
we wish to thank an anonymous referee for very useful and constructive comments , including the suggestion to add more samples from the literature .
this work is based in part on data obtained as part of the ukirt infrared deep sky survey and the subaru / xmm - newton deep survey .
we thank d. elbaz for providing us compiled spectral energy distributions of arp220 and m82 .
we also thank e. daddi for sending us his data points used in figure 7 .
tt acknowledges the japan society for the promotion of science ( jsps pd fellow , no .
18@xmath1057747 ) .
a. j. , murowinski r. , tarenghi m. , 2006 , in ground - based and airborne telescopes . edited by stepp , larry m .. proceedings of the spie , volume 6267 , pp .
626702 ( 2006 ) .
6267 of presented at the society of photo - optical instrumentation engineers ( spie ) conference , the atacama large millimeter / submillimeter array : overview and status t. , nagashima c. , nakajima y. , nagata t. , sato s. , nakaya h. , yamamuro t. , sugitani k. , tamura m. , 2003 , in instrument design and performance for optical / infrared ground - based telescopes . edited by iye , masanori ; moorwood , alan f. m. proceedings of the spie , volume 4841 , pp . 459 - 464 ( 2003 ) .
sirus : a near infrared simultaneous three - band camera .
pp 459464 s. j. , hambly n. c. , dye s. , almaini o. , cross n. j. g. , edge a. c. , foucaud s. , hewett p. c. , hodgkin s. t. , irwin m. j. , jameson r. f. , lawrence a. , lucas p. w. , adamson a. j. , et al .
2007 , , 375 , 213
|
we investigate the relationship between two massive star - forming galaxy populations at redshift @xmath0 ; i.e. submillimetre galaxies ( smgs ) and bzk - selected galaxies ( bzks ) . out of 60
smgs found in the subaru / xmm - newton deep field , we collect optical
nir photometry of 28 radio counterparts for 24 smgs , based on refined sky positions with a radio map for 35 smgs ( ivison et al .
2007 ) .
we find a correlation between their @xmath1-band magnitudes and @xmath2 [ @xmath3 colours : almost all of the @xmath1-faint ( @xmath4 ) radio - detected smgs have @xmath5 , and therefore bzks .
this result gives strong support to perform direct optical identification of smgs by searching for bzks around smgs .
we calculate the formal significance ( @xmath6 value ) for each of the bzk associations around radio - undetected smgs , and find 6 new robust identifications , including one double identification . from this analysis , we obtain the current best estimate on the surface density of bzk - selected smgs , which indicates that only @xmath7 per cent of bzks are smgs .
if bzks are normal disk - like galaxies at @xmath0 as indicated by the correlation between their star formation rate ( sfr ) and stellar mass and also by dynamical properties , smgs are likely to be merging bzks . in this case
, a typical enhancement of sfr due to merging is only a factor of @xmath8 , which is an order of magnitude lower than that of local ulirgs .
this may indicate that most of the merging bzks could be observed as smgs . considering a possible high fraction of mergers at @xmath0
( at least it would be higher than the fraction at @xmath9 of @xmath10 per cent ) , it is rather puzzling to find such a low fraction of smgs in the progenitor population , i.e. bzks .
galaxies : starburst dust , extinction infrared : galaxies submillimetre galaxies : evolution .
|
the role of mergers in galaxy evolution has become increasingly recognized recently , stimulated by the central role of cdm cosmology ( navarro , frenk & white 1994 , 1995 ; power et al .
the importance of mergers was realized from purely observational arguments by toomre & toomre ( 1972 ) ; they argued that as tidal encounters generate short lived features , to yield today s `` peculiar '' galaxies a population of binary galaxies with highly eccentric orbits is required .
if these have a flat binding energy distribution , their merger rate must have declined with time as @xmath2 so that the ten obviously merging objects in the new general catalogue must be the tail end of 750 remnants ( see also toomre , 1977 ) .
zepf & koo ( 1989 ) , prior to the hubble deep fields , and abraham ( 1999 ) using their contents , inferred that galaxy pair density grows as @xmath3 , while brinchmann et al .
( 1998 ) showed that irregular galaxies form 10% of the total at @xmath4 and 30% at @xmath5 .
peculiar morphologies , characteristic of mergers and interactions dominate at high redshift .
the route to ellipticals via major mergers of spirals was first explored by toomre & toomre ( 1972 ) , by toomre ( 1977 ) , and by verraraghavan & white ( 1985 ) .
although most studies focused on the more spectacular major mergers , studies of the galaxy have brought out the importance of minor mergers and accretions .
for example gilmore & feltzing ( 2000 ) in a major review , using arguments based on metallicity and kinematics , discuss its structural components : bulge , thin disc , thick disc ( freeman 1993 ; gilmore & wyse 1998 ) , and stellar halo , showing that the thin disc and bulge are relatively young , while the thick disc and halo are relatively old .
in particular the thick disc formed at a specific epoch , some 10 gyr ago ( gilmore & wyse 1998 ) as a result of a minor merger , and the galaxy is accreting material now as it disrupts its smaller neighbours ( lynden bell 1976 ; mirabel 1982 ; savage et al . 2000 ) . here
we will point up a process which is playing a key role in the evolution of the galaxy , but which tends to be overlooked : the accretion of gas clouds of sub galactic mass .
this must be important in galaxy groups , small clusters , and the outskirts of rich clusters in general .
we will show here , as the main thrust of the article , that the galactic accretion rate has not declined during the disc lifetime and while this may seem surprising , it is a natural consequence of a plausible model for the local group . to maintain the star formation rates ( sfr s ) observed in late type galaxies now a steady inflow of gas from outside the galaxy is maintained .
a key difference between a gas rich and a gas poor galaxy is that the capture cross
section for gas cloud accretion is higher in the former ; this alone could account for the lower mean sfr s seen in late type galaxies but also for their greater relative scatter .
the distribution of star formation across a galaxy disc is also affected strongly by gas flows , but here by in
plane flows .
we bring this out in a semi quantitative way in the final part of the paper , where we present observations of the relation between the velocity gradient of the gas flow in a strongly barred galaxy , ngc 1530 , and the local formation rate of massive stars , inferring that compression enhances the local sfr , while shear reduces it .
long before galaxy mergers were considered important , larson ( 1972a , b ) suggested that long term infall of low metallicity gas to the galactic disc could explain the observed relative dearth of metal poor stars ( the g dwarf problem ) .
an advantage of this scenario over its rivals is that it offers an explanation for the presence of measurable deuterium abundance in the galaxy , notably near the centre ( audouze et al .
1976 , and more recently lubowich et al .
2000 ) . with no known source of d within the galaxy , and as astration destroys it , continuous replenishment by infall can explain its presence .
high redshift d abundances from the ly@xmath6 forest ( e.g. kirkman et al .
2000 ; pettini & bowen 2001 ) are @xmath72 - 4@xmath810@xmath9 , while recent fuse values for the local ism ( e.g. lehner et al .
2002 ; oliveira et al .
2003 ) are @xmath71 - 2@xmath810@xmath9 , and to complete the argument for infall the d abundance recently measured with fuse in the high velocity cloud ( hvc ) `` complex c '' ( sembach et al . 2004 ) is 2.2@xmath810@xmath9 , a value intermediate between the first two .
the idea is that some hvc s are supplying metal
poor non astrated gas continuously to the galaxy : they are the infall component mentioned
. some early models of galactic inflow ( hunt & sciama , 1972 ; tinsley 1977 ) , which derived infall mass rates similar to the galaxy wide sfr , entailed a sweeping up of local group gas , while larson ( 1976 ) suggested collapsing remnants of the pregalactic nebula .
muller , oort & raimond ( 1963 ) detected hvc s in at 21 cm ; oort ( 1970 ) noted that some of them have radial velocities higher than the local galactic escape velocity , and might be primordial .
recent observations that their metallicities are @xmath10 solar ( e.g. lehner et al .
2002 ; wakker et al . 2003 ; savage et al . 2003
; sembach et al .
2004 ) , rule out primordiality , but also rule out a galactic origin , so they are candidates for low metallicity infall . evidence that some hvc s may well belong to the local group but not the galaxy had been accumulating ( mirabel 1982 ; lepine & duvert 1994 ; blitz et al . 1999 ; lpez corredoira et al . 1999 ; casuso & beckman 2001 ; gibson et al .
2001 ; putman et al .
2002 ; and recently braun et al .
2002 ) in a sensitive survey , estimated that a gas mass of a few times 10@xmath11 could have accreted to both the milky way and m31 during their lifetimes .
direct evidence for infall for other galaxies is small .
phookun et al .
( 1993 ) detected falling onto ngc 5254 , while recently fraternali et al .
( 2004 ) detected flows perpendicular to the planes of nearby spirals using as the tracer .
observations of infall to external galaxies are few so far because the detection sensitivity has been limited at @xmath710@xmath12 of gas , while it is probable that the bulk of these flows are in clouds of lower mass . as sensitivities have improved ,
signs of accretion are being reported , see e.g. van der hulst & sancisi ( 2003 ) . as parameter ) from cb97 .
two peaks , for halo and thin disc stars , are clearly seen .
the thick disc contribution , less obvious , lies between them and widens the higher metallicity peak ( for the disc stars).,width=188 ] the g dwarf problem is a strong pointer to infall for the milky way ( van den bergh 1962 ; schmidt 1963 ; pagel & pachett 1975 ; tinsley 1980 ; and pagel 1987 ) .
closed box models for chemical evolution of the galactic disc predict far more low metallicity stars than those observed , and integrated population studies of external galaxies point to the same conclusion ( worthey et al .
1996 ; espana & worthey 2002 ; bellazzini et al .
late f and g stars may be nearly as old as the disc , and preserve a record of its evolution though the oldest have evolved off the main sequence , so as improving technique brought k dwarf statistics within range ( flynn & morrell 1997 ) their local metallicity distribution has given an more accurate history to follow . excluding halo and thick disc stars kinematically , a frequency @xmath13 metallicity plot for g , or better k dwarfs
show low numbers for low disc metallicity ( [ fe / h]@xmath14 ) rising sharply to a narrow peak between [ fe / h]=-0.6 and -0.2 , then falling off to higher metallicities .
the earlier g dwarf statistics ( carney et al .
1990 ; rocha pinto & maciel 1996 ) are well explained in a scenario with near constant infall of low metallicity gas to the galactic plane ( casuso & beckman 1997 ) . the k dwarfs show the same metallicity pattern ( flynn & morrell 1997 ; favata et al .
1997 ; kotoneva et al .
2002 ) and we will discuss this and how it is modelled in sect . 4 below .
1 shows an early stellar metallicity frequency plot , taking in thin disc , thick disc and halo g dwarfs from casuso & beckman ( 1997 ) ( hereinafter cb97 ) .
further chemical evidence explained well by low metallicity infall is the evolution of beryllium and boron , measured @xmath13 iron as standard in local disc stars . a recent compilation of observational data on this is shown in fig . 2 ( left and right ) .
the `` loop back '' seen for disc stars in both plots is well explained by infall , and best explained if the infall rises slowly with time .
recent models for be and b evolution in the galaxy ( fields et al 2000 ; ramaty et al .
2000 ) have focused on their evolution in the halo , i.e. for [ fe / h]@xmath14 , invoking specific mechanisms for their production by high cosmic ray fluxes in star forming zones to explain the observed linear relations of both be and b with fe . but their treatment of the disc evolution of be and b lacks finesse , notably in the chemical evolution model used , mainly because the authors are not really aiming to explain the disc observations , even though over 90% of be and b were made in the disc .
alibs et al .
( 2002 ) use a much better disc model , but even they do not predict very well the observations in fig . 2 . in cb97
we had managed to model earlier versions of these data sets . to our surprise , even concern , the model entailed infall with a rising rate over the disc lifetime . in 1997
the b observations were sparse but as shown in fig .
2 the trends for b and be @xmath13 fe have been strongly confirmed as data has accumulated .
it was these b and be observations which first pushed us towards an increasing inflow rate scenario , but other types of evidence have accumulated to support this as we now describe .
since vaughan & preston ( 1980 ) brought out the possible use of chromospheric activity in late type dwarfs as a chronometer , a number of surveys within 1 kpc of the sun have used activity indices to investigate the time dependence of the local sfr .
barry ( 1988 ) found using this technique that the sfr , averaged in bins of 10@xmath11yr showed evidence of a secular increase during the disc lifetime , but also large amplitude excursions in this general trend . from a much wider stellar sample rocha
pinto et al .
( 2000 ) confirmed both findings , as we show here in fig .
although both samples were taken locally to the sun , wielen ( 1977 , 1996 ) showed that the diffusion of stellar orbits implies that such a sample must include stars which have drifted radially from their birthplaces by over 2 kpc , both inwards and outwards .
thus any large excursion in the sfr must have occurred simultaneously over a major fraction of the galactic disc .
interaction could cause this , but not only with another galaxy ; accretion of a large gas cloud will also work . the data in fig .
3 are well interpreted in a scenario of infall of clouds with a range of sizes .
we can estimate a time scale for single cloud accretion , rough as it depends on distance estimates , using measured properties of extragalactic hvc s . from blitz et al .
( 1999 ) or putman et al .
( 2002 ) we set a lowish scale size for a large cloud at @xmath75 kpc .
measured radial velocity of @xmath7200 km s@xmath15 then gives an accretion time of a few times 10@xmath12 yr , which is consistent with the widths of the surge peaks in fig .
3 . also from fig .
3 we can see a long term slowly rising trend , which is clearly worth further comment .
is this apparently rising sfr compatible with cosmological trends ?
since the madau et al .
( 1996 ) plot it is known that the global sfr in the universe has been falling steadily from @xmath1 to the present epoch . we must note though that the local group is gravitationally bound ; interaction rates between galaxies have been steady since the group condensed
. the hvc accretion rate to the milky way must depend on the density of the clouds within the local group , the gravitational range of the galaxy , and its accretion cross section .
while accretion causes the first parameter to fall with time , it causes the second and third to rise as gas accumulates in the plane . in lpez
corredoira et al . ( 1999 ) we computed that for the current accretion rate to be rising ( or constant in the limiting case ) the present mean cloud density must have a lower limit , which can be converted to an estimate of the fractional mass of the local group in the form of these intracluster clouds .
the value is @xmath750% , which is fully plausible , comparing the masses of its galaxies ( the milky way and m31 as the smaller objects contribute little ) with dynamical estimates of the local group mass , as first done by kahn & woltjer ( 1959 ) ; for a recent estimate see whiting ( 1999 ) .
( note that these conclusions depend on only one assumption about dark matter : that the baryon to dark matter density does not vary much from object to object , which we can not discuss in detail here , but we take as reasonable ) .
we have also shown , in lpez
corredoira et al . (
2002 ) that if the net long term vector of this inflow is directed to the local group barycentre ( blitz et al . 1999 ) its dynamic parameters are those required to yield the known milky way warp amplitude and direction .
a global check on the inflow rate can come from measurements of the sfr .
for example sn ii rates , of order a few per century in the disc , ( e.g. dragicevich et al .
1999 ) agree with an accretion rate of a few yr@xmath15 , using a salpeter imf with mass limits 0.1 and 100 , and that type ii sne come from stars with m@xmath168 .
chemical evolution models require mean accretion rates during the disc lifetime of @xmath72 yr@xmath15 , consistent with estimates of hvc accretion rates ( blitz et al .
1999 ; braun et al .
the dearth of low metallicity k dwarfs in the galactic disc is a more severe test of chemical evolution models than previous similar data for g dwarfs . in a recent article ( casuso & beckman 2004 ) we describe the observational and theoretical developments in this field and this section gives an overview of these .
the g dwarf data by rocha
pinto & maciel ( 1996 ) already showed a much narrower peak in the metallicity frequency distribution than that in fig .
1 . using a chemical evolution model ( cb97 ) developed to explain the be and b data , in casuso & beckman ( 2001 ) ( cb01 ) we modelled this narrow peak well .
the model included a rising infall rate , and predicted even better the g dwarf data of jorgensen ( 2001 ) , published while cb01 was submitted , so this was a genuine prediction and not a post diction as one so ofter finds .
it is significant that we selected the best model for its be , b predictions , although improvements were possible using only the rocha
pinto & maciel ( 1996 ) data , so the improved agreement with jorgensen was remarkable .
one feature of jorgensen s data was the careful exclusion of thick disc stars from the sample . in fig .
4 we compare the two data sets with the model from cb01 ; these fits , especially to jorgensen s data , strengthened our `` increasing infall rate '' hypothesis.the next step was to use k - dwarf metallicity distributions . in fig.5
we show two data sets , by favata et al .
( 1997 ) and kotoneva et al .
( 2002 ) , compared with our model prediction.we can see that agreement is good in the metallicity range close to solar , but less good outside this range .
of the two data sets , one lies above and the other below our prediction , at lower metallicities .
the cause in both cases is the thick disc population .
favata et al . (
1997 ) cut off their graph where the thick disc numbers become appreciable , while kotoneva et al .
( 2002 ) did not take special steps to exclude thick disc stars .
we conclude that the k dwarf metallicity distribution in the thin disc gives some support to our model , but further measurement and analysis are needed here .
the light elements : hydrogen , helium , lithium , beryllium and boron , are exceptional in that their main production sites are principally non - stellar . hydrogen ( h , d and t ) most present day helium ( @xmath17he and @xmath18he ) , and some 10% of lithium are primordial , while the remaining lithium , beryllium and boron were formed in the interstellar medium ( ism ) in processes involving galactic cosmic rays ( gcr ) .
reeves ( 1974 ) reviewed light element nucleosynthesis setting the scene for all subsequent developments .
for the present article the importance of the light elements is that they present sensitive tests of abundance dilution by infall of element poor gas into the ism where they are being produced .
we have presented the be and b @xmath13 fe data plots in fig . 2 , and in fig .
6 we show a similar compilation for li . for be and b
there has been much interest in why they show linear dependences on metallicity ; they are spallation products of cno so initial predictions were for a quadratic dependence , as secondary elements . without entering in depth in this complex field ,
it is clear from fig .
2 that this linear relation holds only in the halo metallicity range ( [ fe / h]@xmath19 - 1.5 ) while for disc metallicities the dependence is far from linear .
clearly halo and disc behaviour are different , and this can be explained independently of the specific gcr production mechanisms for be and b by differences in the surrounding gas flow conditions . for the halo stars , not the subject of this article , a model in which the star forming gas is flowing inwards to the galactic centre as the initial spherical collapse occurs , gives a satisfactory explanation of the pseudo linear be and b @xmath13 fe observations .
for the disc stars we show four model comparisons , a closed box model and three inflow models .
the best fits are for secularly slowly increasing inflow .
the same model , with no fine tuning , was used to fit the g- and k dwarf distributions shown in figs . 4 and 5 , and is used as a framework for predicting the li @xmath13 fe plot in fig . 6 ,
though here the importance of both production and destruction mechanisms is such that the goodness of fit is not a simple test of the gas flow model ( see casuso & beckman 2000 for more details ) .
finally we can return to the case of deuterium which we discussed briefly above . as deuterium is destroyed in stars
, there are no equivalent plots to those in fig . 2 for this nuclide .
however the mere presence of d in the ism of the galaxy today , especially near the galactic centre , implies very recent infall , and we discussed implications of this in casuso & beckman ( 1999 ) . to summarize this section , the best fits to the observed evolutionary plots of light element abundances in the galactic disc
are provided by models in which the infall of has occurred at a constant or slowly rising rate . ) for local disc stars ; compilation from casuso & beckman ( 2000 ) .
the upper envelope represents minimum li depletion in the star observed , and is the datum for model fitting .
compared are a number of models , also from casuso & beckman ( 2000 ) .
the complexity of the production and destruction processes implies that model fits do not tell us too much about the time dependence of gaseous infall for li , but all models showing fair fits to the upper envelope use constant or increasing infall assumptions.,width=207 ]
far less attention has been paid to gas accretion as a driver of galaxy evolution than to mergers , because the latter are more spectacular and much easier to observe .
the lower end of the galaxy mass function should contain clouds whose baryonic mass could be as high as 10@xmath12 , and would range downwards from there .
their numbers should be high , conforming to general laws of mass fractionation , which elmegreen ( 2002 ) has characterized as fractal in his discussion of galactic gas clouds .
estimates at the low end of the galaxy mass function by bell et al .
( 2003 ) confirm that the number of objects is rising as the gas to stars ratio is rising , for the faintest dwarf galaxies , and should keep rising for `` failed galaxies '' which are the gas clouds we `` need '' .
rosenberg & schneider ( 2002 ) find a slope of -1.5 for the mass function for galaxies .
@xmath20cdm model cosmologies ( see kauffmann , white & guiderdoni 1993 , for an early model including @xmath20 ) do predict a steep mass function at the low mass end , though they also overpredict satellite galaxy numbers , which is a problem .
so we would expect galaxy groups and clusters to contain clouds in the mass range proposed here , but most such objects lie below present detectability limits ( @xmath710@xmath21 ) and can be picked up only in nearby galaxy groups . one result of considerable relevance is that by kennicutt , tamblyn & congdon ( 1994 ) . in a survey they found that early type spirals have much lower current sfrs than late types ( as expected ) but a much higher fractional variability in the sfr .
tomita et al .
( 1996 ) inferred that their results imply sfr variability on timescales of 10@xmath12 years , and hirashita & kamaya ( 2000 ) modelled this using a scenario with quasi
periodic limit cycles .
one problem with this mechanism is to explain how sfr surges can occur over a whole galaxy .
we have shown how the sfr excursions in rocha
pinto et al .
( 2000 ) imply galaxy wide changes , which are consistent with triggering by external clouds if these are large enough to affect a large part of the galaxy : sizes of several kiloparsecs for the `` external '' hvc s were estimated by blitz et al . ( 1999 ) and putman et .
al ( 2002 ) .
the variability difference between early and late type galaxies noted above is because a gas rich disc presents a larger cross
section for cloud accretion , and a higher probability for a star formation surge when a cloud is captured .
late types respond only to the most massive incoming clouds , and do so in a less dramatic way . the current gas accretion rate of 2 yr@xmath15 cited in sect
. 3 , assuming constant infall , would imply 2@xmath810@xmath22 of gas accreted in the disc lifetime ; an exponentially rising rate starting from zero would give 1.4@xmath810@xmath22 .
using a dark to baryonic matter ratio of 5 the total mass accreted would be between 7 and 10 times 10@xmath22 .
these total masses will be upper limits as the dark matter component of any hvc with incoming velocity higher than escape velocity will not be retained ( this would give a natural explanation for possible low dark matter : baryonic matter ratios in massive late type spirals ) .
these predictions are just compatible with observationally based estimates .
nikiforov et al .
( 2000 ) estimate a galaxy disc mass of 6@xmath810@xmath22 within a 20.5 kpc radius , in a total galaxy mass of 3.3@xmath810@xmath23 . even if these figures are rough , we find that a major fraction of the disc has been accreted by gas infall .
this is qualitatively compatible with `` inside out '' evolution of the disc , in which growing gas column density as a function of time and radius have produced increased sfrs at smaller radii , which in turn yields a radial metallicity gradient in the disc . inside out
disc formation occurs in many @xmath20cdm based scenarios ( e.g. samland & gerhard 2003 ) . in any model where
the disc is built up by gas accretion the column density threshold for quiescent star formation will give higher sfr s in the inner disc , although diffusion , especially if impelled by a bar , will tend to blur out the radial distribution of stellar age and metallicity . as it is extremely hard , as yet to estimate stellar population age distributions from outside a galaxy , the local measurements we have used to predict the time dependence of infall must be used as the best benchmark for some time to come .
so far we have been discussing accretion inflow perpendicular to the planes of galaxies . in this shorter final section
we will see how gas flows within the plane affect the local sfr on short timescales . for this
we use our recently measured velocity field of the strongly barred galaxy ngc 1530 . in zurita
et al . ( 2004 ) we describe the taurus fabry prot interferometer on the 4.2 m wht , la palma ( spain ) with which we made the two dimensional velocity map of the ionized hydrogen using emission , and give details of the observations and reduction .
the map has a velocity interval of @xmath718.6 km s and an angular resolution of @xmath71 , which mark improvements on previous optical and radio maps by regan et al .
( 1996 ) , by downes et al .
( 1996 ) and by reynaud & downes ( 1997 , 1998 , 1999 ) .
we derived first the classical rotation curve , and used its two dimensional projection to subtract off from the observed velocity field , yielding a residual field of non circular projected velocity over the full face of the galaxy . in fig .
[ mos_gal ] we show how this was done and illustrate our results . ngc 1530 is an excellent subject for kinematic analysis .
its arms , bar and circumnuclear disc are very well separated ; the major axis and bar position angle are well separated , and the inclination permits good velocity and intensity measurement . in fig .
[ mos_gal]c , streaming in the arms due to the disc density wave system shows up as ripples in the two dimensional velocity field , with amplitude 20 - 30 km s , which we will not analyze further here . in fig .
[ mos_gal]b the non circular motions round the bar show up clearly as distorted isovels .
these motions are well seen in fig .
[ mos_gal]d : the residual non circular velocity field , after subtracting off the rotational field seen in fig .
[ mos_gal]c .
cross sections through this field
are shown in fig .
[ perfiles ] , revealing counterflow with amplitude @xmath24100 km s top either side of the bar .
d , showing the amplitude of the flows , @xmath16100 km s in opposite directions at either side of the bar.,width=336 ] although gives patchy velocity fields compared with , we have high spatial resolution to compensate .
this allows us to produce maps of non circular velocity gradient across the whole galaxy from a few hours observations . in a poster complementary to this article ( zurita et al .
2004a ) we show a beautiful result of this work : the complete coincidence of the strong dust lanes along the bar with the loci of maximum velocity gradient perpendicular to the bar .
this shows that lines of maximum shear , which are also lines of net zero non
circular velocity , are zones to which the dust migrates .
we can draw more powerful inferences by superposing cross sections of our maps in velocity gradient on cross sections across the surface brightness map ( fig .
[ mos_gal]a ) ; the results are shown in fig .
[ int+grad ] .
these show clearly an anticorrelation between local sfr ( surface brightness ) and local velocity gradient , but in two distinct regimes . in zones of high shear ( gradient perpendicular to the flow along the bar ) star formation
is suppressed , a conclusion inferred by reynaud & downes ( 1998 ) from co mapping .
this is as expected if shear disrupts large molecular clouds on timescales shorter than those for massive star formation ( see kennicutt 1998 for a discussion of this and related points ) . in zones of high compression star formation
is enhanced , but occurs offset by hundreds of parsecs from the zone of maximum present compression . the offset relation ; compression high local sfr is seen on scales of hundreds of parsecs . on smaller scales
we see the effects of outflow from luminous regions which impinges on the surrounding ism , yielding `` walls '' of velocity gradient around these regions .
these can be seen in fig . 1 of zurita et al .
( 2004a ) , and show up here in fig .
[ int+grad ] , where each peak in the surface brightness due to a bright region coincides with a dip surrounded by symmetrically offset peaks , in the velocity gradient .
these outflows have been reported by relao et al .
( 2003 ) , and have amplitudes of 4080 km s from the centre of the region . we will present a quantitative treatment of these outflows and some initial models in relao et al . (
2004 ) . in the inner 5 of our original rotation curve ( zurita et al .
2004 ) we found a steep gradient which relaxed to a slightly less steep gradient outside this radius . assuming this to be caused by a bulge , we used an hst - nicmos near - ir image to check this , but as seen in fig .
[ mos_centro]a the central spheroidal component is diminutive , and could not cause the steep inner gradient . assuming that we are detecting projected
non circular motion , we extrapolated the outer part of the rising rotation curve linearly to the origin , and subtracted this from the observed inner velocity field , yielding a map of projected non circular velocity in the inner disc , as shown in fig .
[ mos_centro]c .
this `` yin / yang '' pattern reveals spiral inflow to the nucleus . to show this we took an image of the zone in j
k from prez - ramrez et al .
( 2000 ) , see fig .
[ mos_centro]b , which shows interlocking spiral dust lanes .
assuming that these indicate inward flow lines we computed the projected velocity we would see , taking for simplicity a constant velocity amplitude along the flow .
the result , in fig .
[ mos_centro]d agrees nicely with the observational field in fig .
[ mos_centro]c , except at the north and south ends of the inner disc , where we know that the data do not allow us to plot the true circular rotation curve , so the residual map in fig .
[ mos_centro]c will have systematic errors here .
this inward spiralling flow is not due to an inner bar . in
zurita et al . ( 2004 , 2004a ) we show using unsharp masking of the hst image that any circumnuclear bar must be less than 0.5 in length , and with negligible dynamical effect over the inner disc .
however , as modelled by englmeier & shlosman ( 2000 ) in a galaxy with an inner lindblad resonance at a radius within the length of a major bar , as the dominant stellar @xmath25 orbits in the bar give way to perpendicular @xmath26 orbits within the ilr the gas is pulled round into opposing spiral trajectories , very much as seen in fig .
[ mos_centro ] , which take it down to within a couple of hundred parsecs of the nucleus .
prot velocity fields from ionized gas emission give us an excellent tool for exploring these dynamical effects .
the message of this article is the same for the two different astrophysical systems considered : galaxy evolution can not be well understood without a knowledge of gas flows and their effects on star formation . for the milky way our reasoning has been indirect .
chemical evidence is used to show that gas has been accreting to the disc during its whole lifetime , and at rates which do not appear to have declined during this period .
this does not contradict the general picture in which the sfr in the universe has been declining steadily from @xmath1 to the present epoch , as the overall density has fallen .
we note that the ongoing global disc sfr is of order 2 yr , while during the formation of the initial spherical component ( stellar halo ) the rate must have been much higher .
this means that a constant , even slowly increasing sfr in the discs of galaxies and small groups , regulated by inflow and feedback processes , need not contradict inferences about the universal sfr from the madau et al .
( 1996 ) plot and its subsequent refinements .
one feature of star formation in galactic discs is its apparently low efficiency averaged over time . in a model
where the disc mass has grown slowly via accretion , this low efficiency is only apparent , an effect of time averaging .
any gas which arrives as infalling suffers relatively rapid conversion to stars , subject to feedback from the massive stars themselves .
this scenario of gas persistently raining down onto galaxy discs , with the occasional sharper shower , a more massive cloud , is finding support as detection sensitivities increase , and more clouds falling in to other galaxies are detected . and
infall can explain rather well what other mechanisms ( major interactions apart ) can not : how the star formation rate can increase and decrease simultaneously over the whole , or the major part of a milky way sized disc galaxy .
we have considered here the evidence that this does happen , both for the milky way and for galaxies in general . in the second part of the paper
we have shown how gas flow is a key parameter determining the star forming pattern in ngc 1530 .
while shocks , strong velocity gradients along the line of flow , yield massive stars , shear , strong velocity gradients perpendicular to the flow , inhibit massive star production .
ngc 1530 is a specially favourable case for study , for the essentially geometrical and morphological reasons explained in sect . 7 , but we should treat it here as an example revealing processes of general importance . here
we have been able to present a semi quantitative picture , but to disentangle fully the effects produced by gas flow on star formation we will need , as observers , to combine similar information from ionized , neutral ( atomic ) and molecular gas . only then will we have the basis for full quantitative understanding , aimed at the major prize : a physical theory of star formation in different galactic environments . _
the research discussed in this article was supported by grants aya2001 - 0435 ( spanish ministry of science and technology ) and aya2004 - 08251-c02 - 01 ( spanish ministry of education and science ) .
a. zurita acknowledges support by the consejera de educacin y ciencia de la junta de andaluca , spain .
the authors thank the organizers of the conference for the invitation to present this paper in an excellent context , both scientific and social . _ 1 abraham , r. g. 1999 in galaxy interactions at low and high redshifts , ( eds .
barnes and d.b .
sanders ) , dordrecht , kluwer , p.11 alibs , a. , labay , j. , & canal , r. 2002 , apj , 571 , 326 audouze , j. , lequeux , j. , reeeves , h. , & vigroux , l. 1976 , apj , 208 , l51 barry , d. c. 1988 , apj , 324 , 436 bell , e. f. , mcintosh , d. h. , katz , n. , & weinberg m. d. 2003 , apjl , 585 , 117 bellazzini , m. , cacciari , c. , federici , l. , fusi pecci , f. , & rich , m. 2003 , a&a , 405 , 867 blitz , l. , spergel , d. n. , teuben , p. j. , hartmann , d. , & burton , w.b . 1999 ,
apj , 514 , 818 braun , r. , de heij , v. , & burton , w. b. 2002 , baas , 200 , 3304 carney , b. w. , latham , d. w. , & laird , j. b. 1990 , aj , 99 , 572 casuso , e. , & beckman , j. e. 1997 , apj , 475 , 155 ( cb97 ) casuso , e. , & beckman , j. e. 1999 , aj , 118 , 1907 casuso , e. , & beckman , j. e. 2000 , pasp , 112 , 942 casuso , e. , & beckman , j. e. 2001 , apj , 557 , 681 ( cb01 ) casuso , e. , & beckman , j. e. 2004 , a&a , 419 , 181 ( cb04 ) .
downes , d. , reynaud , d. , solomon p. m. , & radford , s.j.e .
1996 , apj , 461 , 186 dragicevich , p. m. , blair d. g. , & burman , r. r. 1999 , mnras , 302 , 693 elmegreen , b. g. 2002 , apj , 564 , 773 englmeier , p. , & shlosman i. 2000 , apj , 528 , 677 espana , l. , & worthey , g. 2002 , aas , 201 , 1408 favata , f. , micela , g. , & sciortino , s. 1997 , a&a , 323 , 809 fields , b. d. , olive , k. a. , vangioni
flam , e. , & cass , m. 2000 , apj , 540 , 930 flynn , c .
, & morell , o. 1997 , mnras , 286 , 617 fraternali , f. , osterloo , t. , & sancisi , r. 2004 , a&a ( in press ) freeman , k. c. 1993 in `` galaxy evolution : the milky way perspective''(ed .
s. r. majewski ) asp converence series , vol .
49 , asp , san francisco , 1993 gibson , b. k. , penton , s. v. , giroux , m. l. , stocke , j. t. , shull , j. m. , & tumlinson , j. 2001 , aj , 122 , 3280 gilmore , g. , & feltzing s. 2000 in `` the evolution of galaxies on cosmological timescales '' ( eds . j. e. beckman & t. j. mahoney ) , asp conference series , vol . 187 , asp , san francisco , p. 20
gilmore , g. , & wyse , r. f. g. 1998 , aj , 116 , 748 hirashita , h. , & kamaya , h. 2000 , aj , 120 , 728 hunt , r. , & sciama , d. w. 1972 , mnras , 157 , 335 jorgensen , b. r. 2001 , a&a , 363 , 947 kahn , f. d. , & woltjer , l. 1959 , apj , 130 , 70 kauffman , g. , white , s. d. m. , & guiderdoni , b. 1993 , mnras , 264 , 201 kennicutt , r. c. 1998 , ara&a , 36 , 189 kennicutt , r. c. , tamblyn , p. , & congdon , c. w. 1994 , apj , 435 , 22 kirkman , d. , tytler , d. , burles , s. , lubin , d. , & omeara , j. m. 2000 , apj , 529 , 655 kotoneva , e. , flynn , c. , chiappini , c. , & matteucci f. , 2002 , mnras , 336 , 879 larson , r.b .
1972a , nature , 236 , 21 larson , r.b .
1972b , nphs , 236 , 7 larson , r.b .
1976 , mnras 176 , 31 lehner , n. , gry , c. , sembach , k. , hbrard , g. , chayer , p. , moos , h. w. , howk , j. c. , desert , j. m. 2002 , apjs , 140 , 81 lepine , j. r. d. , & duvert , g. 1994 , a&a , 286 , 60 lpez
corredoira , m. , beckman , j. e. , & casuso , e. 1999 , a&a , 351 , 920 lpez
corredoira , m. , betancort
rijo , j. , beckman , j. e. 2002 , a&a , 386 , 169 lubowich , d. a. , pasachoff , j. m. , balonek , t. j. , millar , t. j. , tremonti , c. , roberts , h. , & galloway , r. p. 2000 , nature , 405 , 1025 lynden bell , d. 1976 , mnras , 174 , 695 madau , p. , ferguson , h. c. , dickinson , m. e. , giavalisco , m. , steidel , c. c. , fruchter , a. 1996 , mnras , 283 , 1388 mirabel , f. 1982 , apj , 256 , 112 muller , c. a. , oort , j. h. , & raimond , e. 1963 , comptes rendus acad .
paris 257 , 1661 navarro , j. f. , frenk , c. s , & white , s. 1994 , mnras , 267 , 1 navarro , j. f. , frenk , c. s , & white , s. 1995 , mnras , 275 , 56 nikiforov , i. i. , petrovsky , i. v. , & ninkova , s. 2000 , in `` small galaxy groups '' asp conference series 209 , ( eds . m. valtonen & c. flynn ) , p. 399oliveira
, c. m. , hbrard , g. , howk , j. c. , kruk , j. w. , chayer , p. , & moos , h. w. 2003 , apj , 587 , 235 oort , j. h. 1970 , a&a , 7 , 181 pagel , b. e. j. 1987 in `` the galaxy '' ( eds . g. gilmore & r. carswell ) d.reidel , p. 341
pagel , b. e. j. , patchett , b. e. 1975 , mnras , 172 , 13 prez
ramrez , d. , knapen , j. h. , peletier , r. f. et al . 2000 ,
mnras , 317 , 234 pettini , m. , & bowen , d. v. 2001 , apj , 560 , 41 phookun , b. , vogel , s. n. , & mundy , l. g. 1993 , apj , 418 , 113 power , c. , navarro , j. f. , jenkins , a. , frenk , c. s. , white , s. , springel , v. , stadel , j. , & quinn , t. 2003 , mnras , 338 , 14 putman , m. e. et al .
2002 , aj , 123 , 873 ramaty , r. , scully , s. t. , lingenfelter , r. e. , & kozlovsky , b. , 2000 , apj , 534 , 747 reeves h. 1974 , ara&a , 12 , 437 regan , m. , teuben , p. , vogel , s. , & van der hulst , t. 1996 , aj , 112 , 2549 relao , m. , beckman , j. e. , zurita , a. , & rozas , m. 2003 , rev .
mex.astr astrofis . , 15 , 205relao , m. , beckman , j. e. , zurita , a. , & rozas , m. 2004 , a&a , ( submitted ) reynaud , d. , & downes d. 1997 , a&a , 319 , 737 reynaud , d. , & downes d. 1998 , a&a , 337 , 671 reynaud , d. , & downes d. 1999 , a&a , 347 , 37 rocha pinto , h. j. , & maciel , w. j. 1996 , mnras , 279 , 447 rocha
pinto , h. j. , scalo , j. , maciel , w. j. , & flynn , c. 2000 , a&a , 358 , 869 rosenberg , j. l. , & schneider , s. e. 2002 , apj , 567 , 247 samland , m. , & gerhard , o.e .
2003 , a&a , 399 , 961 savage , b. d. et al .
2000 , apjs , 129 , 563 savage , b. d. et al .
2003 , apjs , 146 , 165 schmidt , m. 1963 , apj , 137 , 758 sembach , k. r. et al .
2004 , apjs , 150 , 387 tinsley , b. 1977 , apj , 216 , 548 tinsley , b. 1980 , fund .
phys . , 5 , 287 tomita , a. , tomita , y. , & saito , m. 1996 , pasj , 48 , 285 toomre , a. 1977 in `` the evolution of galaxies and stellar populations '' ( eds .
b. tinsley & r. b. larson ) , new haven , yale u. obs . , p. 401
toomre , a. , & toomre , j. 1972 , apj , 178 , 623 valle , g. , ferrini , f. , galli , d. , & shore , s. n. 2002 , apj , 566 , 252 van den bergh , s. 1962 , aj , 67 , 486 van der hulst , t. j. m. , & sancisi , r. 2003 , in `` recycling intergalactic and interstellar matter '' , iau symposium 217 , p. 140
vaughan , a. , & preston , g. w. 1980 , pasp , 92 , 385 veeraraghavan , s. & white , s. d. m. 1985 , apj , 296 , 376 wakker , b. p. et al .
2003 , apjs , 140 , 91 whiting , a. b. 1999 , proc .
iau symposium 192 `` the stellar content of the local group of galaxies '' ( eds .
whitelock & r. cannon ) , asp , p. 420 wielen , r. f. 1977 , a&a , 60 , 263 wielen , r. f. , fuchs , b. , & dettbarn , c. 1996 , a&a , 314 , 438 worthey , g. , dorman , b. , & jones , l. a. 1996 , aj , 112 , 948 zepf , s. e. , & koo , d. c. 1989 , apj , 337 , 34 zurita , a. , relao , m. , beckman , j. e. , & knapen , j. h. 2004 , a&a , 413 , 73 zurita , a. , relao , m. , beckman , j. e. , & knapen , j. h. 2004a ( these proceedings )
|
in the first part of this article we show how observations of the chemical evolution of the galaxy : g- and k dwarf numbers as functions of metallicity , and abundances of the light elements , d , li , be and b , in both stars and the interstellar medium ( ism ) , lead to the conclusion that metal poor gas has been accreting to the galactic disc during the whole of its lifetime , and is accreting today at a measurable rate , @xmath0 per year across the full disc .
estimates of the local star formation rate ( sfr ) using methods based on stellar activity , support this picture .
the best fits to all these data are for models where the accretion rate is constant , or slowly rising with epoch .
we explain here how this conclusion , for a galaxy in a small bound group , is not in conflict with graphs such as the madau plot , which show that the universal sfr has declined steadily from @xmath1 to the present day .
we also show that a model in which disc galaxies in general evolve by accreting major clouds of low metallicity gas from their surroundings can explain many observations , notably that the sfr for whole galaxies tends to show obvious variability , and fractionally more for early than for late types , and yields lower dark to baryonic matter ratios for large disc galaxies than for dwarfs . in the second part of the article we use ngc 1530 as a template object , showing from fabry
prot observations of its emission how strong shear in this strongly barred galaxy acts to inhibit star formation , while compression acts to stimulate it .
galaxy : evolution , galaxy : accretion , galaxies : ism , galaxies : kinematics galaxies : star formation
|
since the development of _ periodic orbit theory _ by gutzwiller @xcite it has become a fundamental question as to how individual semiclassical eigenenergies and resonances can be obtained from periodic orbit quantization for classically chaotic systems .
a major problem is the exponential proliferation of the number of periodic orbits with increasing period , resulting in a divergence of gutzwiller s trace formula at real energies and below the real axis , where the poles of the green s function are located .
the periodic orbit sum is a dirichlet series @xmath2 where the parameters @xmath3 and @xmath4 are the amplitudes and periods ( actions ) of the periodic orbit contributions . in most applications eq .
[ g ] is absolutely convergent only in the region @xmath5 with @xmath6 the entropy barrier of the system , while the poles of @xmath1 , i.e. , the bound states and resonances , are located on and below the real axis , @xmath7 .
thus , to extract individual eigenstates , the semiclassical trace formula ( [ g ] ) has to be analytically continued to the region of the quantum poles . up to now no general procedure is known for the analytic continuation of a non - convergent dirichlet series of the type of eq .
all existing techniques are restricted to special situations . for bound and ergodic systems
the semiclassical eigenenergies can be extracted with the help of a functional equation and the mean staircase function ( weyl term ) , resulting in a riemann - siegel look - alike formula @xcite .
alternative semiclassical quantization conditions based on a semiclassical representation of the spectral staircase @xcite and derived from a quantum version of a classical poincar map @xcite are also restricted to bound and ergodic systems . for systems with a symbolic dynamics
the periodic orbit sum ( [ g ] ) can be reformulated as an infinite euler product , which can be expanded in terms of the cycle length of the symbolic code .
if the contributions of longer orbits are shadowed by the contributions of short orbits the cycle expansion technique can remarkably improve the convergence properties of the series and allows to extract the bound states and resonances of bound and open systems , respectively @xcite .
a combination of the cycle expansion technique with a functional equation for bound systems has been studied by tanner et al .
however , the existence of a simple symbolic code is restricted to very few systems , and cycle expansion techniques can not be applied , e.g. , to the general class of systems with mixed regular - chaotic classical dynamics .
in this paper we present a general technique for the analytic continuation and the extraction of poles of a non - convergent series of the type of eq .
the method is based on _ harmonic inversion _ by filter - diagonalization .
the advantage of the method is that it does not depend on special properties of the system such as ergodicity or the existence of a symbolic dynamics for periodic orbits
. it does not even require the knowledge of the mean staircase function , i.e. , the weyl term in dynamical systems .
the only assumption we have to make is that the analytic continuation of the dirichlet series @xmath1 ( eq . [ g ] ) is a linear combination of poles @xmath8 , which is exactly the functional form of , e.g. , a quantum mechanical response function with real and complex parameters @xmath9 representing the bound states and resonances of the system , respectively . to demonstrate the general applicability and accuracy of our method
we will apply it to two systems with completely different properties , first the zeros of the riemann zeta function @xcite , as a mathematical model for a bound system , and second the three disk scattering system as a physical example . as pointed out by berry @xcite the density of zeros of riemann s zeta function
can be written , in formal analogy with gutzwiller s semiclassical trace formula , as a non - convergent series , where the `` periodic orbits '' are the prime numbers .
a special property of this system is the existence of a functional equation which allows the calculation of riemann zeros via the riemann - siegel formula @xcite .
an analogous functional equation for quantum systems with an underlying chaotic ( ergodic ) classical dynamics has served as the basis for the development of a semiclassical quantization rule for bound ergodic systems @xcite .
the riemann zeta function has also served as a mathematical model to study the statistical properties of level distributions @xcite .
we will demonstrate that harmonic inversion can reveal the riemann zeros with extremely high accuracy and with just prime numbers as input data .
the most important advantage of our method is , however , its wide applicability , i.e. , it can be generalized in a straightforward way to non - ergodic bound or open systems . our second example , the three disk scattering problem , is an open and non - ergodic system .
its classical dynamics is purely hyperbolic , and the periodic orbits can be classified by a complete binary symbolic code . this system has served as a model for the development of cycle expansion techniques @xcite . when applying the harmonic inversion technique to the three disk scattering system we will highlight the general applicability of our method by not having to make use of its symbolic dynamics in any way .
it is evident that methods invoking special properties of a given system may be more efficient regarding , e.g. , the number of periodic orbits required for the calculation of a certain number of poles of the response function @xmath1 in that particular case .
it is not our purpose to compete with the efficiency of such methods .
rather , the advantage of the harmonic inversion technique lies in its wide applicability , which allows the investigation also of systems not possessing special properties .
this is demonstrated in this paper by solving two completely different problems , viz .
the zeros of the riemann zeta function and the three disk scattering system , with one and the same method .
the paper is organized as follows . in section 2
we explain the general idea of the method by way of example of the riemann zeros .
this is followed by the derivation of the harmonic inversion method in section 3 and the presentation of numerical results for the riemann zeros in section 4 .
the method is extended to the general case of periodic orbit quantization in section 5 , and its usefulness and wide applicability is demonstrated for the three disk scattering system , as a physical example , in section 6 .
our goal is to introduce our method for periodic orbit quantization by harmonic inversion using , as an example , the well defined problem of calculating zeros of the riemann zeta function .
there are essentially two advantages of studying the zeta function instead of a `` real '' physical bound system .
first , the riemann analogue of gutzwiller s trace formula is exact , as is the case for systems with constant negative curvature @xcite , whereas the semiclassical trace formula for systems with plane geometry is correct only to first order in @xmath10 .
this allows a direct check on the precision of the method .
second , no extensive periodic orbit search is necessary for the calculation of riemann zeros , as the only input data are just prime numbers .
it is not our intention to introduce yet another method for computing riemann zeros , which , as an objective in its own right , can be accomplished more efficiently by specific procedures .
rather , in our context the riemann zeta function serves primarily as a mathematical model to illustrate the power of our technique when applied to bound systems . before discussing the harmonic inversion method
we start with recapitulating a few brief remarks on riemann s zeta function necessary for our purposes .
the hypothesis of riemann is that all the non - trivial zeros of the analytic continuation of the function @xmath11 have real part @xmath12 , so that the values @xmath13 , defined by @xmath14 are all real or purely imaginary @xcite .
the riemann staircase function for the zeros along the line @xmath15 , defined as @xmath16 i.e. the number @xmath17 of zeros with @xmath18 , can be split @xcite into a smooth part , @xmath19 and a fluctuating part , @xmath20 substituting the product formula ( [ zeta_def ] ) ( assuming that it can be used when @xmath21 ) into ( [ n_osc_def ] ) and expanding the logarithms yields @xmath22 therefore the density of zeros along the line @xmath15 can formally be written as @xmath23 with the response function @xmath1 given by the series @xmath24 which converges only for @xmath25 .
obviously eq .
[ g_nc ] is of the same type as the response function ( [ g ] ) , with the entropy barrier @xmath26 , i.e. , eq .
[ g_nc ] does not converge on the real axis , where the riemann zeros are located .
the mathematical analogy between the above equation and gutzwiller s periodic orbit sum @xmath27 with @xmath28 the amplitudes and @xmath29 the classical actions ( including phase information ) of the periodic orbit contributions , was already pointed out by berry @xcite . for the riemann zeta function
the primitive periodic orbits have to be identified with the primes @xmath30 , and the integer @xmath31 formally counts the `` repetitions '' of orbits .
the `` amplitudes '' and `` actions '' are then given by @xmath32 both equation ( [ rho_zeta ] ) for the riemann zeros and for most classically chaotic physical systems the periodic orbit sum ( [ rho_po ] ) do not converge . in particular , zeros of the zeta function , or semiclassical eigenstates , can not be obtained directly using these expressions .
the problem is to find the analytic continuation of these equations to the region where the riemann zeros or , for physical systems , the eigenenergies and resonances , are located .
[ g_nc ] is the starting point for our introduction and discussion of the harmonic inversion technique for the example of the riemann zeta function .
the generalization of the method to periodic orbit quantization ( eq . [ rho_po ] ) in section 5 will be straightforward .
although eq .
[ g_nc ] is the starting point for the harmonic inversion method , for completeness we quote the riemann - siegel formula , which is the most efficient approach to computing riemann zeros . for the riemann zeta function it follows from a functional equation @xcite that the function @xmath33\right\ } \zeta\left({1\over 2}-iw\right)\ ] ] is real , and even for real @xmath34 . the asymptotic representation of @xmath35 for large @xmath34 , @xmath36 } \frac{\cos\{\pi\overline n(w)-w\ln n\}}{n^{1/2 } } \nonumber \\ & - & ( -1)^{{\rm int}[\sqrt{w/2\pi}]}\left({2\pi\over w}\right)^{1\over4 } \ ; { \cos\left(2\pi\left(t^2-t-{1/16}\right)\right ) \over\cos\left(2\pi t\right ) } + \dots \ ; , \label{riemann_siegel}\end{aligned}\ ] ] with @xmath37 $ ] is known as the riemann - siegel formula and has been employed ( with several more correction terms ) in effective methods for computing riemann zeros @xcite .
note that the principal sum in ( [ riemann_siegel ] ) has discontinuities at integer positions of @xmath38 , and therefore the riemann zeros obtained from the principal sum are correct only to about 1 to 15 percent of the mean spacing between the zeros .
the higher order corrections to the principal riemann - siegel sum remove , one by one , the discontinuities in successive derivatives of @xmath35 at the truncation points and are thus essential to obtaining accurate numerical results .
an alternative method for improving the asymptotic representation of @xmath35 by smoothing the cut - offs with an error function and adding higher order correction terms is presented in @xcite .
an analogue of the functional equation for bound and ergodic dynamical systems has been used as the starting point to develop a `` rule for quantizing chaos '' via a `` riemann - siegel look - alike formula '' @xcite .
this method is very efficient as it requires the least number of periodic orbits , but unfortunately it is restricted to ergodic systems on principle reasons , and can not be generalized either to systems with regular or mixed classical dynamics or to open systems .
by contrast , the method of harmonic inversion does not have these restrictions .
we will demonstrate that riemann zeros can be obtained directly from the `` ingredients '' of the non - convergent response function ( [ g_nc ] ) , i.e. , the set of values @xmath39 and @xmath40 , thus avoiding the use of the functional equation , the riemann - siegel formula , the mean staircase function ( [ n_bar ] ) , or any other special property of the zeta function . the comparison of results in section 4 will show that the accuracy of our method goes far beyond the riemann - siegel formula ( [ riemann_siegel ] ) without higher order correction terms .
the main goal of this paper is to demonstrate that because of the formal equivalence between eqs .
( [ rho_zeta ] ) and ( [ rho_po ] ) our method can then be applied to periodic orbit quantization of dynamical systems @xcite without any modification . to find the analytic continuation of eq .
( [ g_nc ] ) in the region @xmath41 we essentially wish to fit @xmath1 to its exact functional form , @xmath42 arising from the definition of the riemann staircase ( [ riemann_staircase ] ) . the `` multiplicities '' @xmath43 in eq .
[ g_ex ] are formally fitting parameters , which here all should be equal to 1 .
it is hard to directly adjust the non - convergent ( on the real axis ) series @xmath1 to the form of @xmath44 .
the first step towards the solution of the problem is to carry out the adjustment for the fourier components of the response function , @xmath45 which after certain regularizations ( see below ) is a well - behaved function of @xmath46 .
due to the formal analogy with the results of periodic orbit theory ( see eqs .
[ apm ] and [ spm ] ) , @xmath47 can be interpreted as the recurrence function for the riemann zeta function , with the recurrence positions @xmath48 and recurrence strengths of periodic orbit returns @xmath49 .
the exact functional form which now should be used to adjust @xmath47 is given by @xmath50 @xmath51 is a superposition of sinusoidal functions with frequencies referring to the sinusoidal form of @xmath47 .
we will also use the word `` poles '' in the context of the response function @xmath1 . ]
@xmath9 given by the riemann zeros and amplitudes @xmath52 . fitting a signal @xmath47 to the functional form of eq .
( [ c_ex ] ) with , in general , both complex frequencies @xmath9 and amplitudes @xmath43 is known as _ harmonic inversion _ , with a large variety of applications in various fields @xcite .
the harmonic inversion analysis is especially non - trivial if the number of frequencies in the signal @xmath47 is large , e.g. , more than a thousand .
it is additionally complicated by the fact that the conventional way to perform the spectral analysis by studying the fourier spectrum of @xmath47 will bring us back to analyzing the non - convergent response function @xmath1 defined in eq .
[ g_nc ] . until recently the known techniques of spectral analysis
@xcite would not be applicable in the present case .
it is the filter - diagonalization method @xcite which have turned the harmonic inversion concept into a general and powerful computational tool .
the signal @xmath47 as defined by eq .
[ c_nc ] is not yet suitable for the spectral analysis .
the next step is to regularize @xmath47 by convoluting it with a gaussian function to obtain the smoothed signal , @xmath53 that has to be adjusted to the functional form of the corresponding convolution of @xmath51 .
the latter is readily obtained by substituting @xmath43 in eq . [ c_ex ] by the damped amplitudes , @xmath54 the regularization ( [ c_sigma ] ) can also be interpreted as a cut of an infinite number of high frequencies in the signal which is of fundamental importance for numerically stable harmonic inversion .
note that the convolution with the gaussian function is no approximation , and the obtained frequencies @xmath9 and amplitudes @xmath43 corrected by eq.[d_sigma ] are still exact , i.e. , do not depend on @xmath55 .
the convolution is therefore not related to the gaussian smoothing devised for riemann zeros in @xcite and for quantum mechanics in @xcite , which provides low resolution spectra only . before proceeding
further we note that even though the derivation of eq .
[ c_sigma ] assumed that the zeros @xmath9 are on the real axis , the analytic properties of @xmath56 imply that its representation by eq .
[ c_sigma ] includes not only the non - trivial real zeros , but also all the trivial ones , @xmath57 , @xmath58 , which are purely imaginary .
the general harmonic inversion procedure described below does not require the frequencies to be real .
both the real and imaginary zeros @xmath9 will be obtained as the eigenvalues of a non - hermitian generalized eigenvalue problem .
the harmonic inversion problem can be formulated as a non - linear fit ( see , e.g. , ref .
@xcite ) of the signal @xmath47 defined on an equidistant grid , @xmath59 with the set of generally complex variational parameters @xmath60 .
( in this context the discrete fourier transform scheme would correspond to a linear fit with @xmath61 amplitudes @xmath43 and fixed real frequencies @xmath62 .
the latter implies the so called `` uncertainty principle '' , i.e. , the resolution , defined by the fourier grid spacing , @xmath63 , is inversely proportional to the length , @xmath64 , of the signal c(s ) . ) the `` high resolution '' property associated with eq .
[ eq : complexct ] is due to the fact that there is no restriction for the closeness of the frequencies @xmath9 as they are variational parameters . in ref . @xcite it was shown how this non - linear fitting problem can be recast as a linear algebraic one using the filter - diagonalization procedure .
the essential idea is to associate the signal @xmath65 with an autocorrelation function of a suitable dynamical system , @xmath66 where @xmath67 defines a complex symmetric inner product ( i.e. , no complex conjugation ) .
the evolution operator can be defined implicitly by @xmath68 where the set of eigenvectors @xmath69 is associated with an arbitrary orthonormal basis set and the eigenvalues of @xmath70 are @xmath71 ( or equivalently the eigenvalues of @xmath72 are @xmath73 ) .
inserting eq .
[ eq : u ] into eq .
[ eq : neuhc_n ] we obtain eq .
[ eq : complexct ] with @xmath74 which also implicitly defines the `` initial state '' @xmath75 .
this construction establishes an equivalence between the problem of extracting spectral information from the signal with the one of diagonalizing the evolution operator @xmath76 ( or the hamiltonian @xmath72 ) of the fictitious underlying dynamical system .
the filter - diagonalization method is then used for extracting the eigenvalues of the hamiltonian @xmath72 in any chosen small energy window .
operationally this is done by solving a small generalized eigenvalue problem whose eigenvalues yield the frequencies in a chosen window .
the knowledge of the operator @xmath72 itself is not required , as for a properly chosen basis the matrix elements of @xmath72 can be expressed only in terms of @xmath65 .
the advantage of the filter - diagonalization procedure is its numerical stability with respect to both the length and complexity ( the number and density of the contributing frequencies ) of the signal . here
we apply the method of ref .
@xcite which is an improvement of the filter - diagonalization method of ref .
@xcite in that it allows to significantly reduce the required length of the signal by implementing a different fourier - type basis with an efficient rectangular filter .
such a basis is defined by choosing a small set of values @xmath77 in the frequency interval of interest , @xmath78 , and the maximum order , @xmath79 , of the krylov vectors , @xmath80 , used in the fourier series , @xmath81 it is convenient to introduce the notations , @xmath82 for the matrix elements of the operator @xmath83 , and @xmath84 , for the corresponding small @xmath85 complex symmetric matrix .
as such @xmath86 denotes the matrix representation of the operator @xmath70 itself and @xmath87 , the overlap matrix with elements @xmath88 , which is required as the vectors @xmath89 are not generally orthonormal . now using these definitions we can set up a generalized eigenvalue problem , @xmath90 for the eigenvalues @xmath91 of the operator @xmath83 .
the column vectors @xmath92 with elements @xmath93 define the eigenvectors @xmath94 in terms of the basis functions @xmath95 as @xmath96 assuming that the @xmath95 s form a locally complete basis .
the matrix elements ( [ eq : up ] ) can be expressed in terms of the signal @xmath65 , the explicit knowledge of the auxiliary objects @xmath72 , @xmath94 or @xmath75 is not needed .
indeed , insertion of eq .
[ eq : psi_j ] into eq .
[ eq : up ] , use of the symmetry property , @xmath97 , and the definition of @xmath65 , eq . [ eq : neuhc_n ] , gives after some arithmetics @xmath98,\ \varphi\neq\varphi'\ , \nonumber \\
u^{(p ) } ( \varphi,\varphi ) & = & \sum_{n=0}^{2m}(m-|m - n|+1)e^{in\varphi}c_{n+p}.\nonumber\end{aligned}\ ] ] ( note that the evaluation of * u*@xmath99 requires knowledge of @xmath65 for @xmath100 . )
the solution of the generalized eigenvalue problem ( [ eq : generalized ] ) is usually done by a singular value decomposition of the matrix * u*@xmath101 .
each value of @xmath30 yields a set of frequencies @xmath9 and , due to eqs .
[ eq : dk_def ] , [ eq : psi_j ] and [ eq : bk ] , amplitudes , @xmath102 note that eq .
[ eq : dk ] is a functional of the half signal @xmath103 . even though in all our applications eq .
[ eq : dk ] yields very good results , here we present an even better expression for the coefficients @xmath43 ( see ref .
@xcite ) , @xmath104 ^ 2\nonumber\\ & \equiv & \left[{1\over m+1}\sum_{j=1}^{j}b_{jk}u^{(0 ) } ( \varphi_j , w_k)\right]^2 \label{eq : bestdk}\end{aligned}\ ] ] with @xmath105 defined by eq .
[ eq : upjj ] .
[ eq : bestdk ] is a functional of the whole available signal @xmath106 .
the converged @xmath9 and @xmath43 should not depend on @xmath30 .
this condition allows us to identify spurious or non - converged frequencies by comparing the results with different values of @xmath30 ( e.g. , with @xmath107 and @xmath108 ) .
we can define the simplest error estimate @xmath109 as the difference between the frequencies @xmath9 obtained from diagonalizations with @xmath107 and @xmath108 , i.e.@xmath110
we have introduced in section 2 the method for periodic orbit quantization by harmonic inversion for the example of the riemann zeta function because this model allows a direct check of the precision of our method . in this section
we present and discuss the numerical results obtained for the riemann zeros .
we also discuss the amount of periodic orbit input data , viz . here the prime numbers , required to obtain converged semiclassical eigenenergies and riemann zeros , respectively . for a numerical demonstration
we construct the signal @xmath56 using eqs .
[ c_nc ] and [ c_sigma ] in the region @xmath111 from the first 78498 prime numbers and with a gaussian smoothing width @xmath112 .
parts of the signal are presented in fig
. 1 . up to @xmath113 the gaussian approximations to the @xmath114-functions do essentially not overlap ( see fig .
1a ) whereas for @xmath115 the mean spacing @xmath116 between successive @xmath114-functions becomes much less than the gaussian width @xmath112 and the signal fluctuates around the mean @xmath117 ( see fig .
1b ) . from this signal
we were able to calculate about 2600 riemann zeros to at least 12 digit precision .
for the small generalized eigenvalue problem ( [ eq : generalized ] ) we used matrices with dimension @xmath118 .
some riemann zeros @xmath9 , the corresponding amplitudes @xmath43 , and the estimated errors @xmath109 ( see eq .
( [ eps_def ] ) ) are given in table 1 .
within the numerical error the riemann zeros are real and the amplitudes are consistent with @xmath52 for non - degenerate zeros . to fully appreciate the accuracy of our harmonic inversion technique we note that zeros obtained from the principal sum of the riemann - siegel formula ( [ riemann_siegel ] ) deviate by about 1 to 15 percent of the mean spacing from the exact zeros . including the first correction term in ( [ riemann_siegel ] ) the approximations to the first five zeros read @xmath119 , @xmath120 , @xmath121 , @xmath122 , and @xmath123 , which still significantly deviates from the exact values ( see table 1 ) .
considering even higher order correction terms the results will certainly converge to the exact zeros .
however , the generalization of such higher order corrections to ergodic dynamical systems is a nontrivial task and requires , e.g. , the knowledge of the terms in the weyl series , i.e. , the mean staircase function after the constant @xcite .
the perfect agreement of our results for the @xmath9 with the exact riemann zeros to full numerical precision is remarkable and clearly demonstrates that harmonic inversion by filter - diagonalization is a very powerful and accurate technique for the analytic continuation and the extraction of poles of a non - convergent series such as eq .
a few @xmath9 have been obtained ( see table 2 ) which are definitely not located on the real axis . except for the first at @xmath124 they can be identified with the trivial real zeros of the zeta function at @xmath125 ; @xmath126 in contrast to the nontrivial zeros with real @xmath9 , the numerical accuracy for the trivial zeros decreases rapidly with increasing @xmath127 .
the trivial zeros @xmath128 are the analogue of resonances in open physical systems with widths increasing with @xmath127 .
the fact that the trivial riemann zeros are obtained emphasizes the general applicability of our method and demonstrates that periodic orbit quantization by harmonic inversion can be applied not only to closed but to open systems as well .
the decrease of the numerical accuracy for very broad resonances is a natural numerical consequence of the harmonic inversion procedure @xcite .
the value @xmath124 in table 2 is special because in this case the amplitude is negative , i.e. , @xmath129 .
writing the zeta function in the form @xcite @xmath130 where @xmath131 is a constant and @xmath132 a regularizing function which ensures convergence of the product , integer values @xmath43 are the multiplicities of _ zeros_. therefore it is reasonable to relate negative integer values with the multiplicities of _ poles_. in fact , @xmath133 has a simple pole at @xmath134 consistent with @xmath124 in table 2 .
we have calculated riemann zeros by harmonic inversion of the signal @xmath56 ( eq . [ c_sigma ] ) which uses prime numbers as input .
the question arises what are the requirements on the signal @xmath56 , in particular what is the required signal length . in other words , how many riemann zeros ( or semiclassical eigenenergies ) can be converged for a given set of prime numbers ( or periodic orbits , respectively ) .
the answer can be directly obtained from the requirements on the harmonic inversion technique .
in general , the required signal length @xmath135 for harmonic inversion is related to the average density of frequencies @xmath136 by @xcite @xmath137 from eq .
[ s_max ] the required number of primes ( or periodic orbits ) can be directly estimated as @xmath138 or @xmath139 .
for the special example of the riemann zeta function the required number of primes to have a given number of riemann zeros converged can be estimated analytically . with the average density of riemann zeros derived from ( [ n_bar ] ) , @xmath140 we obtain @xmath141
the number of primes with @xmath142 can be estimated from the prime number theorem @xmath143 on the other hand the number of riemann zeros as a function of @xmath34 is given by eq .
( [ n_bar ] ) . the estimated number of riemann zeros which can be obtained by harmonic inversion from a given set of primes is presented in fig . 2 .
for example , about 80 zeros @xmath144 can be extracted from the short signal @xmath56 with @xmath145 ( 168 prime numbers ) in agreement with the estimates given above .
obviously , in the special case of the riemann zeta function the efficiency of our method can not compete with that of the riemann - siegel formula method ( [ riemann_siegel ] ) where the number of terms is given by @xmath146 $ ] and , e.g. , 5 terms in eq.[riemann_siegel ] would be sufficient to calculate good approximations to the riemann zeros in the region @xmath147 .
our primary intention is to introduce harmonic inversion by way of example of the zeros of the riemann zeta function as a _
tool for periodic orbit quantization , and not to use it as an alternative method for solving the problem of finding most efficiently zeros of the riemann zeta function .
a functional equation can only be invoked for the semiclassical quantization of _ bound _ and _ ergodic _ systems . in this case
the required number of periodic orbits can be estimated from the condition @xmath148 @xcite , which differs by a factor of 4 from the required signal length ( [ s_max ] ) for harmonic inversion .
periodic orbit quantization by harmonic inversion will be of particular advantage in situations where special properties such as a functional equation can not be invoked , e.g. , for bound systems with non - ergodic , i.e. , regular or mixed classical dynamics , and for open ( scattering ) systems .
we conclude this section with a remark on the famous riemann hypotheses mentioned in section 2.1 . applying our method of harmonic inversion to the signal @xmath56 ( eq . [ c_sigma ] ) the riemann hypotheses for the _ zeros _ of the zeta function , @xmath149 , is directly related to an equivalent statement for the _ eigenvalues _
@xmath150 of the operator @xmath151 , i.e. , the generalized eigenvalue problem ( [ eq : generalized ] ) .
speculations that the operator @xmath152 can be regarded as the hamiltonian of a quantum mechanical system have been presented by berry @xcite .
unfortunately , @xmath152 is not known as it is only defined implicitly by way of its matrix representation ( [ eq : upjj ] ) , which is a linear functional of the signal ( [ c_sigma ] )
. however , the very fact that the riemann zeros are obtained as eigenvalues of some matrix with analytically known coefficients is already intriguing .
as mentioned in section 2 the basic equation ( [ g_nc ] ) used for the calculation of riemann zeros has the same mathematical form as gutzwiller s semiclassical trace formula . both series , eq .
[ g_nc ] and the periodic orbit sum ( [ rho_po ] ) , suffer from similar convergence problems in that they are absolutely convergent only in the complex half - plane outside the region where the riemann zeros , or quantum eigenvalues , respectively , are located . as a consequence , in a direct summation of periodic orbit contributions
smoothing techniques must be applied resulting in low resolution spectra for the density of states @xcite . to extract individual eigenstates the semiclassical trace formula has to be analytically continued to the region of the quantum poles . here
dynamical zeta functions have turned out to be of particular interest . for _ bound _ and
_ ergodic _ systems one technique is to apply an approximate functional equation and generalize the riemann - siegel formula ( [ riemann_siegel ] ) to dynamical zeta functions @xcite .
the riemann - siegel look - alike formula has been applied , e.g. , for the semiclassical quantization of the hyperbola billiard @xcite . for bound ergodic systems alternative semiclassical quantization conditions based on a semiclassical representation of the spectral staircase @xmath153 @xcite and derived from a quantum version of a classical poincar map @xcite
have also been discussed .
these quantization techniques can not be applied to _ open _ systems . however , if a symbolic dynamics for the system exists , i.e. , if the periodic orbits can be classified with the help of a complete symbolic code , the dynamical zeta function , given as an infinite euler product over entries from classical periodic orbits can be expanded in terms of the cycle length of the orbits @xcite . the _ cycle expansion _
series is rapidly convergent if the contributions of long orbits are approximately shadowed by contributions of short orbits .
the cycle expansion technique has been applied , e.g. , to the three disk scattering system @xcite , the three body coulomb system @xcite , and to the hydrogen atom in a magnetic field @xcite .
a combination of the cycle - expansion method with a functional equation has been applied to bound systems in @xcite . however , the existence of a complete symbolic dynamics is more the exception than the rule , and the cycle expansion can not be applied , in particular for systems with mixed regular - chaotic classical dynamics . in this section
we apply the same technique that we used for the calculation of riemann zeros , to the calculation of semiclassical eigenenergies and resonances of physical systems by harmonic inversion of gutzwiller s periodic orbit sum for the propagator .
the method only requires the knowledge of all orbits up to a sufficiently long but finite period and does not rely on either an approximate semiclassical functional equation , nor does it depend on the existence of a symbolic code for the orbits .
the method will therefore allow the investigation of a large variety of systems with an underlying chaotic , mixed , or even regular classical dynamics .
the derivation of an expression for the recurrence function to be harmonically inverted is analogous to that in section 2.2 . following gutzwiller @xcite the semiclassical response function for chaotic systems
is given by @xmath154 where @xmath155 is a smooth function and the @xmath29 and @xmath28 are the classical actions and weights ( including phase information given by the maslov index ) of periodic orbit contributions .
( [ g_sc ] ) is also valid for integrable @xcite and near - integrable @xcite systems but with different expressions for the amplitudes @xmath28 .
it should also be possible to include complex `` ghost '' orbits @xcite and uniform semiclassical approximations @xcite close to bifurcations of periodic orbits in the semiclassical response function ( [ g_sc ] ) .
the eigenenergies and resonances are the poles of the response function but , unfortunately , its semiclassical approximation ( [ g_sc ] ) does not converge in the region of the poles , whence the problem is the analytic continuation of @xmath156 to this region . in the following
we make the ( weak ) assumption that the classical system has a scaling property , i.e. , the shape of periodic orbits does not depend on the scaling parameter , @xmath34 , and the classical action scales as @xmath157 examples of scaling systems are billiards @xcite , hamiltonians with homogeneous potentials @xcite , coulomb systems @xcite , or the hydrogen atom in external magnetic and electric fields @xcite .
eq . [ s_po ]
can even be applied for non - scaling , e.g. , molecular systems if a generalized scaling parameter @xmath158 is introduced as a new dynamical variable @xcite .
quantization yields bound states or resonances , @xmath9 , for the scaling parameter . in scaling systems the semiclassical response function @xmath159 can be fourier transformed easily to obtain the semiclassical trace of the propagator @xmath160 the signal @xmath161 has @xmath114-peaks at the positions of the classical periods ( scaled actions ) @xmath162 of periodic orbits and with peak heights ( recurrence strengths ) @xmath28 , i.e. , @xmath161 is gutzwiller s periodic orbit recurrence function .
consider now the quantum mechanical counterparts of @xmath159 and @xmath163 taken as the sums over the poles @xmath9 of the green s function , @xmath164 @xmath165 with @xmath43 being the multiplicities of resonances , i.e. , @xmath52 for non - degenerate states .
in analogy with the calculation of riemann zeros from eq .
( [ c_sigma ] ) the frequencies , @xmath9 , and amplitudes , @xmath43 , can now be extracted by harmonic inversion of the signal @xmath161 after convoluting it with a gaussian function , i.e. , @xmath166 by adjusting @xmath167 to the functional form of eq .
[ c_qm ] , the frequencies , @xmath9 , can be interpreted as the semiclassical approximation to the poles of the green s function in ( [ g_qm ] ) .
note that the harmonic inversion method described in section 3 allows studying signals with complex frequencies @xmath9 as well .
for open systems the complex frequencies can be interpreted as semiclassical resonances .
note also that the @xmath9 in general differ from the exact quantum eigenvalues because gutzwiller s trace formula ( [ g_sc ] ) is an approximation , correct only to first order in @xmath10 .
therefore the diagonalization of small matrices in ( [ eq : generalized ] ) does not imply that the results of periodic orbit quantization are more `` quantum '' in any sense than those obtained , e.g. , from a cycle expansion @xcite .
the eigenvalues are solutions of non - linear equations and the diagonalization is equivalent to the search of zeros of the dynamical zeta function in the cycle expansion technique .
numerical calculation of the zeros is also a non - linear problem and , in contrast to the matrix diagonalization , might encounter a problem of missing roots .
the procedure described above can be generalized in a straightforward manner to the calculation of semiclassical diagonal matrix elements @xmath168 of a smooth hermitian operator @xmath169 . in this case
we start from the quantum mechanical trace formula @xcite @xmath170 which has the same functional form as ( [ g_qm ] ) , but with @xmath171 instead of @xmath52 . for the quantum response function @xmath172 ( eq .
[ ga_qm ] ) a semiclassical approximation has been derived in @xcite , which has the same form as gutzwiller s trace formula ( [ g_sc ] ) but with amplitudes @xmath173 where @xmath174 is the monodromy matrix and @xmath175 the maslov index of the periodic orbit , and @xmath176 is the classical average of the observable @xmath132 over _ one _ period @xmath177 of the _ primitive _ periodic orbit . note that @xmath178 and @xmath179 are functions of the classical action instead of time for scaling systems @xcite .
gutzwiller s trace formula for the density of states is obtained with @xmath169 being the identity operator , i.e. , @xmath180 . when the semiclassical signal @xmath161 ( eq . [ c_sc ] ) with amplitudes @xmath28 given by eqs .
[ a_po , a ] and [ a_cl ] is analyzed with the method of harmonic inversion the frequencies and amplitudes obtained are the semiclassical approximations to the eigenvalues @xmath9 and matrix elements @xmath171 , respectively .
let us consider a billiard system consisting of three identical hard disks with unit radii , @xmath181 , displaced from each other by the same distance @xmath182 .
this simple , albeit nontrivial , scattering system has served as a model for periodic orbit quantization in many investigations in recent years @xcite . after symmetry reduction
the periodic orbits can be classified by a binary symbolic code @xcite .
for @xmath183 there is a one - to - one identity between the periodic orbits and the symbolic code , whereas for @xmath184 pruning of orbits sets in . for @xmath185 semiclassical resonances
were calculated by application of the cycle expansion technique including all periodic orbits up to cycle length @xmath186 @xcite . in order to demonstrate the usefulness of the harmonic inversion technique
we first apply it to the case @xmath187 studied before .
note that the ratio corresponds to the very favorable regime for the cycle expansion ( see below ) . in billiards
the scaled action @xmath46 is given by the length @xmath188 of orbits @xmath189 and the quantized parameter is the absolute value of the wave vector @xmath190 .
3a shows the periodic orbit recurrence function , i.e. , the trace of the semiclassical propagator @xmath191 .
the groups with oscillating sign belong to periodic orbits with adjacent cycle lengths . to obtain a smooth function on an equidistant grid , which is required for the harmonic inversion method ,
the @xmath114-functions in ( [ c_sc ] ) have been convoluted with a gaussian function of width @xmath192 .
as explained in section 2 this does not change the underlying spectrum .
the results of the harmonic inversion analysis of this signal are presented in fig .
3b and table 3 . the crosses in fig .
3b represent semiclassical poles , for which the amplitudes @xmath43 are very close to @xmath193 , mostly within one percent . because the amplitudes converge much slower than the frequencies
these resonance positions can be assumed to be very accurate within the semiclassical approximation .
in fact , a perfect agreement to many significant figures is achieved for these poles with the results obtained by cycle expansion @xcite , and this agreement confirms that the results in fig .
3b and table 3 are the true semiclassical resonances , i.e. , deviations from the exact quantum poles are solely due to the semiclassical approximation in gutzwiller s trace formula . for some broad resonances marked by diamonds in fig .
3b and table 3 the @xmath43 deviate strongly from @xmath193 , within 5 to maximal 50 percent .
it is not clear whether these strong deviations are due to numerical effects , such as convergence problems caused by too short a signal , or if they are a direct consequence of the semiclassical approximation .
of course , in the exact expression ( [ c_qm ] ) all multiplicities @xmath43 are 1 , but there is no proof that this is still true within the semiclassical approximation . however , for the lowest @xmath194 eigenvalues ( see , e.g. , the first three resonances in table 3 ) , where the agreement with the exact resonance energies is worst @xcite @xmath52 still holds , indicating that there is no @xmath10-dependence for the multiplicities .
the cycle expansion technique is based on the idea that the contributions of long periodic orbits are shadowed by those of short orbits .
for the three disk scattering system this is ideally fulfilled in the limit of a large ratio @xmath195 .
this is no longer true for short distances between the disks and hence the convergence of the conventional cycle expansion becomes rather slow @xcite . as a second example of periodic orbit quantization by harmonic inversion we therefore study the three disk scattering system with a short distance ratio @xmath196 , and thus in a situation where the assumption of the cycle expansion that contributions of long orbits are shadowed by short orbits is no longer a good approximation .
the results are presented in fig . 4 and table 4 .
for large @xmath188 groups of orbits with the same cycle length of the symbolic code strongly overlap and can not be recognized in fig .
. the signal is obtained from periodic orbits with cycle length @xmath197 .
note that only 356 periodic orbits with @xmath198 are included in the signal ( fig .
4a ) whereas the complete set of orbits with cycle length @xmath197 consists of 1377 orbits .
the resonances obtained by harmonic inversion of the semiclassical recurrence function are in good agreement with results of the cycle expansion @xcite .
we finally remark that when a symbolic dynamics exists and contributions of long orbits are shadowed by short orbits the cycle expansion techniques are very efficient and , e.g. , for the three disk scattering system with @xmath199 a large number of resonances can be obtained from just 8 periodic orbits with cycle length @xmath200 .
again , we do not aim at competing with the efficiency of the cycle expansion method in such ideal situations . as already mentioned in section 4
the advantage of periodic orbit quantization by harmonic inversion is its _ general _ applicability : it does not depend on the existence of a symbolic code and the shadowing of orbits .
we have extracted the semiclassical resonances of the three disk scattering problem directly from the periodic orbits with exactly the same method as the riemann zeros in section 4 from the primes as periodic orbits , for which no symbolic dynamics exists .
we have introduced harmonic inversion as a new and general tool for semiclassical periodic orbit quantization .
the method requires the complete set of periodic orbits up to a given maximum period as input but does not depend on special properties of the orbits , as , e.g. , the existence of a symbolic code or a functional equation .
we have demonstrated the wide applicability of the method by applying it to two systems with completely different properties , namely the zeros of the riemann zeta function and the three disk scattering problem .
both systems have been treated before by efficient methods , which , however , are restricted to bound ergodic systems or systems with a complete symbolic dynamics .
the harmonic inversion technique allows to solve both problems with one and the same method .
therefore the method can also serve as a tool for , e.g. , the semiclassical quantization of systems with mixed regular - chaotic classical dynamics , which still is a challenging and unsolved problem .
the signal @xmath161 can be composed as the sum of a signal related to the irregular part of the classical phase space with periodic orbit amplitudes given by gutzwiller s trace formula @xcite and a signal related to stable @xcite or nearly integrable @xcite torus structures .
it should also be possible to include , e.g. , creeping orbits @xcite , ghost orbit contributions @xcite , and higher order @xmath10 corrections @xcite into the signal @xmath161 , which can then be inverted to reveal the semiclassical poles .
the method can even be used for a semiclassical periodic orbit quantization of systems with non - homogeneous potentials such as the potential surfaces of molecules when a generalized scaling technique @xcite is applied .
we are grateful to b. eckhardt who kindly communicated to us his periodic orbits for the three disk scattering system .
j. m. thanks the alexander von humboldt - stiftung for a feodor - lynen scholarship and h. taylor and the university of southern california for their kind hospitality and support .
99 gutzwiller m c 1967 _ j. math . phys . _ * 8 * 1979 and 1971 _ j. math
* 12 * 343 gutzwiller m c 1990 , _ chaos in classical and quantum mechanics _
( new york : springer ) berry m v 1986 , riemann s zeta function : a model for quantum chaos ? _ quantum chaos and statistical nuclear physics _ , ed t h seligman and h nishioka ( _ lecture notes in physics _ * 263 * ) ( berlin : springer ) pp 1 - 17 berry m v and keating j p 1990 _ j. phys . _ a * 23 * 4839 keating j 1992 _ chaos _ * 2 * 15 berry m v and keating j p 1992 _ proc .
lond . _ * a 437 * 151 aurich r , matthies c , sieber m and steiner f 1992 , _ phys .
lett . _ * 68 * 1629 aurich r and bolte j 1992 _ mod .
* b 6 * 1691 bogomolny e b 1992 _ nonlinearity _ * 5 * 805 cvitanovi p and eckhardt b 1989 _ phys .
lett . _ * 63 * 823 artuso r , aurell e and cvitanovi p 1990 _ nonlinearity _ * 3 * 325 ; 361 eckhardt b and russberg g 1993 _ phys . rev .
_ e * 47 * 1578 eckhardt b , cvitanovi p , rosenqvist p , russberg g and scherer p 1995 pinball scattering , in _ quantum chaos _
, ed g casati and b v chirikov , ( cambridge : university press ) , pp 405 - 434 tanner g , scherer p , bogomolny e b , eckhardt b and wintgen d 1991 _ phys .
lett . _ * 67 * 2410 edwards h m 1974 _ riemann s zeta function _
( new york : academic press ) titchmarsh e c 1986 _ the theory of the riemann zeta - function _ oup , 2nd ed .
odlyzko a m 1990 _ the @xmath201th zero of the riemann zeta function and 70 million of its neighbours _ ( at&t bell laboratories ) bohigas o and giannoni m j 1984 chaotic motion and random - matrix theories _ mathematical and computational methods in nuclear physics _
ed j s dehesa j m g gomez and a polls ( _ lecture notes in physics _ * 209 * ) ( berlin : springer ) pp 1 - 99 bogomolny e and keating j 1995 _ nonlinearity _ * 8 * 1115 main j , mandelshtam v a and taylor h s 1997 _ phys .
_ * 79 * 825 marple s , jr . 1987 , _ digital spectral analysis with applications _ , prentice - hall , englewood cliffs wall m r and neuhauser d 1995 _ j. chem .
phys . _ * 102 * 8011 mandelshtam v a and taylor h s 1997 _ phys
* 78 * 3274 mandelshtam v a and taylor h s 1997 _ j. chem . phys . _ * 107 * 6756 main j , mandelshtam v a and taylor h s 1997 _ phys .
lett . _ * 78 * 4351 delsarte j 1966 _ j. anal . math .
( jerusalem ) _ * 17 * 419 aurich r , sieber m and steiner f 1988 _ phys .
_ * 61 * 483 sieber m and steiner f 1991 _ phys .
* 67 * 1941 keating j p and sieber m 1994 _ proc .
_ * a 447 * 413 wintgen d 1988 _ phys .
lett . _ * 61 * 1803 ezra g s , richter k , tanner g and wintgen d 1991 _ j. phys .
_ b * 24 * l413 wintgen d , richter k and tanner g 1992 _ chaos _ * 2 * 19 tanner g , hansen k t and main j 1996 _ nonlinearity _ * 9 * 1641 tanner g and wintgen d 1992 _ chaos _ * 2 * 53 berry m v and tabor m 1976 _ proc
* a349 * 101 tomsovic s , grinberg m and ullmo d 1995 _ phys .
* 75 * 4346 ullmo d , grinberg m and tomsovic s 1996 _ phys .
_ e * 54 * 136 ku m , haake f and delande d 1993 _ phys .
_ * 71 * 2167 main j and wunner g 1997 _ phys .
a _ * 55 * 1743 ozorio de almeida
a m and hannay j h 1987 _ j. phys .
* 20 * 5873 main j and wunner g 1998 _ phys .
e _ * 57 * in press heller e j 1984 _ phys .
lett . _ * 53 * 1515 martens c c , waterland r l and reinhardt w p 1989 _ j. chem .
phys . _ * 90 * 2328 tomsovic s 1991 _ j. phys . _ a * 24 * l733 main j , wiebusch g , welge k h , shaw j and delos j b 1994 _ phys . rev . _ a * 49 * 847 main j , jung c and taylor h s 1997 _ j. chem
. phys . _ * 107 * 6577 eckhardt b , fishman s , mller k and wintgen d 1992 _ phys .
a _ * 45 * 3531 boos d , main j , mehlig b and mller k 1995 _ europhys .
* 32 * 295 gaspard p and rice s a 1989 _ j. chem .
phys . _ * 90 * 2225 , 2242 , and 2255 wirzba a , private communication wirzba a 1992 _ chaos _ * 2 * 77 gaspard p and alonso d 1993 _ phys .
rev . _ a * 47 * r3468
.[table1 ] non - trivial zeros @xmath9 , multiplicities @xmath43 , and error estimate @xmath109 for the riemann zeta function . [ cols= "
> , > , > , > , > , > " , ]
|
in semiclassical theories for chaotic systems such as gutzwiller s periodic orbit theory the energy eigenvalues and resonances are obtained as poles of a non - convergent series @xmath0 .
we present a general method for the analytic continuation of such a non - convergent series by harmonic inversion of the `` time '' signal , which is the fourier transform of @xmath1 .
we demonstrate the general applicability and accuracy of the method on two different systems with completely different properties : the riemann zeta function and the three disk scattering system .
the riemann zeta function serves as a mathematical model for a bound system .
we demonstrate that the method of harmonic inversion by filter - diagonalization yields several thousand zeros of the zeta function to about 12 digit precision as eigenvalues of small matrices . however , the method is not restricted to bound and ergodic systems , and does not require the knowledge of the mean staircase function , i.e. , the weyl term in dynamical systems , which is a prerequisite in many semiclassical quantization conditions .
it can therefore be applied to open systems as well .
we demonstrate this on the three disk scattering system , as a physical example .
the general applicability of the method is emphasized by the fact that one does not have to resort a symbolic dynamics , which is , in turn , the basic requirement for the application of cycle expansion techniques .
|
once a conductor forms contact with an insulator , an energy barrier is formed between the two materials , which impedes the charge - carrier injection into the insulator .
although this injection barrier in general sways charge transport through the conductor / insulator system , only the two limiting cases of very low or very high injection barriers are often considered . for low injection barriers ,
one expects the contact to be ohmic , meaning that the contact is able to supply more charges per unit time than the bulk of the insulator can support . in this case , a space - charge region is formed and the electric field at the interface vanishes @xcite . because excess charge - carriers dominate charge transport in insulators , one observes a space - charge - limited current ( sclc ) density of the form @xmath0 ( in the absence of charge - carrier traps ) , where @xmath1 is the sample thickness and @xmath2 is the applied voltage .
the current - voltage characteristic ( iv - characteristic ) is determined by the bulk properties of the material with no influence of the contact properties @xcite . for high injection barriers ,
one anticipates the injection rate across the conductor / insulator interface to dominate the iv - characteristic of the system .
the models to describe injection are the fowler - nordheim ( fn ) tunneling model @xcite or the richardson - schottky ( rs ) model @xcite for thermionic injection .
the fn model describes the charge - carrier injection by tunneling through a triangular barrier into an unbound continuum of states .
the current takes then the temperature independent form : @xmath3 , \label{fowler}\ ] ] where @xmath4 is the elementary charge , @xmath5 being the effective mass of a charge - carrier , @xmath6 is planck s constant , @xmath7 is a constant and @xmath8 is the height of the barrier .
the rs model on the other hand describes charge injection as a thermally activated hopping over the potential barrier , where barrier lowering due to the superposition of the external electrostatic potential and the image - charge potential is considered .
the rs model predicts the iv - characteristics to follow : @xmath9}\exp { \left [ \frac{1}{kt}\sqrt{\frac{ev}{4\pi\epsilon\epsilon_0 l } } \right ] } .
\label{richardson}\ ] ] here , @xmath10 is the temperature , @xmath11 is the boltzmann constant , @xmath12 represents the effective richardson constant , @xmath13 is the relative permittivity of the insulator and @xmath14 is the permittivity of vacuum . an alternative description of the injection process is given by the drift - diffusion theory involving electron - electron interaction in a mean - field approximation @xcite . in this homogeneous continuum model widely used for the description of conventional crystalline semiconductors ,
the steady state current density exemplarily given for holes in one dimension is : @xmath15 where @xmath16 is the hole mobility , @xmath17 is the hole density , @xmath18 is the electric field in the insulator , and @xmath19 is the diffusivity of holes . @xmath18 and @xmath17 are coupled by the poisson equation .
the drift - diffusion equation in combination with the poisson equation involves space - charge effects , but meets the problem of self - consistency in the boundary conditions .
the electrostatic potential generated by the injected charge - carriers modifies the injection barrier , but on the other hand the amount of charge - carriers injected per unit time depends strongly on the barrier height .
therefore , a proper description of the injection process by means of the drift - diffusion theory has to involve both , the insulator and the conductor side of the system .
this can be seen in analogy to a strongly asymmetric _ pn_-junction .
different attempts to numerically describe the injection process in insulators with account of the space charge were made recently @xcite .
in transitive fe - simulations in one- and two - dimensional geometries by christen et al .
@xcite both injection - limited and space charge - limited regimes were presented . however , the injection description was still not self - consistent imposing some phenomenological values for the field and the carrier density at the contact .
a comprehensive one - dimensional numerical model accounting for space - charge effect was developed by tuti et al .
@xcite which comprised the hopping transport and the tunnel injection from electrodes . unfortunately , this sophisticated numerical tool does not allow analytical fitting of current - voltage characteristics which gives insight in major mechanisms controlling injection . in the work of neumann et al .
@xcite , a self - consistent numerical treatment of the injection and transport processes in a conductor / insulator / conductor device was presented in which continuity of the electrochemical potential and the electric displacement was assumed everywhere in the system , in particular at the contacts .
the conductor and the insulator were characterized by their specific density of state ( dos ) distributions . in present work
we obtain an exact analytical solution for the injection across the conductor / insulator interface in a self - consistent manner . to focus on the consequences of the self - consistent treatment , we discuss our results on the basis of a simplified indium tin oxide / organic semiconductor system used as hole - injecting contact in organic optoelectronic devices , shifting the comparison with experimental data to a future time .
let us consider a single conductor / insulator interface located at a position @xmath20 .
the conductor is supposed to extend over the semi - space with @xmath21 , whereas the insulator covers the semi - space with @xmath22 .
all energies are measured downwards with respect to the top of the valence band in the conductor , in order to account for hole transport in the system . in the following sections ,
the theoretical models describing the insulator and the conductor are introduced .
the conductor electrode is characterized by its dos as a function of the energy @xmath23 . for our purpose
it is sufficient to assume the free electron approximation , in which the dos function reads : @xmath24 where @xmath5 is the effective mass in the conductor . in the thomas - fermi approximation @xcite
, one can deduce the electrochemical potential @xmath25 of the conductor as a function of the spatial coordinate @xmath26 , @xmath27 where @xmath28 is the hole density in the electrode and @xmath29 is the electrostatic potential . in general
, the electrochemical potential @xmath30 relates the steady state current density @xmath31 with the charge - carrier density . for a one - dimensional geometry ,
the current remains constant across the whole space and @xmath31 is given by the conductivity @xmath32 and the derivative of @xmath30 @xcite , @xmath33 ( notice the direction of the energy axis ) .
the conductivity of a conductor @xmath34 can be expressed in terms of the hole mobility @xmath35 and the hole density @xmath36 in the valence band at an infinite distance from the conductor / insulator interface . since charge - carriers are transferred from the electrode to the insulator
, a space - charge region emerges near the interface which modifies the electric field @xmath37 in the conductor according to gauss law , @xmath38 where @xmath39 is the relative permittivity of the electrode . in eq.([gaussmetal ] ) @xmath40 is the excess hole density .
however , charge - carrier densities in degenerate conductors are rather high and consequently , the value for the excess hole density is small in comparison with the background hole density , @xmath41 . hence , the well - known linearized thomas - fermi approximation @xcite can be applied , leading with eqs.([kappametal]-[gaussmetal ] ) to a differential equation for @xmath42 : @xmath43 with @xmath44 being the thomas - fermi screening length , defining the typical length scale of the system .
here @xmath45 is the electrochemical potential at an infinite distance from the conductor / insulator interface .
since space - charge zones in conductors are of finite thickness , gradients of @xmath37 have to vanish at an infinite distance from the contact .
therefore , the solution for the electric field reads @xmath46 e^{x / l_{tf } } + \frac{j}{\sigma_{c } } , \label{solutionmetal}\ ] ] where the electric field in the conductor at the conductor / insulator interface , @xmath47 , is the only unknown quantity .
the validity of the used approximations has to be reviewed critically .
an exact quantum - mechanical theory of the inhomogeneous electron gas accounting for the ionic lattice of the material and electron correlations , gives a comparable scale for the electric field penetration in conductors and demonstrates the usability of the uniform positive background model for simple metals @xcite .
while in typical metals the thomas - fermi screening length is about @xmath48 making the application of the quasi - classical thomas - fermi approximation , eq.([kappametal ] ) , inappropriate , substantially larger thomas - fermi screening lengths can be found in degenerate semiconductors such as transparent conducting oxides used in optoelectronic devices @xcite .
similarly , the insulator is characterized by a dos function , @xmath49 , describing extended states in which charge transport takes place .
in contrast to the conductor the electrochemical potential , @xmath50 , is situated well above the top of the band of extended hole states . in other words ,
the insulator is supposed to be non - degenerate . introducing a band - edge
means that the dos function @xmath51 when @xmath52 .
hence , the density of holes in extended states can be calculated using boltzmann statistics : @xmath53 where the energy scale has been adjusted to the top of the valence band in the conductor .
the injection barrier @xmath8 is defined as the energetic difference of the top of the extended state distribution , @xmath49 , to the electrochemical potential in the conductor at an infinite distance from the interface , @xmath45 . the electrochemical potential of the insulator is then given by @xmath54 where the quantity @xmath55 can be understood as the effective total density of states available in the insulator at a given temperature .
we note that the temperature dependence of @xmath56 becomes weak in the case of a narrow - band insulator .
since band - gap energies are much larger than @xmath57 , thermal excitation of a charge carrier from the valence band to the conduction band of a typical insulator is virtually impossible .
charge - carriers contributing to the electrical current are therefore excess charge - carriers injected from the conductor and their total density has to appear in gauss law , @xmath58 with @xmath59 the relative dielectric permittivity of the insulator .
eqs.([currentgeneral]),([kappaorganic ] ) and ( [ gaussorganic ] ) with @xmath60 lead to a nonlinear differential equation for the electric field @xmath61 , @xmath62 where @xmath16 is the hole mobility in the insulator . the same result is obtained by employing the drift - diffusion model , eq.([drift - diffusion ] ) , and the einstein relation , relating @xmath16 and @xmath19 @xcite . introducing the following dimensionless quantities , @xmath63 eq.([dglorganic ] ) converts into a dimensionless form , @xmath64 the solution of eq .
( [ dglorgdimless ] ) must be separately formulated for the two cases of thermal equilibrium and an applied steady - state current . in equilibrium ,
the dimensionless current density @xmath65 vanishes and eq.([dglorgdimless ] ) can be integrated in elementary functions .
the first integration results in : @xmath66 with an arbitrary constant @xmath67 . at an infinite distance from the contact the field and its derivative ( i.e. the charge carrier density ) vanish , thus , @xmath68 and a solution for the electric field reads , @xmath69 in eq .
( [ orgequiinf ] ) the field at the contact in the insulator , @xmath70 , is the only unknown quantity . considering a net current density , a general solution of eq.([dglorgdimless ] )
is known in terms of airy functions @xmath71 and @xmath72 @xcite : @xmath73 + c_2 \mathrm{bi}^{\prime}\left[(\iota/2)^{1/3}(x+c_1)\right ] } { \mathrm{ai}\left[(\iota/2)^{1/3}(x+c_1)\right]+ c_2 \mathrm{bi}\left[(\iota/2)^{1/3}(x+c_1)\right ] } , \label{solexact}\ ] ] where primes denote derivatives of airy functions with respect to their arguments , @xmath74 and @xmath75 are unknown constants . when @xmath76 the gradient of the charge carrier density has to vanish so that , @xmath77 eq.([bcinfinity ] ) demonstrates that , in presence of a constant current , the magnitude of the electric field @xmath78 rises asymptotically since the charge carrier density @xmath79 vanishes .
considering the asymptotic behavior of the solution of eq .
( [ dglorgdimless ] ) it is convenient to account explicitly for the sign of the current density @xmath65 .
since we assume injection of holes from the left semi - space to the right one , @xmath65 together with @xmath31 must be positive .
this imposes asymptotes of the field and the hole density @xmath80 respectively , resembling the characteristic behavior for sclc @xcite .
taking into account asymptotic properties of airy functions @xcite it is easy to establish that , to satisfy the asymptotic conditions for the field , eqs .
( [ bcinfinity ] ) and ( [ asymptotics ] ) , the constant @xmath75 must equal zero . then , the general solution , eq .
( [ solexact ] ) , transforms to @xmath81 } { \mathrm{ai}\left[(\iota/2)^{1/3}(x+c_1)\right]}. \label{solspecial}\ ] ] the constant @xmath74 must be determined from the boundary conditions derived in the next section .
the equations of the electric field distributions in the conductor and in the insulator have to be solved self - consistently .
assuming steady - state , self - consistency is achieved by adjusting the integration constants ( i.e. @xmath70 , @xmath82 and @xmath74 ) with respect to the maxwell equations and continuity of the electrochemical potential : @xmath83 eq.([bcgeneralfield ] ) requires that the electric displacement in the insulator at the interface , @xmath84 , is equal to @xmath85 .
this holds as long as no interface charges or dipole - layer exist at the contact , being problematic in many systems considered here @xcite . yet , for simplicity and without loss of generality eq.([bcgeneralfield ] ) is assumed , leading with eq.([bcgeneralkappa ] ) and eq.([solutionmetal ] ) to a nontrivial boundary condition relating the electric field and its first derivative in the respective media at the contact . in the insulator
the following boundary condition holds : @xmath86 where the dimensionless constant @xmath87 is defined as @xmath88 this boundary condition contains parameters of both media , being specified in the bulk of the respective material .
in this section the solution of the injection problem using the presented model is discussed .
we distinguish between the equilibrium condition , where space charge zones are formed as a consequence of diffusive charge carrier transfer , and the case of a steady state current , where charge carriers are driven through the system by a time invariant external electric field . as an example for an insulator ,
an organic semiconductor can be considered .
organic semiconductors show many typical characteristics of insulators like relatively large band gaps up to @xmath89 ev and hence , the absence of intrinsic charge carriers .
however , it is well established that disordered organic semiconductors possess a gaussian dos @xcite compromising the applicability of eq.([nsboltzmann ] ) - organic semiconductors are degenerate systems , the tail states acting as charge carrier traps . as a consequence
, one has to distinguish between trap states and transport states and fermi statistics has to be considered . yet , for weak disorder , the gaussian width is small and in the limiting case of a vanishing disorder , charge - carrier trapping in tail states is negligible and boltzmann statistics is valid . in organic light
emitting diodes ( oleds ) or field effect transistors ( ofets ) organic semiconductors are contacted with metals like au , ca , al or transparent conducting oxides like indium tin oxide ( ito ) to allow for charge carrier injection in the otherwise charge - carrier free organic semiconductor . while in metals the thomas - fermi approximation is disputable due to the prevailed low screening length of @xmath48 , the characteristic scale @xmath90 varies in ito , depending on doping @xcite , from 2.4@xmath91 to a few @xmath92 .
therefore , we choose in our self - consistent consideration an ito electrode and assume that it can be described in terms of the thomas - fermi approximation .
ito is typically employed as anode in oleds @xcite , since it provides a decent conductivity and a sufficient high workfunction ( 5ev ) to allow for efficient hole injection while being transparent in the visible range of the optical spectrum to enable light outcoupling . due to the importance of charge carrier injection for the device performance of oleds and ofets ,
the description of the involved charge carrier injection process has been advanced in recent years considering surface recombination of charge carriers at the interface @xcite or stochastic hopping in a surface - barrier potential @xcite . even more suitable for organic semiconductors microscopic models
account for the mobility of electrons within and transfer between molecular strands and for the interaction of electrons with molecular vibrational modes @xcite .
however , all these models work within the single electron picture , so far it concerns the injection process , which means that interaction between injected electrons is not incorporated and , thus , space - charge effects on the injection are not properly taken into account . from now on it is assumed that the material specific quantities of the organic semiconductor and ito adopt the typical values given in table [ materialparameters ] .
._typical material parameters for an organic semiconductor and ito @xcite .
the parameters are deduced assuming @xmath93 .
@xmath94 is the electron mass and @xmath95 is determined by parameters of both materials . _ [ cols="^,^,^,^,^,^,^,^,^ " , ] thereby , the injection barrier @xmath8 is given by the energetic difference between @xmath45 and the band edge in the organic semiconductor and , thus , is determined by both media .
changing the value of the barrier height while leaving the electrode unchanged can therefore be understood as considering a different organic semiconductor .
specifying the material parameters for the insulator and the conductor leads to substantial consequences for the boundary condition given in eq.([bcssx0 ] ) .
@xmath65 is multiplied by a small factor , @xmath96 and hence , the boundary condition does not depend directly on @xmath65 in most practical cases .
neglecting @xmath97 , eq.([bcssx0 ] ) can be reformulated to @xmath98 , \label{nnull}\ ] ] @xmath99 being the density of injected holes at the contact . as a consequence ,
the dependence of @xmath70 on the current is only due to eq.([dglorgdimless ] ) .
apparently , there exists a barrier variation @xmath100 leading to the definition of an effective injection barrier @xmath101 of @xmath102 the modification of the injection barrier corresponds to the amount of energy a charge carrier gains ( or loses ) in the electric field at the electrode side of the interface .
this change , as is seen from eq.([deltaeffdim ] ) , may be positive or negative , since the space - charge region at the interface represents a potential barrier itself . using the analytical solution ( [ orgequiinf ] ) valid for thermal equilibrium
, the boundary condition ( [ bcssx0 ] ) can be simplified to a transcendental relation for @xmath70 : @xmath103 @xmath70 together with eq.([orgequiinf ] ) yields the solution of the electric field distribution in thermal equilibrium . while the charge carrier mobilities of the respective material determine the time needed to reach equilibrium they apparently do not influence the final equilibrium electric field distribution .
eq.([orgequiinf ] ) and eq.([bcequi ] ) are generally valid for non - degenerate systems . the solution for the dimensionless electric field is shown in fig.[infequid0 ] , where barrier - free injection ( @xmath104 ) from an ito electrode in an organic semiconductor is considered .
the solution can be interpreted as follows .
holes diffuse from the electrode into the organic semiconductor .
this results in a negative electric field causing a drift current opposite but equal in absolute value to the diffusion current , so that the net current is zero . as a consequence
, a space - charge zone is established in the electrode and in the organic semiconductor , which in total is neutral .
this space - charge region is very thin in the electrode but extends far into the organic semiconductor .
the weak decay of the electric field ( @xmath105 ) in the organic semiconductor is due to a missing charge - carrier background in the insulator . in units of @xmath106 in equilibrium ( @xmath107 ) and
barrier - free case ( @xmath104 ) as function of the coordinate @xmath108 in units of @xmath90 .
the inset shows the discontinuity of the electric field at the interface._,width=283,height=241 ] the distribution of the electric field changes when a non - vanishing injection barrier is introduced . due to an impeded charge carrier injection , the electric field
is solely reduced close to the interface , leaving the field far in the organic material invariant .
depending on the material parameters chosen , the parameter @xmath87 , eq.([alpha ] ) , may be very large or very small ( here : @xmath95 ) . in these limits the nonlinear equation ( [ bcequi ] ) can be solved approximately .
if @xmath109 , two characteristic regions arise depending on the relation between @xmath8 and @xmath110 ( here : @xmath111 ) .
if the barrier height is so small that @xmath112 , @xmath70 is given by : @xmath113 , \label{fnullappr1}\ ] ] with @xmath114 .
if the barrier is so large that @xmath115 then @xmath116 the latter relation is also valid for @xmath117 and arbitrary @xmath8s .
eq.([fnullappr2 ] ) demonstrates the exponential suppression of the electric field within the space - charge region by the injection barrier , since less charge carriers are transferred across the contact .
the integration of the electric field leads to a voltage @xmath118 to maintain equilibrium .
the equilibrium voltage @xmath118 is often referred to as the contact potential @xcite . since the equilibrium field decreases with @xmath119 , the integration over the entire infinite semi - space diverges so that a cut - off length @xmath1 has to be introduced .
this results from the fact that a perfect insulator excluding any intrinsic charge carriers is considered .
the analytical expression for @xmath118 reads , @xmath120 , \label{vequi}\ ] ] where @xmath70 results from the boundary condition in equilibrium , eq .
( [ bcequi ] ) .
the existence of such an equilibrium voltage has practical consequences for organic electronic devices . in organic photovoltaic cells ,
the contact potential is known to reduce the open - circuit voltage considerably once the barrier height is small . in bulk- heterojunction solar cells based on [ 6,6]-phenyl c@xmath121-butyric acid methyl ester ( pcbm ) as electron acceptor and poly[2-methoxy-5-(3,7-dimethyloctyloxy)-p - phenylene vinylene ] ( oc@xmath122c@xmath123-ppv ) as electron donator sandwiched between ito / pedot : pss and lif / al electrodes a reduction of the open circuit voltage of @xmath124 is observed @xcite .
now , the steady state situation , where a constant current @xmath31 flows across the ito / organic semiconductor system , will be considered .
the solution is given by eq.([solspecial ] ) and the unknown constant @xmath74 has to be found numerically from eq.([bcssx0 ] ) .
in fig.[infssd0 ] , the solution for the dimensionless electric field @xmath125 is depicted for current densities of @xmath126 and barrier - free injection ( @xmath104 ) . in units of @xmath106 for barrier - free charge - carrier injection and a constant current of @xmath127 ( solid line ) and @xmath128 ( dashed line ) as function of the coordinate @xmath26 in units of @xmath90 .
_ , width=283,height=241 ] comparison with the equilibrium solution shows that the distribution of the electric field is , near the electrode , virtually independent of the net current density , but is substantially effected by the current deep in the dielectric bulk . for @xmath129
many holes diffuse into the organic semiconductor ( @xmath130 large ) and a strong negative field in the vicinity of the interface emerges .
this negative field compensates for the diffusion into the organic semiconductor to such an extend that the net current reaches @xmath31 .
however , far away from the contact where diffusion is negligible , the electric field has to be positive ( together with the current ) .
thus , there is a position @xmath131 where the electric field changes sign , as can be seen in fig.[infssd0 ] . since a vanishing electric field strength
is assigned to the ohmic contact itself , the position @xmath132 is often referred to as the virtual electrode @xcite .
for a detailed discussion of the field distribution in the insulator , the dependence of the position of the virtual electrode , the electric field at the contact @xmath70 , and the charge - carrier density ( represented by @xmath133 ) are shown in fig.[infe0 ] as functions of @xmath8 and for two typical current densities .
additionally , the distribution of the electric field is depicted in fig.[infssdmult ] for different injection barriers and a current density of @xmath128 . in units of @xmath90 for current densities of @xmath127 ( solid line ) and @xmath128 ( dashed line ) as a function of barrier height @xmath8 .
b ) electric field strength at @xmath134 inside the semiconductor for current densities of @xmath127 ( solid line ) and @xmath128 ( dashed line ) as a function of barrier height @xmath8 .
the axis on the right hand side indicates the corresponding values for the derivative of the electric field for current densities of @xmath127 ( dashed - dotted line ) and @xmath128 ( dotted line ) .
_ , width=359 ] in units of @xmath106 for a constant current of @xmath135 and barrier height of @xmath136 ( solid line ) , @xmath137 ( dashed line ) , @xmath138 ( dotted line ) and @xmath139 ( dash - dotted line ) as function of the coordinate @xmath140 in units of @xmath90 .
_ , width=283,height=241 ] for small @xmath8s , the electric field at the interface is negative , the charge - carrier density is large , and the virtual electrode is far inside the insulator .
a charge - carrier reservoir is formed in the near contact region and the current across the residual insulator @xmath141 is supplied by this reservoir . from fig.[infssd0 ] and [ infssdmult ] it becomes evident that the electric field follows @xmath142 for @xmath141 and hence , resembles the field distribution of sclc assuming ohmic boundary conditions @xcite .
increasing the injection barrier , the position of the virtual electrode is hardly effected until the injection barrier exceeds @xmath143 .
for a further increased barrier it approaches the physical electrode rapidly .
simultaneously , @xmath70 approaches zero and is pinned there for quite a wide @xmath8-range . however
, this does not mean that the ohmic boundary conditions ( @xmath144 and @xmath145 ) assigned to an ideal contact @xcite is a good approximation in this case .
in fact , the density of the charge - carriers at @xmath134 is relatively low , though @xmath144 holds .
@xmath146 reflects neither @xmath147 nor a constant form . by further increasing @xmath8 ,
the amount of injected charge carriers remains small , @xmath148 is located at the physical electrode and @xmath70 becomes positive , while the field in the insulator is weakly dependent on @xmath140 .
the charge - carrier reservoir in the insulator is depleted and due to the few charge - carriers at the contact , a strong positive field is required to drive the current across the near contact region .
note that for @xmath149 ( or @xmath150 ) , the position of the virtual electrode depends strongly on the induced current density while @xmath70 and @xmath133 do not , reflecting the fact that close to the contact the equilibrium field distribution is hardly effected by the current but further in the insulator it is . for barrier - free injection , the virtual electrode is shifted from approximately @xmath151 to @xmath152 once the current increases from @xmath127 to @xmath128 .
this is illustrated in fig .
[ infssd0 ] and [ infe0 ] .
vice versa , the influence of the current density on the field at the contact becomes recognizable for larger barriers , once the charge carrier reservoir is depleted and the virtual electrode coincides with the physical one .
this is due to the fact that a strongly increased positive field is required to support an additional current density , since the charge - carrier density is small . from fig.[infe0
] it can be seen that the minimal injection barrier required to result in a match of virtual and physical electrode ( @xmath144 ) is shifted to smaller barriers for higher current densities . by means of the solution ( [ solspecial ] ) ,
this may be formulated as an exact relation between the quantities involved .
the requirement @xmath144 results in the particular value of the constant @xmath153 where @xmath154 is the first zero of the airy function @xmath155 . using eq.([bcssx0 ] ) , the current magnitude @xmath156 , which suppresses the electric field at the interface , is determined to @xmath157 from the above equation it becomes evident that for increased injection barriers , the minimal current density required to obtain a match of virtual and physical electrode is exponentially reduced .
knowledge about the distribution of the electric field gives access to the voltage drop @xmath2 across the system for a given current density @xmath31 and hence , to its iv - characteristics .
as the model consists of infinite semi - spaces , integration over the electric field diverges when being carried out over the entire space , i.e. from @xmath158 to @xmath159 .
this holds in equilibrium as well as in steady state .
however , since the focus of this work is to analyze the contact phenomena arising from the conductor / insulator junction , two simplifications are introduced .
firstly , according to the definition of the contact potential the voltage drop is calculated by introducing a cut - off length @xmath1 corresponding to the typical thickness of the organic layer .
hence , the bulk conduction in an organic layer with finite thickness is taken into account . secondly ,
since the voltage drop across a conductor bulk of finite thickness is small due to its high conductivity , the constant part equal to the asymptotic constant field value times the macroscopic conductor thickness is subtracted from the integral over the conductor .
then , the voltage drop over the entire system reads : @xmath160 } { \mathrm{ai}\left[(\iota/2)^{1/3}c_1\right ] } \right| } \right\ } - v_{equi}. \label{voltinf}\ ] ] here , the first two terms in braces represent the voltage drop in the electrode . the third term in braces results from the voltage drop within a distance @xmath1 from the contact inside the organic semi - space and the last term corrects the voltage by its equilibrium value , eq.([vequi ] ) .
the two ( not independent ) constants @xmath70 and @xmath74 result from the steady - state boundary condition , eq .
( [ bcssx0 ] ) . in fig.[infivdmult ] , the resulting iv - characteristics for different barrier heights are presented assuming @xmath161 . in the displayed voltage region ,
the barrier - free contact ( @xmath104 ) is able to supply more charge - carriers than the bulk of the organic semiconductor can transport .
hence , the entire system appears to be space - charge limited with a current approximately @xmath162 for all voltages .
deviations from the mott - gurney law are due to the formation of the space - charge region emerging by charge - carrier diffusion at @xmath163 . as the current increases , the position of the virtual electrode approaches the real electrode and the width of this space charge region decreases .
thus , the calculated curve reproduces the mott - gurney law @xcite more exactly for higher voltages . .
the curves for @xmath104 and @xmath164 nearly merge in the above representation._,width=283,height=241 ] in the presence of non - vanishing barriers , the current is substantially reduced . in the low voltage regime
this reduction is about one order of magnitude once the height of the barrier increases from @xmath104 to @xmath165 . for high @xmath8s , one observes a transition from a @xmath162 to a @xmath166 dependence , followed by an exponential current increase with the applied voltage .
since the sclc represents an upper limit for the current flow through the system , all iv - curves approach the sclc regime for even higher voltages .
the calculated iv - characteristics for high injection barriers can be understood by considering some simple approximations .
once the electric field at the interface becomes positive , the charge - carrier reservoir established in the organic semiconductor is exhausted and no virtual electrode is present .
hence , diffusion is negligible all over the space and an electrical current is virtually due to charge carrier drift only , @xmath167 the solution depends on @xmath70 and reads @xcite , @xmath168 analyzing the drift equation ( [ drift ] ) at @xmath134 by using the boundary condition , eq.([bcssx0 ] ) leads to an exponential dependence of the dimensionless current @xmath65 on the local value of the electric field @xmath169 where the right hand side of the boundary condition , eq.([bcssx0 ] ) , has been neglected . solving eq.([approxcurrent ] ) numerically with respect to @xmath70 for a given current density @xmath65 determines the electric field distribution , eq.([driftfield ] ) , and integrating the field over the entire system leads to the iv - characteristic .
this approximation offers a simple way to calculate iv - characteristics @xcite displayed exemplary for a barrier height of @xmath165 in fig.[infivapprox ] . .
the solid line has been obtained using the exact solution ( [ solspecial ] ) of the complete drift diffusion equation ( [ dglorgdimless ] ) .
the circles show the iv - characteristic calculated with the help of the semi - analytical approach derived for high injection barriers , eqs.([driftfield])-([approxcurrent ] ) , and the triangles display the injection current @xmath170 calculated with the help of eq.([injeccurrent ] ) .
the dashed line shows a characteristic current of the virtual electrode appearance._,width=283,height=241 ] the agreement with the iv - characteristic calculated using the exact solution , eq.([solspecial ] ) , is perfect for all voltages where diffusive transport of charge carriers is negligible .
this includes the sclc regime at high bias . only at low voltages , where the charge carrier reservoir in the organic semiconductor is not exhausted ,
deviations are recognizable . according to the exact relation , eq .
( [ fieldnode ] ) , one can see that below a characteristic current @xmath171 the virtual electrode at @xmath150 appears and the diffusion becomes important . for a purely injection - limited regime space charge effects
are of no importance , resulting in a constant electric field all over the semiconductor semi - space .
this field coincides with the electric field strength at the interface .
in such a case @xmath172 and eq.([approxcurrent ] ) governs the injection limited iv - characteristics alone , @xmath173 .
\label{injeccurrent}\ ] ] eq.([injeccurrent ] ) resembles the result of the drift - diffusion equation @xcite . yet
, the self - consistent treatment of the injection problem yields directly a barrier lowering arising from the potential energy a charge carrier gains on the conductor side of the interface .
in contrast to the @xmath174 dependence of the schottky - lowering predicted from the image - charge potential in the single - electron picture , the current density depends exponentially on the external voltage @xmath2 .
the injection - limited current @xmath175 is depicted in fig.[infivapprox ] .
once space charge effects become predominant , @xmath70 can be neglected in eq.([driftfield ] ) and the temperature - independent mott - gurney law is reproduced .
this occurs as soon as the field - induced barrier lowering has proceeded to such extent that the contact can again establish space charge in the organic semiconductor .
hence , a transition from injection - limited to space - charge - limited current is observed . as may be seen from fig.[infivapprox ]
, the crossover from injection - limited to sclc occurs at rather high voltages in presence of a medium injection barrier of @xmath176 .
this has its origin in the weak barrier lowering for increasing external voltage . due to the low screening length in the conductor , a high voltage drop across the entire system
is required to result in an energy gain in the conductor being comparable to the injection - barrier energy .
note that in contrast to a space - charge formation resulting from diffusion , the electric field at the contact is positive once the space - charge formation is solely due to a drift - controlled injection .
such a sclc regime occurs also for barrier - free injection as soon as the diffusive filling of the space - charge region is weaker than its depletion due to the external applied field .
this , however , requires a very high bias . as a result
, the sclc is directly supplied by an efficient charge - carrier drift out of the conductor . from eq.([injeccurrent ] ) and mott - gurney law one can estimate the upper limit of the barrier required to retain sclc over the whole voltage range : @xmath177 assuming the parameters introduced before , we can predict a critical barrier height of @xmath178 .
this is in slight contradiction to device models in which a schottky - type barrier lowering is assumed @xcite .
here , one would expect the crossover between sclc and injection - limited current to occur at an injection barrier of @xmath179 .
however , up to now the experimental data available predict the crossover to be at a barrier height between @xmath180 and @xmath181 , being consistent with both approaches .
in the calculation of the charge - carrier transport through insulators , a fundamental question about the boundary conditions generally arises when a charge - carrier injecting interface has to be involved .
typically , boundary conditions at the interface are chosen , fixing there the charge carrier density and/or the electric field . however , at the conductor / insulator contact the system is ill - defined , meaning that especially at the interface the charge carrier density and the electric field strongly depend on the condition of the system . in this paper ,
a one - dimensional analytical model describing the charge - carrier transport across a conductor / insulator junction was presented , where boundary conditions are defined far into the conductor and the insulator , respectively . here
, the influence of the two materials on each other is negligible so that they can be regarded as independent . considering the poisson equation and assuming the electric displacement as well as the electrochemical potential being continuous over the entire system
, the electric field distribution and the current - voltage characteristic were derived .
the model predicts sclc , injection - limited conduction as well as the crossover between the two limits . in most current regimes ,
the influence of the self - consistent treatment is noticeable . for pure injection - limited conduction , an injection current
similar to the prediction of the drift - diffusion theory was derived . however , due to the consistent treatment of the injection problem in one dimension , a barrier lowering different to the one predicted from the three - dimensional image - charge potential in the single - electron picture emerges .
this injection - barrier lowering results from the potential energy gain ( or loss ) of the charge - carriers in the electrode . also for sclc , a deviation of the mott - gurney law is observable as long as the virtual electrode does not match the physical one . only at high bias
the self - consistent treatment has no influence on the iv - characteristic and the well known mott - gurney law is fully reproduced .
useful discussion with r. schmechel is gratefully acknowledged .
the authors would express their thanks to the deutsche forschungsgemeinschaft for financial support of the sonderforschungsbereich 595 .
the coauthors dedicate this article to the memory of frederick neumann whose life was taken by a tragic accident .
lambert and p. mark , _ current injection in solids _
( academic press , new york 1970 ) . p.w.m .
blom , m.j.m . de jong , and j.j.m .
vleggaar , appl .
. lett . * 68 * , 3308 ( 1996 ) .
blom , m.j.m . de jong , and m.g .
van munster , phys .
b * 55 * , r656 ( 1997 ) .
f. neumann , y.a .
genenko , r. schmechel , and h. von seggern , synth . met . *
150 * , 291 ( 2005 ) .
sze , _ physics of semiconductor devices _ ( john wiley & sons , inc .
, new york 1969 ) .
n.w . ashcroft and n.d .
mermin , _ solid state physics _ ( thomson learning , inc . , florence 1976 ) .
a.b . walker , a. kambili and s.j .
martin , j. phys . : condes .
matter * 14 * , 9825 ( 2002 ) .
martin , a.b .
walker , a.j .
campbell , and d.c .
bradley , j. appl .
phys . * 98 * , 063709 ( 2005 ) .
e. tuti , m.n .
bussac , b. masenelli , m. carrard , and l. zuppiroli , j. appl . phys . * 89 * , 430 ( 2001 ) .
f. neumann , y.a .
genenko , c. melzer , and h. von seggern , j. appl . phys . * 100 * , 084511 ( 2006 ) .
s. christen and m. seeger , journal of electrostatics * 65 * , 11 ( 2007 ) .
e.m . lifshitz and l.p .
pitaevskii , _ physical kinetics _ ( butterworth - heinemann ltd . , oxford 1999 ) .
n.d . lang and w. kohn , phys .
b * 1 * , 4555 ( 1970 ) .
n.d . lang and w. kohn , phys . rev .
b * 3 * , 1215 ( 1971 ) .
d. mergel and z. qiao , j. phys .
phys . * 35 * , 794 ( 2002 ) .
d. mergel and z. qiao , j. appl .
phys . , * 95 * , 5608 ( 2004 ) .
h. fujiwara and m. kondo , phys .
b * 71 * , 075109 ( 2005 ) .
a.d . polyanin and v.f .
zaitsev , _ handbook of exact solutions for ordinary differential equations _ ( chapman & hall / crc , new york 2003 ) . , edited by m. abramovitz and i. stegun , ( dover publications , new york , 1970 )
. h. ishii , k. sugiyama , e. ito , and k. seki , adv .
, * 11 * , 605 , ( 1999 ) .
veenstra , u. stalmach , v.v .
krasnikov , g. hadziioannou , h.t .
jonkman , a. heeres , and g.a .
sawatzky , appl .
lett . * 76 * , 2253 ( 2000 )
. h. peisert , m. knupfer , t. schwieger , j.m .
auerhammer , m.s .
golden , and j. fink , j. appl . phys . * 91 * , 4872 ( 2002 ) .
h. bssler , phys . stat
( b ) * 175 * , 15 ( 1993 ) .
scott and g.g .
malliaras , chem .
* 299 * , 115 ( 1999 ) .
malliaras and j.c .
scott , j. appl .
* 85 * , 7426 ( 1999 ) .
arkhipov , u. wolf , and h. bssler , phys .
b * 59 * , 7514 ( 1999 ) .
arkhipov , e.v .
emelianova , y.h .
tak , and h. bssler , j. appl . phys . *
84 * , 848 ( 1999 ) .
arkhipov , h. v. seggern , and e.v .
emelianova , appl .
phys . lett . *
83 * , 5074 ( 2003 ) .
a.m. stoneham , m.m.d .
ramos , a.m. almeida , h.m.g .
correia , r.m .
ribeiro , h. ness and a.j .
fisher , j. phys . :
matter * 13 * , 2411 ( 2002 ) .
mihailetchi , p.w.m .
blom , j.c .
hummelen , and m.t .
rispens , j. appl .
* 94 * , 6849 ( 2003 ) .
a. rose , phys .
rev . * 97 * , 1538 ( 1955 ) .
emtage and j.j .
odwyer , phys .
* 16 * , 356 ( 1966 ) .
blom , m.c.j.m .
vissenberg , materials science and engeneering * 27 * , 53 ( 2000 ) .
crowell and s.m .
sze , solid state electron .
* 9 * , 1035 ( 1966 ) .
campbell , p.s .
davids , d.l .
smith , n.n .
barashkov , and j.p .
ferraris , j. appl .
phys . * 72 * , 1863 ( 1998 ) .
|
we present a closed description of the charge carrier injection process from a conductor into an insulator .
common injection models are based on single electron descriptions , being problematic especially once the amount of charge - carriers injected is large .
accordingly , we developed a model , which incorporates space charge effects in the description of the injection process .
the challenge of this task is the problem of self - consistency .
the amount of charge - carriers injected per unit time strongly depends on the energy barrier emerging at the contact , while at the same time the electrostatic potential generated by the injected charge- carriers modifies the height of this injection barrier itself . in our model , self - consistency is obtained by assuming continuity of the electric displacement and the electrochemical potential all over the conductor / insulator system .
the conductor and the insulator are properly taken into account by means of their respective density of state distributions .
the electric field distributions are obtained in a closed analytical form and the resulting current - voltage characteristics show that the theory embraces injection - limited as well as bulk - limited charge - carrier transport .
analytical approximations of these limits are given , revealing physical mechanisms responsible for the particular current - voltage behavior .
in addition , the model exhibits the crossover between the two limiting cases and determines the validity of respective approximations .
the consequences resulting from our exactly solvable model are discussed on the basis of a simplified indium tin oxide / organic semiconductor system .
|
spectral estimation constitutes a classical problem that has found applications in a large variety of fields ( including astronomy , radar , communications , economics , medical imaging , spectroscopy , , to name but a few ) .
one important category of spectral estimation problems arises for signals that can be well represented by the parametric model @xmath0 where @xmath1 is an additive noise term . given a vector of (
typically equally spaced ) samples , @xmath2 where @xmath3 is the sampling period , the goal is to estimate the complex frequency nodes @xmath4 i.e. , the damping and frequency parameters @xmath5 and @xmath6 .
note that once the parameters @xmath7 have been computed , determining @xmath8 reduces to a simple linear regression problem .
thus , the focus lies on the estimation of the nodes @xmath9 .
the signal model includes the case of exponentially damped signals defined by @xmath10 .
this case has recently received significant interest , notably in applications involving nuclear quadrupole resonance signals for which the @xmath5 ( together with @xmath6 ) are used as a signature of the chemical composition of an analysed substance ( see for instance @xcite and references therein ) .
the literature on methods for estimating @xmath9 in this setting is rich , cf .
@xcite for an overview . here
, we focus on so - called _ high - resolution _ methods , i.e. , methods whose frequency resolution is not tied to a pre - defined grid ( and can attain numerical precision in the noise free case ) .
+ one important class of techniques is based on the principle of nonlinear least squares ( nls ) @xcite and aims at solving the nonlinear regression problem associated with , i.e. , at minimizing @xmath11 with respect to the parameter vectors @xmath12 and @xmath9 . although nls enjoy desirable theoretical properties ( e.g. , those of maximum likelihood estimators when @xmath1 is a gaussian white noise ; robustness to coloured noise ) , their practical use remains limited due to the extreme difficulty of globally minimizing @xmath13 due to its pronounced multimodal shape with one very sharp global minimum ( cf .
, e.g. , @xcite and references therein ) .
+ a second prominent and popular class is given by subspace methods ( such as music @xcite , esprit @xcite and min - norm @xcite ) .
these methods are based on estimates of the covariance of @xmath14 and rely on the assumption that the noise @xmath15 is white . in the present contribution
, we develop a novel high - resolution methodology for the estimation of @xmath9 in .
similar to nls , we aim at approximating @xmath14 as good as possible by a linear combination of complex exponentials .
however , the proposed methodology fundamentally departs from any of the above methods in the following ways : first , it is based on _
kronecker s theorem _ for complex symmetric matrices , which essentially states that if a function @xmath16 is uniformly sampled , then the _ hankel matrix _ that is generated by the vector of samples @xmath14 has rank @xmath17 if and only if @xmath16 coincides at the sample points with a function that is a linear combination of @xmath17 exponential functions .
this fact has been used for the sparse approximation of functions by sums of exponentials in the con - eigenvalue approach @xcite and the alternating projections method @xcite . here , it is used to formulate an nls type minimization problem in which the model ( and its parameters @xmath12 and @xmath9 ) does not enter explicitly . instead , the model is imposed _
implicitly _ by constraining the rank of the hankel matrix generated by the vector @xmath18 approximating @xmath14 to equal @xmath17 . consequently , residual minimization is not performed over the parameter space @xmath19 directly , but over the space of vectors @xmath18 which generate hankel matrices of rank @xmath17 .
the frequency nodes @xmath9 are then obtained by considering the con - eigenvectors of the solution hankel matrix .
second , we reformulate the minimization problem such that it can be effectively solved by the _ alternating direction method of multipliers _
( admm ) @xcite .
admm is an iterative technique that is recently gaining popularity due to its robustness , versatility and applicability to large - scale problems .
moreover , this technique enjoys performance comparable to problem - specific state - of - the - art methods . while the optimization problem considered here is nonconvex , it is shown numerically that the solutions obtained with the admm procedure generically provide excellent approximations and parameter estimates for the model .
the resulting hankel matrix admm frequency estimation procedure is practically extremely appealing due to its simplicity and ease of implementation .
it enjoys excellent performance , outperforming subspace methods while at the same time alleviating the practical limitations of classical nls . unlike subspace methods
, it does not rely on a specific noise model .
another interesting property is that it applies to situations with _ missing _ ( or censored ) data .
the remainder of this work is organized as follows . in section
[ sec : admm ] , we define an optimization problem based on kronecker s theorem for the approximation by sums of complex exponentials and formulate the admm - based procedure for obtaining its solution . section [ sec : freqest ] summarizes the procedure for obtaining frequency node estimates from the solution hankel matrix . in section [ sec : results ] , we analyze the performance of the method by means of numerical simulations . section [ sec : conclusions ] concludes this contribution and points to future work .
in this section , we approximate the vector @xmath14 as good as possible by a vector @xmath18 that is a linear combination of complex exponential functions . using kronecker s theorem
, we express the solution to this approximation problem in terms of the vector @xmath18 whose hankel matrix has rank @xmath17 and minimizes the residual .
the resulting optimization problem will then be reformulated and solved by admm . for completeness , we recall that a hankel matrix @xmath20 has constant values on the anti - diagonals , i.e. , @xmath21 it can thus be generated elementwise from a vector @xmath22 , such that @xmath23 we denote this by @xmath24 . _ kronecker s theorem _
@xcite states that if the hankel matrix @xmath24 generated by the vector @xmath18 is of rank @xmath17 then , with the exception of degenerate cases , there exists @xmath25 and @xmath26 such that @xmath27 the converse holds as well .
it follows that the best approximation ( in the @xmath28 sense ) of @xmath14 by a linear combination of @xmath17 complex exponentials is given by the vector @xmath18 which satisfies @xmath29 and minimizes the @xmath28 norm of the residual @xmath30 . in other words , @xmath18 can be obtained as the solution of the optimisation problem @xmath31 the parameter @xmath17 can be selected by considering the singular values of the hankel matrix @xmath32 generated by the samples @xmath14 , cf .
, @xcite . at this stage ,
we assume @xmath17 to be given .
let @xmath33 be the indicator function for square matrices @xmath34 given by @xmath35 then , problem can be reformulated as follows @xmath36 the optimization problem has a cost function that consists of two terms , each depending only on one of the variables @xmath37 and @xmath38 , along with a linear constraint .
the problem formulation is therefore well suited to be addressed using admm .
admm is an iterative technique in which a solution to a large global problem is found by coordinating solutions to smaller subproblems .
it can be seen as an attempt to merge the benefits of dual decomposition and augmented lagrangian methods for constrained optimization .
for an overview of the admm method see @xcite . for a convex cost function admm
is guaranteed to converge to the unique solution . for non - convex problems
it can get stuck at local minima , although it can work quite well in practice also in these situations , cf .
* chapter 9 ) . since the rank constraint @xmath33 is nonconvex , convergence of an algorithm solving is not guaranteed .
our numerical simulations indicate that the method works substantially better than established high resolution techniques like esprit , and moreover that it can be applied to cases where esprit is non - applicable ( e.g. , missing data ) by the nuclear norm , at the cost of biased solutions . ] .
an admm iterative step ( from iteration @xmath39 to iteration @xmath40 ) for ( [ hankel_eq ] ) reads @xmath41 where @xmath42 are the lagrange multipliers , and @xmath43 is the augmented lagrangian associated with ( [ hankel_eq ] ) , @xmath44.\end{gathered}\ ] ] the first minimization step is nonconvex due to the projection onto a non convex set but can be solved analytically .
deriving closed form expressions for both minimization steps steps and is straightforward .
the solution @xmath45 to the first minimization step is equal to the best rank @xmath17 approximation of the matrix @xmath46 defined elementwise by @xmath47 denoting by @xmath48 the singular value decomposition of @xmath46 , then by the eckart - young theorem , @xmath49 where @xmath50 is obtained from the diagonal matrix @xmath51 by replacing the diagonal elements @xmath52 for @xmath53 with zeros .
+ the solution to the second minimization step is given by @xmath54}{1 + \rho q(l ) } , \ ] ] where @xmath55 stands for @xmath56 and @xmath57 with these explicit expressions for the admm minimization steps , the procedure for the approximation by sums of exponentials is given by the following simple matlab function : -0.75 .... function f_appro = expo_sum_admm(f , p , rho , it_max ) % best approximation to f using p exponentials .
n=(length(f)-1)/2;r = zeros(2*n+1,1 ) ; lambda = zeros(n+1);q=[1:n+1,n:-1:1 ] ' ; for iter=1:it_max , b = form_hank(f - r)-lambda / rho;[u , s , v]=svd(b ) ; s = diag(s);s(p+1:end)=0;a = u*diag(s)*v ' ; r=(rho*q.*f - sum_hank(lambda+rho*a))./(1+rho*q ) ; lambda = lambda+rho*(a - form_hank(f - r ) ) ; end ; f_appro = sum_hank(a)./q ; function h = form_hank(f ) % form hankel matrix h = hankel(f(1:(end+1)/2),f((end+1)/2:end ) ) ; function f = sum_hank(h ) % sum anti - diagonals n = size(h,1)-1 ;
f = zeros(2*n+1,1 ) ; h = flipud(h ) ; for j =- n : n , f(j+n+1,1)=sum(diag(h , j ) ) ; end ; .... it is straightforward to replace @xmath58 in problem by a weighted norm @xmath59 and to derive the corresponding admm procedure
. the only change in the above final expressions is that the denominator in the second minimization step ( and in the corresponding line 8 of the matlab code ) is replaced by @xmath60 .
let @xmath61 denote the set of indices @xmath62 of the missing data samples and set @xmath63 .
the use of weights @xmath64 defined by @xmath65 yields an admm procedure for data with missing samples . in order to highlight the simplicity of the proposed algorithm , the admm procedure was described above to terminate after a preset number @xmath66 of iterations . commonly used stopping criteria are based on the magnitude of the primal and dual residuals and are straight - forward to incorporate in the admm procedure ( cf .
, @xcite for details ) .
the admm procedure would converge to the global solution of the problem if the problem was convex .
since it is non convex , it is only guaranteed to converge locally and can converge to different points depending on the choice of @xmath67 and initial points @xmath68 and @xmath69 .
numerical experiments indicate , however , that the admm solution @xmath70 of the problem provides in general an excellent approximation for @xmath14 . here
, we have initialized @xmath68 and @xmath69 with zeros .
once the solution hankel matrix @xmath24 to has been found , the parameter vector @xmath9 can be obtained by considering the so - called con - eigenvectors of @xmath37 .
hankel matrices belong to the class of complex symmetric matrices satisfying @xmath71 , which generically can be decomposed as @xcite @xmath72 where @xmath73 are the decreasingly ordered con - eigenvalues of @xmath37 and where the _ con - eigenvectors _
@xmath74 are orthogonal and satisfy the relation @xmath75 . similarly to the eckart young theorem , it can be shown that the best rank @xmath76 approximation of @xmath37 is given by @xcite @xmath77 since @xmath18 is a sum of @xmath17 exponentials , @xmath78 can be expressed in the form ( [ rank_n_rep ] ) , and therefore each con - eigenvector @xmath79 will also be a sum of the same exponentials .
let @xmath80 .
it is then possible to write @xmath81 , where @xmath82 is the @xmath83 vandermonde matrix generated by @xmath84 ( i.e. , @xmath85 ) and @xmath86 is some ( invertible ) @xmath87 matrix .
denote as @xmath88 ( resp .
@xmath89 ) the matrix @xmath90 whose first row ( resp .
last row ) has been dropped .
clearly , we have @xmath91 , @xmath92 and the vandermonde structure of @xmath82 leads to @xmath93 it follows that @xmath94 where @xmath95 stands for conjugate transpose , where @xmath96 denotes the pseudo - inverse operator .
therefore , we can compute the nodes @xmath9 by computing the eigenvalues of @xmath97 .
.[tab : param]parameters for the @xmath98 ( top ) and @xmath99 ( bottom ) complex exponentials used in the numerical simulations . [ cols= " > , > , > , > , > , > , > , > " , ] performance for the estimation of @xmath6 ( left ) and @xmath5 ( right ) , @xmath100 , for esprit ( top ) and the admm procedure ( bottom ) .
the crbs are plotted in solid lines.,title="fig : " ] performance for the estimation of @xmath6 ( left ) and @xmath5 ( right ) , @xmath100 , for esprit ( top ) and the admm procedure ( bottom ) .
the crbs are plotted in solid lines.,title="fig : " ] + performance for the estimation of @xmath6 ( left ) and @xmath5 ( right ) , @xmath100 , for esprit ( top ) and the admm procedure ( bottom ) .
the crbs are plotted in solid lines.,title="fig : " ] performance for the estimation of @xmath6 ( left ) and @xmath5 ( right ) , @xmath100 , for esprit ( top ) and the admm procedure ( bottom ) .
the crbs are plotted in solid lines.,title="fig : " ] performance for the estimation of @xmath6 , @xmath100 , for the admm procedure when @xmath101 out of @xmath102 samples are missing at random positions ( left ) and as one consecutive block ( right ) .
the solid lines indicate the crbs.,title="fig : " ] performance for the estimation of @xmath6 , @xmath100 , for the admm procedure when @xmath101 out of @xmath102 samples are missing at random positions ( left ) and as one consecutive block ( right ) .
the solid lines indicate the crbs.,title="fig : " ] approximation of @xmath99 exponentials for @xmath103 samples of which two blocks of @xmath104 samples are missing ( @xmath105 ) .
real parts of noise - free available samples ( red dots ) and missing samples ( red circles ) ; mean and @xmath106 standard deviation error band ( blue solid lines ) . ]
we analyze the estimation performance of the proposed frequency estimation algorithm by considering different numerical simulations conducted with @xmath98 and @xmath99 complex exponentials embedded in circular white gaussian noise ( @xmath107 independent realisations ) for different signal - to - noise ratios ( snr ) .
the model parameters @xmath12 and @xmath9 are summarized in tab .
[ tab : param ] ( units correspond to uniform sampling of the interval @xmath108 $ ] ) .
the performance are compared with the theoretical cramr - rao bounds ( crbs ) for the estimation problem ( see , e.g. , @xcite ) . in figure
[ fig : perf ] , the standard deviations ( stds ) of the estimates of @xmath5 and @xmath6 , @xmath100 , obtained with the admm procedure and with the esprit method , are plotted as a function of snr together with the ( square root of the ) crbs .
note that the components @xmath109 and @xmath110 are close in frequency but significantly differ in amplitude ( cf . ,
[ tab : param ] ) .
we observe that the proposed admm procedure provides estimates attaining the corresponding crbs , indicating that the procedure realizes the minimum variance estimator .
it consistently outperforms the esprit method in terms of std , and especially so for small snr and for the two components that are close in frequency . in fig .
[ fig : perfmissing ] , we summarize the performance of the admm procedure for estimating @xmath6 , @xmath100 , for @xmath111 samples of which @xmath101 samples are missing at random positions ( left ) and as one consecutive block centred at random position ( right ) .
the stds are again found to reach the ( square root of the ) crbs , demonstrating that the admm procedure is not affected by missing data .
similar results are obtained for @xmath5 and are not reported here for space limitation reasons .
finally , figure [ fig : appro ] illustrates the approximation performance of the admm procedure for @xmath103 samples of @xmath99 exponentials of which two blocks of 60 consecutive samples are missing ( @xmath105 , mean over 100 realizations ) .
despite the fact that @xmath112 of the samples are missing , the obtained approximations are notably consistent with the underlying noise - free function .
the maximum observed differences @xmath113 are found to be below @xmath114 times the sampling frequency .
a high - resolution parametric frequency estimation procedure that is based on approximation with sums of @xmath17 complex exponentials was proposed . in the proposed algorithm ,
the kronecker theorem was used to cast the approximation problem in terms of generating functions for hankel matrices of rank @xmath17 .
the resulting optimization problem was addressed by an admm procedure .
the hankel matrices obtained from the admm procedure were then used to compute the parameters of the complex exponential model .
this is in contrast to other methods , such as classical nls or subspace methods .
although the optimization problem considered in this paper is non - convex ( and the problems of local minima are inherited with non - convexity , as is the case for classical nls ) numerical simulations indicated that the admm procedure yields excellent practical performance .
the admm procedure can be applied to equally spaced samples , including situations with missing data , and does not rely on explicit noise model assumptions . despite its versatility , the resulting algorithm is simple and easy to implement .
the method has been presented here for complex - valued data .
however , its real - valued counterpart can be obtained in a similar fashion as outlined in this work .
the procedure will be applied to frequency node estimation in nuclear quadrupole resonance applications .
future work includes the extension to multiple time series and to the estimation of directions of arrival .
the work was supported by the swedish foundation for international cooperation in research and higher education , the swedish research council and the craaford foundation .
part of this work was conducted within the labex cimi during visits of f. andersson at university of toulouse .
|
we develop a parametric high - resolution method for the estimation of the frequency nodes of linear combinations of complex exponentials with exponential damping .
we use kronecker s theorem to formulate the associated nonlinear least squares problem as an optimization problem in the space of vectors generating hankel matrices of fixed rank .
approximate solutions to this problem are obtained by using the alternating direction method of multipliers .
finally , we extract the frequency estimates from the con - eigenvectors of the solution hankel matrix .
the resulting algorithm is simple , easy to implement and can be applied to data with equally spaced samples with approximation weights , which for instance allows cases of missing data samples . by means of numerical simulations ,
we analyze and illustrate the excellent performance of the method , attaining the cramr - rao bound .
* keywords : * frequency estimation , nonlinear least squares , hankel matrices , kronecker s theorem , missing data , alternating direction method of multipliers
|
recent observations show a multitude of physical effects that occur when active galactic nuclei ( agn ) interact with the ambient intracluster medium ( icm ) .
while these effects are widely believed to be crucial for the formation of structure in the universe , they are still poorly understood .
the central galaxy in almost every strong cooling core contains an active nucleus and a jet driven radio galaxy .
the radio power of these cooling cores is somewhat correlated with the x - ray luminosity , although the range of the radio power is much greater than the range of the x - ray core power .
this is supported by a recently discovered correlation between the bondi accretion rates and the jet power in nearby , x - ray luminous elliptical galaxies @xcite .
these results show that the agn at the centres of large elliptical galaxies feed back enough energy to quench cooling and star formation , thus providing a possible explanation for the observed cut - off at the bright end of the galaxy luminosity function @xcite .
the study of agn - icm interactions is of great interest in a much broader context . given the magorrian relation @xcite , clusters hosting a giant cd galaxy should have a massive black hole in their centre .
the growth of these black holes is an integral part of feedback regulated cooling of the icm .
radio - loud agn inflate kpc - sized bubbles of relativistic , underdense plasma ( a.k.a .
radio lobes ) that displace the hot intra - cluster medium and thus appear as depressions in the x - ray surface brightness .
high - resolution x - ray observations of cooling flow clusters with _
chandra _ have revealed a multitude of x - ray holes often coincident with patches of radio emission , a compiled list of which is presented by @xcite .
numerical simulations of hot , underdense bubbles in clusters of galaxies have been performed by a number of authors ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
common to all of these simulations is that they use a spherically symmetric , analytical profile for the icm .
@xcite simulated agn feedback in a cosmologically evolved cluster but did not model the jets .
most recently , @xcite performed three - dimensional simulations of a jet in a hydrostatic , spherically symmetric cluster model . in their simulation
, the jet power was modulated by the mass accretion rate across the inner boundary . in all of their models , jet heating failed to prevent the catastrophic cooling of gas at the centre because the jet preferentially heated gas lying along the jet axis but failed to heat matter in the equatorial plane . however , in reality the dynamics of the jet may be strongly affected by the ambient medium .
bulk velocities may advect and distort the jet and the radio lobe .
density inhomogeneities can affect both the jet propagation and the deposition of entropy in the icm by the jet .
for the purpose of studying the extent of cluster heating by agn , the crucial question is _ exactly where and how much _ energy is deposited by the jet in the icm .
the primary aim of this work is therefore to study the dynamics of the jet and the extent of icm heating subject to the motions and inhomogeneities of the icm . in this letter , we present first results from hydrodynamical simulations of agn heating in a realistic galaxy cluster .
the primary objectives of this work is to study the interaction of the jet with a dynamic , inhomogeneous icm .
the letter is organised as follows : [ sec : code ] describes the technical setup of the simulations , [ sec : discussion ] presents the results and discusses the implications for cluster physics , and
[ sec : summary ] summarised and concludes .
the initial conditions of our simulation are based on a rerun of the s2 cluster from @xcite , whose properties are sufficiently close to a typical , massive , x - ray bright cluster with a mass of @xmath2 and a central temperature of 6 kev .
the output of the gadget sph simulation serves as the initial conditions for our simulation .
we use the flash code @xcite which is a modular block - structured adaptive mesh refinement code , parallelised using the message passing interface .
it solves the riemann problem on a cartesian grid using the piecewise - parabolic method .
our simulation includes @xmath3 dark matter particles .
the particles are advanced using a cosmological variable - timestep leapfrog - method .
gravity is computed by solving poisson s equation with a multigrid method using isolated boundary conditions . for the relatively short physical time of the jet simulation ( 160 myrs ) , radiative cooling and star formation
are neglected , though they were included in the constitutive sph simulation .
the cluster shows significant signs of anisotropy as well as some net rotation of the central cluster gas and contains a dynamically induced cold - front .
we specifically picked a non - relaxed cluster to investigate the possible impact of cluster gas dynamics on jet cluster interactions , to be compared with hydrostatic , spherical atmospheres previously investigated ( e.g. * ? ? ? * ) .
the computational domain is a @xmath4 box around the cluster s centre of mass .
the maximum resolution at the grid centre corresponds to a cell size of @xmath5 , implying 11 levels of refinement .
the total computing time on our 26 processor cluster amounted to 2 processor years .
the simulations presented in this letter were performed assuming an adiabatic equation of state with a uniform adiabatic index of @xmath6 and no radiative cooling .
the jet is injected through a nozzle placed at the centre of the gravitational potential , coincident with the gas density peak of the central elliptical galaxy .
the nozzle is modeled as two circular back - to - back inflow boundaries @xmath7 or 12 resolution elements in diameter .
the nozzle faces obey inflow boundary conditions fixed by the jet s mass- , momentum- , and energy fluxes .
this treatment avoids the entrainment of cluster gas into the jet which is unavoidable in simpler schemes where the jet is approximated by injecting mass , momentum , and energy into a finite volume of the cluster that contains thermal gas and is part of the active computational grid .
we were thus able to cleanly separate jet fluid and cluster fluid and to calculate accurate maps of synchrotron and thermal emission from the two components , respectively , as well as cleanly measuring the impact of the interaction on the thermal cluster gas .
vlbi observations of jets ( e.g. * ? ? ?
* ) and theoretical arguments about the morphology of radio lobes @xcite imply that agn jets suffer from a range of dynamical instabilities as they travel through the igm and their own accumulated exhaust in the lobe .
these instabilities occur on kpc scales and are thus not resolved in our simulation , leading to measurable deviations from straight , axial trajectories , which has been dubbed the `` dentist - drill '' effect @xcite .
mimicking it is critical for accurately modeling injection of momentum and energy by jets .
we included this effect by imposing a random - walk jitter on the jet axis confined to a 20 degree half - opening angle .
the jet material is injected equally in opposite directions with velocity @xmath8 and an internal mach number of 32 .
the jet power of the simulation presented in this letter was chosen to be @xmath9 , corresponding to a rather powerful source , comparable to cyg a , and in line with the powers now implied by the large scale moderate shocks found around , e.g. , hercules a @xcite . in order to quantitatively study the jet s effect on the cluster , we ran a simulation without a jet but otherwise identical parameters for the same length of time , which we refer to as the control simulation .
having thus described the numerical setup , we will now proceed to present and discuss the results .
while the jet is active , the jet - inflated cocoon and the surrounding swept up shell generally follow the morphological evolution discussed previously in the literature : the jets inflate two oblong cocoons , which push thermal gas aside .
the early expansion is supersonic , shocking the surrounding gas directly , while at later times the expansion slows down and eventually becomes sub - sonic .
[ fig : maps1 ] shows a time series of density and entropy cuts as well as simulated x - ray maps ( using the chandra acis - i response and assuming an apec thermal model ) and low - frequency radio maps ( assuming equipartition and neglecting radiative cooling ) .
the aspect ratio of the lobes is roughly 3 and decreases with time , indicating the lateral spreading and shear of the radio plasma . during the active source evolution , while the jet is on , the source closely resembles the best studied example of a source of comparable power : cyg a. the fact that aspect ratio and x - ray and radio morphology of the simulated source so closely resemble cyg a indicates that we properly modeled the effect of the dentist drill effect ( which can , in fact , be directly observed in cyg a , where the jet axis is clearly mis - aligned with the radio and x - ray hot spots ) .
we verified this by running a test case with half the opening angle of the jitter cone ( @xmath10 instead of @xmath11 ) which produced too narrow a cocoon ( with an aspect ratio of around 7 ) .
the x - ray maps show a clear increase in the surface brightness in the outgoing shock / compression wave and a wispy , turbulent wake behind the wave ( fig .
[ fig : maps2 ] ) .
this is best demonstrated by un - sharp masking of the images .
an unsharp masked movie of the x - ray images can be viewed at http://space.mit.edu/@xmath12heinzs/jetpower/ , showing a network of roughly spherical outgoing sound waves that are excited behind the major initial shock wave , consistent with the complex network of sound waves seen in perseus @xcite .
they also show bud - like bubbles appear at the edges of the main shell when the jet axis moves due to `` denstist - drill '' , reminiscent of the budding bubbles in m87 @xcite . two distinctions to previously published
results are important to point out regarding the morphological evolution of the source : first , it is clear from the series of snapshots shown in fig .
[ fig : maps1 ] that the non - sphericity of the atmosphere induces significant asymmetries in the two sides of the cocoon .
given that the difference is so strong , it is clear that the impact on the morphology as a whole must be significant .
this justifies our initial motivation for this study : in order to investigate the evolution of radio sources and their impact on galaxy cluster atmospheres , one can not neglect the dynamic nature of the atmospheres prior to jet injection .
second , the dynamic nature of the cluster atmosphere significantly alters the late stage evolution , after the jet switches off ( between the fourth and the fifth frame at @xmath13 ) and the source evolution becomes sub - sonic . during the sub - sonic stage ,
the impact of the pressure and density of the surrounding gas become more and more important .
in particular , the rotation that is present in the cluster ( see fig . [
fig : maps2 ] ) has clearly sheared the plasma away from the jet axis .
we will quantify these results below .
it is already clear , however , that this effect can solve the dilemma of launching multiple jet episodes into the same cluster : any evacuated channel created by previous outbursts will have been buffeted around sufficiently to move enough cluster gas back into the path of the new jet to provide a sufficient cross section for interaction and to couple the jet to the inner cluster gas .
the spiral - like morphology of the radio source at late stages is very reminiscent of the large scale ( @xmath14 ) morphology of m87 and abell 4059 @xcite , where the large scale radio bubbles are clearly mis - aligned with the inner jet .
this leads us to suggest that there might be significant rotation or shear present in the gas of these clusters . before proceeding to a more quantitative analysis
, it is worth pointing out that the episode of jet activity _ does not _ disrupt the cluster or blow the central region apart , despite the fact that we are simulating a powerful source ( certainly at the upper end of the power range expected from central cluster radio sources ) .
contrary to what might be expected naively , the expansion shock weakens and does not super - heat the central atmosphere beyond convective stability .
this does not mean , however , that the jet does not induce significant turbulent motion in the central cluster .
as found in earlier investigations of jet - cluster interactions , the impact of the jet on the cluster leads to an increase in the central entropy and a decrease in density , i.e , a net heat input .
this is qualitatively apparent form the entropy panels in fig .
[ fig : maps1 ] . figures [ fig : massprofiles ] and [ fig : lagrangian ] quantify this result . the bottom panel of fig . [ fig : massprofiles ] shows the cumulative radial mass , i.e. , the mass @xmath15 contained within radius @xmath16 from the cluster centre .
the initial time step lies above all other curves out to a radius of about 60 kpc , indicating that the cluster has been inflated ( i.e. , mass moved out ) .
the bottom panel of fig .
[ fig : lagrangian ] shows the increase in the mean entropy @xmath17 of the cluster gas at different time steps over the control simulation , plotted against the lagrangian coordinate @xmath15 from fig .
[ fig : massprofiles ] .
this plot clearly shows that the entropy of the central @xmath18 has been increased by about 0.1 , corresponding to an increase in @xmath19 by about 25% .
the top panel of fig .
[ fig : lagrangian ] shows the ratio of the cooling times with and without jet . at late times ( @xmath20 ) the cooling time for the inner few@xmath21 has been increased by about 50% , more at earlier times .
this indicates that , on average , the jet can halt cooling for a time significantly longer than the time the jet is active .
future simulations including radiative cooling will be necessary to investigate whether this increase is sufficient to offset cooling entirely and whether this result is consistent with _
xmm_-newton and _ chandra _ constraints on cluster cooling rates .
one of the criticisms of models of cluster heating by jets is the non - spherical nature of the energy input .
the problem is twofold : firstly , if the energy is predominantly injected into the axial direction , the equatorial cluster gas does not benefit from the energy injection and can cool unimpeded , while the axial gas is super - heated , making agn heating very inefficient . secondly , the channel of jet exhaust and super - heated gas along the jet axis would allow subsequent episodes of jet activity to punch easily through the inner cluster and deposit their energy well outside the central region , thus allowing the inner regions to cool unimpeded .
we already suggested that the shear and rotation in the central cluster can rectify the second problem .
the top panel of fig .
[ fig : massprofiles ] shows the radial profile of the target mass @xmath22 contained in a @xmath11 angle from the jet axis , relative to that of the control simulation . clearly , the amount of mass is significantly reduced during the early stages of the simulation . however , at late times , @xmath22 increases back up to 80% of the unperturbed value , consistent with the general reduction of density in the central cluster shown in the bottom panel of fig .
[ fig : massprofiles ] .
such a small reduction is dynamically insignificant , i.e. , the amount of target mass is more than sufficient for a subsequent jet episode to couple to the cluster at _ any _ radius .
this proves quantitatively that cluster dynamics can solve the problem of evacuated channels raised by @xcite . if cluster mass can be moved around to re - fill the jet channel and to buffet the fossil radio plasma around , it is plausible to assume that the energy input of the jet might be isotropised more rapidly as well .
the entropy panel of fig .
[ fig : maps1 ] shows this qualitatively : there is no clear high entropy channel along the axis of the jet .
in fact , the draft of the rising bubbles draws low entropy material out of the central cluster , as previously discussed by @xcite .
this is borne out by a quantitative analysis : fig .
[ fig : angle ] shows the excess energy in the thermal cluster gas ( once again excluding non - thermal jet exhaust ) as a function of polar angle @xmath23 away from the jet axis ( measured around the cluster centre ) . at early times
, the excess energy is clearly peaked around the jet axis ( within about 30@xmath24),but at late times ( 160 myrs , solid black line ) , the distribution is consistent with being isotropic ( dotted black line ) .
this indicates that individual episodes of jet activity can , in fact , distribute their energy rather isotropically into the cluster .
more detailed simulations including radiative cooling and multiple jet episodes will be necessary to develop a detailed picture of how heating and cooling in realistic cluster atmospheres proceed .
we have presented simulations of jet - cluster feedback that integrate a correct representation of cluster dynamics , including dark matter , and the collimated input of energy and momentum from an agn .
the simulations accurately reproduce the observed morphological appearance of powerful radio sources like cyg a , if a jitter is imposed on the jet to account for unresolved dynamical instabilities of the jet commonly referred to as the `` dentist drill effect '' .
we find that the dynamic structure of the cluster in the form of rotation and shear in the velocity field and the inhomogeneity and anisotropy of the cluster have significant impact on the evolution of the jets and the radio lobes they inflate . at late times
, we find that the excess energy deposited into the cluster by the action of the jets is distributed well away from the axis , consistent with roughly spherically symmetric energy input . after long times ( > 160 myrs ) , the inner cluster has been sufficiently rearranged to essentially erase the low density channel blasted out by the jet and move enough material into the way of subsequent jet outbursts to couple efficiently with the inner cluster .
this solves the problem of low efficiency feedback found in simulations of spherically symmetric , static atmospheres @xcite .
|
we present numerical simulations investigating the interaction of agn jets with galaxy clusters , for the first time taking into account the dynamic nature of the cluster gas and detailed cluster physics .
the simulations successfully reproduce the observed morphologies of radio sources in clusters .
we find that cluster inhomogeneities and large scale flows have significant impact on the morphology of the radio source and can not be ignored a - priori when investigating radio source dynamics .
morphological comparison suggests that the gas in the centres of clusters like virgo and abell 4059 show significant shear and/or rotation .
we find that shear and rotation in the intra - cluster medium move large amounts of cold material back into the path of the jet , ensuring that subsequent jet outbursts encounter a sufficient column density of gas to couple with the inner cluster gas , thus alleviating the problem of evacuated channels discussed in the recent literature .
the same effects redistribute the excess energy @xmath0 deposited the jet , making the distribution of @xmath1 at late times consistent with being isotropic .
galaxies : clusters : general galaxies : jets galaxies : intergalactic medium
|
the phase volume in @xmath1-dimensional space - time for the decay process @xmath2 can be reduced to the integral @xmath3^{d/2 - 2}\theta(\phi),\ ] ] where @xmath4 , @xmath5 , @xmath6 are the three mandelstam variables and , in the physical region , the kibble cubic @xmath7 stays positive .
clearly @xmath8 is symmetric in the three masses .
however , when one eliminates one of the mandelstam variables ( or any linear combination ) one is left , for even @xmath1 , with an elliptic integral which is not _
explicitly _ symmetrical , although it must be so implicitly .
the aim of this article is to exhibit three routes for getting a satisfyingly symmetrical result in the @xmath9 case , where @xmath8 is nothing but the area of the dalitz - kibble plot @xcite divided by @xmath10 .
detailed derivations are not given , for lack of space , but will be published elsewhere .
the complicated expressions found by almgren @xcite and bauberger et al @xcite , obtained by finally integrating [ @xmath11 , @xmath12 below ] @xmath13 involve the complete elliptic integrals , @xmath14 , @xmath15 and @xmath16 , where @xmath17 the main point is that the result is symmetric under @xmath18 interchange but _ not obviously so _ under @xmath19 and @xmath20 interchanges .
the key to converting these expressions into symmetric form is to pass to the jacobian zeta - functions @xmath21 via @xmath22 with @xmath23 and to use the addition formula ( 142.01 ) of byrd and friedman @xcite : @xmath24 after numerous manoeuvres , including identification of the three angles @xmath25 \ ; , \ ] ] etc . and noting the remarkable connection @xmath26 one obtains the beautifully symmetric form : @xmath27 \nonumber \\ & & \hspace*{21 mm } + \sqrt{q_{++ } q_{-- } } \ ; k(k ) \biggl [ \frac{z(\varphi_1,k)}{\sin^2\varphi_1 } + \frac{z(\varphi_2,k)}{\sin^2\varphi_2 } + \frac{z(\varphi_3,k)}{\sin^2\varphi_3 } \biggl ] \biggl\ } .\end{aligned}\ ] ]
in the triangular phase plot , pictured in figure 1 , points @xmath28 refer to configurations where the mandelstam variables assume their minimum values , points @xmath29 where they attain their maximum values and the @xmath30-lines and @xmath31-lines are tangents at @xmath29 and @xmath28 . needing to integrate within the cubic outline @xmath32 , one
is guaranteed to obtain a symmetric answer by finding the area of the hexagon @xmath33 with @xmath30 vertices and cutting out the six @xmath34 crescents . now area @xmath35 \ ; , \ ] ] while a typical crescent has an area @xmath36\end{aligned}\ ] ] that is definitely given by an elliptic function .
these crescents are readily evaluated in the limit of large @xmath37 and of small @xmath38 by analytical means .
one finds to leading order @xmath39 \\ & & -\frac{m_2m_3 ^ 2p^2}{12(m_2+m_3)^4 } \left[12m_3 ^ 3 + 23m_2m_3 ^ 2 + 20m_2 ^ 2m_3 + 6m_2 ^ 3\right ] + { \cal o}(p)\ ; , \\ a_{o_tp_sq_{su}}&=&2q^2 \left [ m_1m_2-\sqrt{m_1m_2m_3(m_1+m_2+m_3 ) } \ ; \theta_{13}\right ] + { \cal o}(q^3)\ ; , \\ \sin\theta_{13}&=&\sqrt{\frac{m_1m_3}{(m_1+m_2)(m_2+m_3)}}\ ; , \quad { \rm etc.}\end{aligned}\ ] ] subtracting all crescents and noting that @xmath40 we get the full phase space area ( in the two respective limits ) , @xmath41 + { \cal o}(p)\ ] ] or @xmath42 higher order terms can be calculated quite systematically in this manner from the original @xmath43 integrals for the crescents .
this method relies on the observation that @xcite phase space is twice the imaginary part of the sunset feynman integral @xcite , possessing 3 internal lines , according to cutkosky s cutting rules .
thus @xmath44 , where in @xmath1-dimensions , @xmath45 \ ; , \\ { \rm i }
\delta_c(x|m_i ) & = & \frac{1}{(2\pi)^{d/2 } } \left(\frac{m}{r}\right)^{d/2 - 1 } k_{d/2 - 1}(mr ) \ ; , \qquad r=\sqrt{-x^2+{\rm i}\epsilon } \ ; .\end{aligned}\ ] ] the fourier integral leads to an integral of a bessel function @xmath46 of the first kind with a product of three @xmath47 functions of the second kind ; balanced progress may be achieved via the expansion , @xmath48 \ ; , \ ] ] which terminates for odd @xmath1 but is asymptotic for even @xmath1 . since for example ,
@xmath49 } { ( 8m_1m_2m_3 r^3)^{1/2 } } \left [ 1 + \frac{3}{8r}\left(\frac{1}{m_1 } \!+\ ! \frac{1}{m_2}\ ! + \ !
\frac{1}{m_3}\right ) \!+\ ! \ldots \right]\ ] ] we are left with a doable integral , that produces a series of hypergeometric functions , whose leading term is @xmath50 with @xmath51 , @xmath52 . as imaginary parts of hypergeometric functions are @xcite @xmath53 we deduce that @xmath54 and recognize the threshold behaviour found by other routes .
this is a particular case of the general @xmath55-body threshold behaviour , @xmath56 .
one last comment : it is best to input the massless propagators @xmath57 at the stage of fourier integration , rather than attempt to take the zero - mass limit of the above massive formulae , because that procedure is extremely tricky .
we are pleased to acknowledge support from the australian research council under grant number a00000780 .
dalitz , phil . mag . * 44 * ( 1953 ) 1068 .
kibble , phys . rev .
* 117 * ( 1960 ) 1159 .
b. almgren , arkiv fr physik * 38 * ( 1968 ) 161 .
s. bauberger , f. berends , m. bhm and m. buza , nucl . phys . *
b434 * ( 1995 ) 383 .
byrd and m.d .
friedman , _ handbook of elliptic integrals for engineers and physicists _ , springer , 1954 .
s. groote , j.g .
krner and a.a .
pivovarov , nucl . phys .
* b542 * ( 1999 ) 515 .
s. groote and a.a .
pivovarov , nucl . phys .
* b580 * ( 2000 ) 459 .
a. bashir , r. delbourgo and m.l .
roberts , j. math .
42 ( 2001 ) 5553 .
davydychev and v.a .
smirnov , nucl .
* b554 * ( 1999 ) 391 .
|
we derive expressions for three - body phase space that are explicitly symmetrical in the masses of the three particles , by three separate methods .
utas - phys-02 - 09 + hep - th/0209233 + july 2002 * three - body phase space : symmetrical treatments * a.i .
davydychev and r. delbourgo + @xmath0_school of mathematics and physics , university of tasmania , + gpo box 252 - 21 , hobart , tasmania 7001 , australia _ +
|
accurate and reproducible measurement of material properties at the nanoscale is the main goal of dynamic atomic force microscopy ( afm ) .
extraction of material properties from the measurable quantities in dynamic afm requires a deep understanding of both the tip - surface interaction and the dynamics of the afm cantilever when it is close to the sample surface .
we propose a method that uses fourier analysis to measure and compensate for background forces , which are long range and not local to the afm tip .
these interactions produce artifacts in the measurement of tip - surface force , leading to overestimation of its attractive and dissipative components .
background forces are observed when measuring the quality factor of a cantilever resonance , which drops by as much as @xmath0 when the tip - sample distance becomes comparable to the cantilever width ( fig .
[ fig : qf0 ] ) .
this phenomenon has been attributed to an additional squeeze - film damping force@xcite , arising when the fluid surrounding the cantilever is squeezed between the cantilever body and the sample surface .
we also observe a slight decrease in the resonance frequency @xmath1 , due to increased hydrodynamic load .
other long range forces appearing at tip - sample distances of a few micrometers have been attributed to electrostatic contributions@xcite .
we have also observed in different commercial afm systems , a dependence of the cantilever s acoustic excitation on the extension of the scanning z - piezo , resulting in a change of the drive force which can be mistaken as a long range interaction .
whatever their origin , be it hydrodynamic , electrostatic , or afm design , these background forces influence force reconstruction in dynamic afm . here
we describe how to compensate for these background forces , assuming that they all share the following properties : they are linear as shown by a lack of intermodulation distortion@xcite ; they act over a long range , comparable to the cantilever width ; and they do not depend on the @xmath2 tip position over the sample surface , as they originate from the cantilever body rather than being local to the tip .
we can easily compensate for any such background force to reveal the true tip - surface interaction , using linear response theory in the frequency domain .
due to its generality and ease of implementation , we expect our method to be readily applied to a variety of force reconstruction methods essential for afm researchers . and resonance frequency @xmath1 as a function of tip - sample distance over a homogeneous pdms surface .
the values are obtained by fitting the thermal noise power spectral density , and plotted as the shift relative to a tip - sample distance of @xmath3 .
the fitted @xmath4 drops by as much as @xmath0 , whereas the decrease in @xmath1 is three orders of magnitude smaller ( plotted values of @xmath1 are multiplied by @xmath5 ) .
the cantilever was a mikromasch hq : nsc15/albs . ]
when an afm cantilever is far away from the sample surface , its fundamental flexural mode is well - modeled as a linear system .
the frequency dependent linear response function @xmath6 relates the frequency components of any force @xmath7 to the resulting frequency components of the cantilever deflection @xmath8 : @xmath9 in our notation @xmath10 denotes a complex valued function of the real variable @xmath11 , the frequency . in particular
, @xmath8 is the fourier transform of @xmath12 . in the following
, we will drop the explicit @xmath11 dependence for the sake of compact notation . when the driven and measured frequency components of @xmath13 are concentrated around the cantilever resonance frequency ,
@xmath14 can be well - modeled as a dampened simple harmonic oscillator : @xmath15 the parameters @xmath16 , @xmath17 and @xmath4 are the mode stiffness , resonance frequency and quality factor , respectively .
these parameters , together with the optical lever responsivity , can be obtained by a non - invasive calibration procedure traceable to the measurement of the thermal fluctuations of the cantilever deflection when it is far from the surface@xcite . the drive force with multiple frequency components @xmath18
is applied to the cantilever by means of a shaker piezo in the case of acoustic actuation , or a pulsed laser beam in the case of photo - thermal excitation .
regardless of the means of excitation , the drive force is determined by measuring the cantilever motion far away from the surface ( larger than @xmath19 ) at what we call the `` free '' position ( fig .
[ fig : spectra](a ) ) . thus , we extract the driving force from a measurement of the cantilever free motion @xmath20 and the calibrated linear response function @xmath14 @xmath21 as the afm probe approaches the sample surface , background forces begin to affect the cantilever body when its separation from the surface becomes comparable to its width , as shown in fig . [
fig : qf0 ] .
background forces change the cantilever motion ( compare insets of fig .
[ fig : spectra](a ) and fig .
[ fig : spectra](b ) ) , but they are clearly linear , as seen from the lack of intermodulation@xcite in the spectrum of fig .
[ fig : spectra](b ) . when the afm tip starts interacting with the surface at what we call the `` engaged '' position ( fig .
[ fig : spectra](c ) ) , the measured motion is affected by all the forces at play @xmath22 where @xmath23 is the nonlinear tip - surface force , carrying all the information about the material properties , and @xmath24 are the background forces
. we can use eq .
( [ eq : f_drive ] ) to account for the drive force , but in oder to solve for the tip - surface force we must eliminate the background forces .
note that , while the components of the drive force @xmath18 can be treated as constant , the components of the background forces @xmath24 depend on the motion @xmath13 which changes from pixel to pixel .
for example , a squeeze - film damping force will depend on the velocity of the cantilever . motivated by experimental observation ( lack of intermodulation distortion )
, we treat the problem of a general linear background force without regard to its particular origin , by expressing it in terms of a linear response function @xmath25 @xmath26 equation ( [ eq : def_chi_bg ] ) allows for the calculation of @xmath24 for any motion @xmath13 .
our treatment assumes that there exists a linear differential equation of the cantilever deflection which describes the background forces .
this equation can in principle be very complicated , _ e.g. _ involve fractional derivatives and have many parameters , but we assume that it does not change as the probe scans over the sample , consistent with the idea that the background forces act on the body of the cantilever . lifting the probe slightly away from the surface , we find that the short ranged @xmath23 goes to zero with an abrupt drop in intermodulation , while the long ranged @xmath24 is barely affected .
we define the `` lift '' position as the closest distance for which the forces acting on the cantilever are linear ( see sect . [
sec : algorithm ] ) . at the lift position ( fig .
[ fig : spectra](b ) ) , the total force is given by the drive force @xmath18 and the linear background forces @xmath24 only .
the lift motion is therefore @xmath27 solving for @xmath24 gives @xmath28 combining eq .
( [ eq : f_drive ] ) , eq .
( [ eq : def_chi_bg ] ) , and eq . ( [ eq : f_bg_lift ] ) , we determine @xmath25 from the measured @xmath20 and @xmath29 @xmath30 going back to the engaged position , we can now compensate for the background forces in eq .
( [ eq : all_forces ] ) using @xmath25 from eq .
( [ eq : chi_bg ] ) .
thus we obtain the tip - surface force : @xmath31 equation ( [ eq : correction ] ) allows for the compensation of the background forces , and thus the calculation of the tip - surface force at every pixel of an afm image , provided the knowledge of @xmath18 and of @xmath25 , both being constant during the afm scan .
) for different number of drive tones @xmath32 is measured as function of the afm feedback amplitude set - point .
as the afm probe moves away from the surface ( increasing set - point ) , @xmath33 gradually decreases until the probe suddenly breaks free from the tip - surface interaction @xmath34 and a sharp drop in @xmath33 is observed .
the lift motion is measured at this drop . ] to accurately determine the linear response function of the background forces @xmath25 , we need to measure the cantilever motion @xmath29 as close to the surface as possible without tip - surface interaction .
we use the notion that the tip - surface force is strongly nonlinear , while the background forces are linear , as evidenced by the measurements of fig .
[ fig : spectra ] and fig .
[ fig : imd ] .
we apply a multifrequency drive with a number of discrete components @xmath32 at the set of frequencies @xmath35 with @xmath36 , _ i.e. _ @xmath37 is non - zero for @xmath38 . when the linear forces @xmath18 and @xmath24 act on the cantilever , the response @xmath8 will be non - zero only at @xmath38 ( see fig .
[ fig : spectra](a , b ) ) . on the other hand ,
when the cantilever is experiencing the nonlinear @xmath23 , response will arise at intermodulation product frequencies @xmath39 given by integer linear combinations of the drive frequencies : @xmath40 where @xmath41 is an integer ( see fig . [
fig : spectra](c ) ) .
we introduce the intermodulation distortion @xmath33 , as the ratio of the power at undriven frequencies to the power at driven frequencies : @xmath42 in principle , we want to measure @xmath29 at the minimum distance from the surface such that @xmath43 . in practice , however , we will always measure some non - zero noise power .
we therefore choose a threshold ( typically @xmath44 ) and measure @xmath29 at the minimum distance from the surface such that @xmath45 .
figure [ fig : imd ] shows the intermodulation distortion as a function of amplitude set - point measured for drive schemes with different number of drive tones @xmath32 . as the afm feedback set - point increases , a characteristic behavior is visible showing a gradual decrease of @xmath33 with increasing set - point , due to the decrease of the non - linear tip - surface interaction @xmath23 .
when the set - point reaches a value of about @xmath46 , a sharp drop in @xmath33 is observed , indicating a transition from an overall nonlinear force to an overall linear force .
this sharp transition allows for unambiguous measurement of @xmath29 , from which we calculate the linear response function of the background forces @xmath25 .
as discussed in sect . [ sec : algorithm ] , @xmath29 and @xmath20 will be non - zero only at the drive frequencies . calculating the linear response function of the background forces from eq .
( [ eq : chi_bg ] ) will therefore yield @xmath47 only at the drive frequencies @xmath38 . on the other hand , to apply the compensation to the measured data with eq .
( [ eq : correction ] ) we require the knowledge of @xmath47 at all the frequencies in the spectrum of engaged motion @xmath48 . to overcome this issue
we use the notion of narrow band measurement on a resonant system . due to the high q resonance in the cantilever linear response function
, the motion will be concentrated at frequencies close to the resonance frequency @xmath17 , within a narrow band @xmath49 @xmath50 where @xmath51 is the number of measured frequencies ( typically 32 ) and @xmath52 is typically chosen as the measurement bandwidth .
we perform a polynomial expansion of the complex function @xmath47 in this narrow band : @xmath53 where @xmath54 is the imaginary constant and @xmath55 and @xmath56 are sets of real coefficients to be determined .
a drive force with @xmath32 frequency components allows for the determination of up to @xmath57 coefficients , corresponding to two polynomial fits of degree @xmath58 of the real and imaginary parts of @xmath25 .
it is possible to perform a low degree fit with a high number of drives , @xmath59 , in which case the coefficients are obtained with a least square optimization method .
while a higher order fit could in principle describe a more complex @xmath25 , we find that a linear approximation of each of the two quadratures ( 4 coefficients , requiring 2 or more drive frequencies ) is sufficient to describe the background forces .
a higher order fit is not always numerically stable , and it can introduce artifacts in the compensated data .
equation ( [ eq : polyfit ] ) is quite general , allowing for a good approximation to any type of linear background force .
a special case of eq .
( [ eq : polyfit ] ) is a polynomial with only two coefficients of the form @xmath60 where @xmath16 is the mode stiffness and @xmath61 and @xmath62 are fit parameters . in this case
it can be shown that the background forces result in an effective cantilever with a renormalized linear response function @xmath63 of the form of eq .
( [ eq : chi_cant ] ) , where the resonance frequency and quality factor are given by @xmath64 this special case is often assumed when analyzing forces in dynamic afm@xcite .
we have shown how to mathematically treat the problem of compensating for arbitrary linear background forces , and we proposed a simple method to obtain their response function .
we now show an application of this method on soft material surfaces where background forces are typically rather large .
figure [ fig : ps_ldpe ] shows dynamic force quadratures@xcite on two areas of a polystyrene low - density polyethylene polymer blend ( bruker ) .
@xmath65 is the force in phase with the cantilever motion integrated over one oscillation cycle , representing the conservative forces experienced by the cantilever at different oscillation amplitudes .
@xmath66 is the force quadrature to the cantilever motion integrated over one oscillation cycle , showing the dissipative interaction of the cantilever with its environment and with the surface .
the force quadratures represent a direct transformation of the measured data without any model assumptions , providing a physically intuitive way of analyzing the measured cantilever dynamics in terms of conservative and dissipative interactions . at low amplitudes the uncompensated force quadratures ( dashed lines in fig .
[ fig : ps_ldpe](b , c ) ) show a positive slope in @xmath65 at low amplitude , the signature of a long range attractive force .
the negative slope in @xmath66 is a signature of a linear damping , in addition to the damping contained in @xmath14 which is calibrated far away from the surface .
some hysteresis is also present in both sets of curves , indicating that the background forces are not purely of the type described by the special case of eq .
( [ eq : polyfit_ho ] ) .
notably , the low amplitude background forces are the same for the two sets of curves , despite being measured over different areas of the sample with very different material properties , confirming that the background forces are not local to the afm tip .
we used a drive scheme with @xmath67 , measured @xmath29 over a polystyrene area of the sample ( black circle in fig .
[ fig : ps_ldpe](a ) ) , then calculated @xmath25 with @xmath68 in eq .
( [ eq : polyfit ] ) and applied its compensation to the measured data .
the solid lines in fig .
[ fig : ps_ldpe](b , c ) show the force quadratures compensated for background forces .
the slope at low amplitudes is now missing , as well as most of the hysteretic effects ( see inset of fig . [
fig : ps_ldpe](b ) ) .
we note that the compensation calculated over a polystyrene area of the sample has this effect not only for the force quadratures on the polystyrene , but also for those on polyethylene .
taken together , these observations confirm the validity of the assumption that while the background forces change for every pixel of the image , their linear response function does not change during the whole scan .
figure [ fig : hydrogel ] shows the same measurement and compensation procedure on a silicone hydrogel sample in a high humidity environment .
the sample presents alternating solid and liquid - like domains ( fig .
[ fig : hydrogel](a ) ) with very different mechanical response as shown by the peculiar shapes of the force quadratures ( fig .
[ fig : hydrogel](b , c ) ) .
also in this sample it is clear how the background force compensation corrects for the long - range attractive forces and increased dissipation , while at the same time preserving the interesting features of the tip - surface force and the peculiar hysteresis .
this will be discussed further in a forthcoming publication .
we have shown the effect of applying the background force compensation on the dynamic force quadratures @xmath65 and @xmath66 .
the compensation procedure is however general and can be applied to any force reconstruction method .
once the frequency components of the compensated tip - surface force @xmath23 are obtained from eq .
( [ eq : correction ] ) , @xmath23 can be fed to any force reconstruction algorithm without further modifications . as an example , in fig . [
fig : adfs ] we calculate the tip - surface force on the polystyrene and on the solid domain of the hydrogel sample with amplitude - dependent force spectroscopy@xcite ( adfs ) , with and without compensation for the background forces .
not compensating for background forces would lead to overestimating the adhesion force by @xmath69 for polystyrene , and by @xmath70 for the solid domain of the hydrogel .
the observed peak - force would be instead underestimated by about @xmath71 in both cases ( not shown in fig .
[ fig : adfs ] ) .
due to the nonlinear nature of the tip - surface interaction in dynamic afm , the dynamics is typically simulated by numerically integrating the differential equation for the cantilever deflection @xmath12 @xmath72 @xmath73 is a known function of time , whereas @xmath34 and @xmath74 are unknown .
a variety of models for @xmath34 are available@xcite to simulate different types of both conservative and dissipative tip - surface interaction as function of the cantilever motion @xmath12 and its velocity @xmath75 , and even on the effective position of a moving surface model@xcite . on the other hand ,
no general model for simulating background forces is available , due to the different types of interaction that can give rise to this effect . in the frequency domain ,
the background forces can be treated as an effective cantilever linear response function @xmath63 : @xmath76 where @xmath77 transforming @xmath63 into a differential equation is in general very difficult , however in the special case of eq .
( [ eq : polyfit_ho ] ) it is possible to simply replace @xmath17 and @xmath4 in eq .
( [ eq : simulation ] ) with @xmath78 and @xmath79 as defined by eq .
( [ eq : w0_eff ] ) and eq .
( [ eq : q_eff ] ) : @xmath80 alternatively , it is possible to more generally treat the effect of the background forces on the nonlinear response by introducing an effective drive force @xmath81 : @xmath82 where @xmath83 once the free , lift , and engaged motions are measured experimentally , eq . ( [ eq : f_eff ] ) is used to determine the effective drive force @xmath84 which is readily transformed to the time domain @xmath85 via the inverse fourier transform . the new differential equation @xmath86 can now be integrated numerically . a comparison of the simulated motion @xmath87 using eq .
( [ eq : sim_chi_eff ] ) or eq .
( [ eq : sim_f_eff ] ) , with the measured motion @xmath88 , allows for numerical optimization to find the best - fit parameters of a nonlinear tip - surface force model .
we derived a mathematical procedure to account for long - range background forces in dynamic afm , under the assumption of linear interaction and in the limit of a narrow band measurement . using intermodulation distortion to detect the onset of tip - surface forces
, we accurately measure the background forces at the driven frequencies and extrapolate their linear response function to undriven frequencies . applying our procedure to experimental data we demonstrated compensation for background forces on dynamic force quadratures and adfs force curves , measured on two different soft materials . given the generality of the compensation procedure and its ease of application to any type of dynamic force reconstruction
, our method will be very useful for the determination of material properties with quantitative dynamic afm .
the authors acknowledge financial support from the swedish research council ( vr ) , and the knut and alice wallenberg foundation .
gunilla hgg ( star - lens ab , ml , sweden ) is acknowledged for providing the silicone hydrogel sample .
14ifxundefined [ 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 10.3390/s90503854 [ * * , ( ) ] link:\doibase 10.1103/physreve.81.056305 [ * * , ( ) ] link:\doibase 10.1063/1.4962866 [ * * , ( ) ] , link:\doibase 10.1103/physrevb.66.035402 [ * * , ( ) ]
link:\doibase 10.1063/1.2909569 [ * * , ( ) ] link:\doibase 10.1063/1.4757398 [ * * , ( ) ] link:\doibase 10.1063/1.2162455 [ * * , ( ) ] link:\doibase 10.1088/0957 - 4484/18/43/435503 [ * * , ( ) ] link:\doibase 10.1021/la203434w [ * * , ( ) ] link:\doibase 10.1063/1.4882755 [ * * , ( ) ] link:\doibase 10.3762/bjnano.4.5 [ * * , ( ) ] link:\doibase 10.1038/ncomms2365 [ * * , ( ) ] @noop `` , '' ( ) link:\doibase 10.1039/c5sm02154e [ * * , ( ) ]
|
background forces are linear long range interactions of the cantilever body with its surroundings that must be compensated for in order to reveal tip - surface force , the quantity of interest for determining material properties in atomic force microscopy .
we provide a mathematical derivation of a method to compensate for background forces , apply it to experimental data , and discuss how to include background forces in simulation .
our method , based on linear response theory in the frequency domain , provides a general way of measuring and compensating for any background force and it can be readily applied to different force reconstruction methods in dynamic afm .
|
the dynamics of driven manifolds in random media presents many examples of non - equilibrium phase transitions .
the interest lies often in how an object interacts with a quenched disorder environment .
if the driving force is small enough and the temperature zero , the manifold , e.g. a domain wall in a magnet , or an interface between two phases , can get pinned .
this means , quite simply , that its velocity becomes zero .
therefore one can discuss the physics in terms of an order parameter ( velocity ) and a control parameter ( external force ) . at and close to the critical force value @xmath2
the driven interface can develop critical correlations due to the frustration that arises from the competition between , typically , elastic forces and the randomness of the medium @xcite .
an example is provided by the quenched kardar parisi
zhang ( qkpz ) equation , @xmath3 where @xmath4 is the interface height , @xmath5 a surface tension , @xmath0 the strength of the nonlinearity characterizing the kpz class @xcite , @xmath6 a driving force , and @xmath7 is the quenched noise . if @xmath0 is positive and stays finite in the limit @xmath8 , the depinning phase transition in 1 + 1 dimensions can be mapped to directed percolation ( dp ) via a geometrical construction . in this way
one can compute the critical exponents through the knowledge concerning the dp exponents , and the particular model where this is explicitly possible has been coined directed percolation depinning one for obvious reasons @xcite .
if @xmath0 is negative , however , the character of the transition changes completely .
1 shows an example of an interface at @xmath9 .
the depinning in this case was investigated by jeong , kahng , and kim @xcite , who showed that the transition becomes of a _ first - order _ kind and that this persists even in the presence of finite tilts below a critical @xmath10 .
the aim of this article is to use a new geometrical construction to discuss the morphology in the case depicted in fig . 1 , similarly in spirit to the dpd - mapping of the @xmath11-case .
the essential physics in the @xmath12 qkpz class , at and below @xmath2 arises from `` facet formation '' , the fact that the slopes of the triangular mounds are just dependent on the parameters of the equation , and from `` valley formation '' . a look at fig . 1 where the interface is shown at various times @xmath13 implies that it is these two mechanisms that are responsible for the final shape of an interface that gets pinned . in this article
we map the two - dimensional growth of the qpkz interfaces to a one - dimensional sub - monolayer polynuclear growth model ( png ) @xcite based on this picture of a statistically constant growth velocity of the facets , and random pinning events on the roughly speaking planar areas of growth .
this implies that one can understand the pinning of an interface in a quenched environment by considering a model with thermal noise , only .
the reason why this is possible is the first - order character of the transition : there is no diverging correlation length , and therefore the physics of the depinning can be understood based on a local picture of events .
notice that the png in the multilayer version is actually in the thermal kpz - class , and that the relation in our case between a growth model and the png is indirect . .
the snapshots have been taken every 1000th time step .
the vertical slicing lines designate the positions of `` valleys '' and `` peaks '' located by the algorithm described in the text . ]
section two of this paper introduces the actual numerical qkpz model and the relevant parameters . in section three ,
we discuss in detail the mapping between the qkpz and the png models including a measurement of the model parameters for use in the png model .
we also discuss the relevant parts of the png literature . in section four the free parameters ( growth velocity , pinning probability per unit time )
are discussed and computed , and a basic comparison of the qkpz interfaces with those that come from the png representation is presented .
section five finishes the paper with a discussion .
the model we focus on is a @xmath14 dimensional qkpz system with the spatial coordinate @xmath15 and time @xmath16 discretized , while the interface height @xmath4 is a continuous number .
we use periodic boundary conditions with flat initial configuration .
the forthcoming results have been obtained from systems in which the equation coefficients are invariably @xmath17 , @xmath18 , and @xmath19 .
the system is updated simultaneously for all lattice sites in every time step according to the finite differences version of the governing equation ( [ ekpz ] ) with @xmath20 : @xmath21 + { } \nonumber\\ { } + { \lambda \over 8}[h_{i+1}(t)-h_{i-1}(t)]^2+f+g \eta_i([h_i(t ) ] ) \nonumber\\ h_i(t+\delta t ) = \left\ { \begin{array}{ll } h_i(t ) + f_i \delta t & \textrm{if $ f_i(t)>0$}\\ h_i(t ) & \textrm{otherwise } \end{array } \right . \label{esim_1d_cont_lesch}\end{aligned}\ ] ] the value of the noise term @xmath22 at integer positions of the height field denoted by @xmath23 $ ] is chosen randomly from the interval @xmath24 $ ] with a uniform distribution and is unchanged in time .
the choice of the noise term is meant to mimic the simulation carried out in @xcite .
nevertheless , the particular form of @xmath22 has a mere quantitative influence on some measurables such as the critical driving force @xmath2 ; the qualitative behavior of the systems have been checked to yield consistent results independently of the various options for the representation of @xmath22 .
we study the morphology of slightly subcritical interfaces with the mapping to the png model in mind . to this end
we identify the macroscopic facets along the interface locating the `` valleys '' and `` peaks '' in the entirely evolved and stopped systems .
the algorithm to find the extrema is composed as follows . as a first step the pinned interface is scanned for all of the local minima and maxima positions .
subsequently after that , the pair of adjacent minimum / maximum with the smallest distance to each other is removed and this step repeated until the distance between the two closest extremal positions is no more than a specific parameter .
applying this procedure we eliminate the noise - like small height fluctuations of the interface .
the resolution of the facets depends on the actual choice of the smallest distance parameter and we use a few lattice units for that .
following the time evolution of some interface configurations like the one in fig .
[ snapshot ] , it suggests itself that the pinning events resulting in the valleys of the final morphology take place in a seemingly random way on the growing terraces .
the interface is observed to propagate with a uniform average vertical velocity at the plateaus of the facets , in other words , the height of the separated terraces of the interface is on the average on the same level at all times .
the straight sides of the facets mean that the horizontal distance between the end points of neighboring facets at the level of propagation grows in a linear fashion with time , following to first order from a balance between the nonlinearity and the driving force in eq .
( [ ekpz ] ) . if we consider the _
axis of fig .
[ snapshot ] as time and the sides of the final interface facets as phase boundaries in this model , we can outline a picture of a model mapping as follows .
we define a one dimensional lattice of cells on the grounds of the png model with two - state filling possibilities : a site is either occupied by a particle or empty , with an initially empty configuration . during the simultaneous updating steps of the lattice ,
a seed particle may be deposited to any of the empty cells with a certain probability in every time step .
the actual deposition probability and place may depend on various factors which will be discussed later . with a system size rescaling factor of @xmath25 .
] deposition of seeds creates growing islands with one particle added per time step at both ends according to the formal rule of @xmath26 @xmath27 denotes the filling state of the cell @xmath28 at time @xmath16 .
an occupied site is marked by 1 while an empty one is marked by 0 .
in other words ( [ efilling_rule ] ) means that a cell gets filled if either of its neighbors is already filled .
the system has periodic boundary conditions so @xmath29 for @xmath30 and @xmath31 for @xmath32 .
@xmath33 is the lattice size and the indices run from @xmath34 to @xmath33 .
it is obvious that if we associate the empty cells in this model with the still - growing parts of the interface and the filled cells with the uncovered regions , the time evolution of the system would give a very similar picture to the final configuration of fig .
[ snapshot ] , if the way the seeds are deposited is chosen the right way .
an example of the mapping is demonstrated by fig . [ png_mapping ] .
[ first_pinning ] suggests that the probability the first seed is introduced at a given time should decay exponentially in time , which is the earmark of a geometric process .
this means that the probability of a seed deposition ( the creation of the first pinned site ) has a constant probability per each time step .
if this is true , then the growth closely resembles the submonolayer polynuclear growth model ( png ) , or the kolmogorov avrami johnson
mehl ( kajm ) model @xcite .
( the png model is originally defined so that the growth itself takes place in many layers atop each other . ) of the first pinning height @xmath4 in the kpz system , rescaled by the system size of 500 , 1000 , and 2500 for @xmath35 , @xmath36 , and @xmath37 , respectively .
the driving force is @xmath38 .
the averaging is done for 10000 interface configurations in each case .
the rescaled distribution is found to decay exponentially with an associated exponent of @xmath39 in the range of the parallel line shown in the figure . ]
another crucial question is : how is the pinning probability or the nucleation probability in the png picture dependent on @xmath6 ?
[ force_function ] illustrates the progress of two quantities close to the estimated @xmath2 . both the average number of pinning events and the average height ( time ) of the first pinning event scale roughly as power laws with @xmath40 .
since the depinning transition is of first order , it is not a priori obvious how these quantities should behave as @xmath41 .
the most natural guess is that the pinning probability per site per time - step goes to zero continuously , which would indicate that @xmath42 and @xmath43 , as depicted in fig .
[ force_function ] , should approach infinity and zero , respectively with cut - offs imposed by the maximum time the interfaces are followed , and the system size , respectively . as function of the driving force @xmath44 for a system of 500 lattice units .
the saturation is caused by limitations imposed on the maximum time configurations are allowed to evolve because of practical reasons .
the inset shows the number of pinnings taking place . ] as function of the driving force @xmath44 for a system of 500 lattice units .
the saturation is caused by limitations imposed on the maximum time configurations are allowed to evolve because of practical reasons .
the inset shows the number of pinnings taking place . ]
the valleys and peaks are identified like described in section ii .
the side gradients of the facets are measured on both sides and are displayed for various system configurations in table [ tmorphology_res ] . the height and position of the pinning events
are also determined and recorded .
@xmath6 is chosen so that the system develops only a limited number of facets ; this way , the uncertainty of facet identification and measurement stays under control .
on the other hand , we are mainly interested in the behavior when @xmath45 . we do not determine @xmath2 with very great accuracy ; good estimates can be found in @xcite . according to table [ tmorphology_res ] , the facet side gradients are independent of system size for a given @xmath6 .
additionally , the left and right hand side slopes are the same as well , as expected in the absence of a global tilt .
in contrast to this , the number of pinning events taking place in the systems is subject to finite size scaling .
the error values associated with the averages are apparently large ( 3040% ) . for the next section where
the growth statistics are compared with the one - dimensional picture some results are needed about the interesting quantities . in the ordinary kajm theory islands
are nucleated on a line randomly , with a rate @xmath46 and grow at a constant velocity . to map to the kpz case one
needs a function that relates that velocity to the kpz mound slopes as discussed below .
if one has @xmath47 nucleated islands ( here pinned parts of the interface ) then the fraction of uncovered space ( here growing fraction of interface ) @xmath48 satisfies the rate equation @xmath49 with a proportionality factor chosen as unity .
it is easy to understand that in a large system the temporal fluctuations of the slopes of the growing kpz interface average out , to an average growth velocity in the kajm picture .
this holds if the slopes do not depend on e.g. the size of the terrace they are adjacent to which seems trivially true .
then the only factor that needs to be checked is whether the nucleation events happen randomly , with a constant density as in the simple kajm as discussed above . in the one - dimensional picture
, one can proceed by defining size distributions for the islands , for the gaps between the island , and finally the density of islands that contain a fixed number of seeds , @xmath50 .
the last one implies that such islands result from the coalescence of @xmath50 single - seed islands .
such distributions describe exactly the kpz interface in the kajm language ( assuming that the mapping to png / kajm works out ) .
the last of these distributions is hard to compute since it involves a solution of a smoluchowski - like equation for the densities for different @xmath50 .
we quote here some of the appropriate results from @xcite .
first , the island number density , @xmath51 corresponds to the number of valleys in the kpz picture ( as , also , a function of average interface height , @xmath52 ) .
second , the uncovered fraction gives directly the interface velocity after a proper normalization , and reads @xmath53 implying a fast decay of the growth .
following the considerations above , let us denote the probability of a seed deposition per lattice unit in the png model @xmath54 for every time step .
@xmath54 should be tuned appropriately so that the average number of seed depositions equals the average number of pinning events or valleys in the kpz system .
this is the only parameter that affects the `` morphology '' in this system , while another one , an artificial `` side gradient '' @xmath55 has also to be introduced to transform the time scale of the png model to length scale in the kpz counterpart .
notice that the virtual valleys and summits from the png model after the scale transformation are not rounded .
the presence of the surface tension term in the kpz equation counteracts such sharp corners , though .
still , if the strength of the surface tension coefficient @xmath5 is sufficiently small , the resulting interface is very similar to the one obtained from the png model . to achieve proper quantitative predictions and match between the kpz morphology statistics and the png picture
, we need to tune @xmath54 and @xmath55 so that measuring the first pinning height ( seed deposition time ) distribution , we end up with the same decay exponents .
we relate the general height in the kpz model @xmath56 to time in the png system @xmath57 as @xmath58 .
the probability of the first seed deposition follows a geometric process : @xmath59 if @xmath60 denotes the seed depositing probability for the entire length of the png system , that is , @xmath61 .
@xmath62 is the first pinning height ( the altitude of the lowest valley ) in the corresponding kpz system . comparing fig .
[ first_pinning ] to eq .
( [ geometric ] ) we find that the assumption of @xmath63 is valid .
the experimental value of the probability decay exponent @xmath64 in the kpz system should thus equal @xmath65 expressed by the png parameters .
the taylor series expansion of the later results in @xmath66 up to first order .
if the horizontal scale transformation between the kpz and png systems is given by @xmath67 , then @xmath55 can simply be calculated from the measured facet side gradients in the kpz system , summarized in table [ tmorphology_res ] . then if @xmath68 denotes the empirical value of the side slope , @xmath69 .
according to these , the normalized seed deposition probability @xmath54 is obtained as @xmath70 the discussion above assumed that the parameter matching is performed between two particular lattices of specific features , most importantly the number of cells in the systems . considering the data collapse of fig . [ first_pinning ] ,
though , an additional relation for @xmath64 emerges instantly : @xmath71 , provided that the equation parameters ( @xmath5 , @xmath0 , @xmath6 , @xmath72 ) are otherwise fixed . since @xmath68 does not explicitly depend on @xmath73 , and if @xmath74 , @xmath75 a further consequence of the fact that @xmath76 for the first pinning probability with a suitable scaling function @xmath77 , is that the general pinning probability per time unit in propagating fronts of a qkpz system is _ linearly proportional _ to the integrated local interface velocity , @xmath78 , @xmath79 using the parameters above it is possible to completely recover the exact scaling law of the first pinning event in a system conforming to the kpz equation .
it may still be uncertain , though , whether the subsequent pinning events follow the same governing rules as the first one , in particular the assumption that the seed creation probability is linearly proportional to the island length .
calculated for a kpz system according to the text .
since the interface propagation velocity is constant , @xmath4 can be appropriately rescaled so as to yield time in a png system .
the system size is 2500 , while the driving force is @xmath80 .
the number of averaging iterations is @xmath81 .
the theoretical function of @xmath82 is fitted to the empirical data points minimizing the quadratic error and giving the results of @xmath83 and @xmath84 .
the slight shift of the distribution maximum may come from the fact that an initially flat interface can not get pinned before the the necessary fluctuations develop . on the other hand
, the omission of very small facets can have an unwanted effect on the data as well . ] in order to verify this assumption , we perform an `` island counting '' measurement on the kpz ensembles with the ultimate goal of deriving the number of islands versus time statistics , @xmath47 .
islands are interpreted in a straightforward way : an island is created in the associated png system whenever a pinning event occurs , and two islands adjoin to form one at the facet summits of the interface , reducing the actual number of islands by one .
an illustration of the island construction is shown in fig . [ png_mapping ] .
this thought experiment leaves us with a way to construct the associated island number distribution in function of the time ( height ) in the kpz model .
the @xmath47 function for the individual samples is naturally composed of step functions taking on the appropriate integer constants between the respective time ( height ) intervals .
the png and kpz behaviors are in good correspondence , which is a strong evidence for the validity of our assumptions .
the experimental relation for @xmath47 calculated in kpz systems is shown in fig . [ island_number ] together with the curve fitted based on the kajm result , eq .
( [ n1 ] ) .
an associated quantity is the average height of the interface in time , which can be measured in a very simple way in contrast to @xmath47 mentioned in the preceding paragraphs .
let @xmath48 denote the uncovered fraction of a png system , eq .
( [ n2 ] ) .
thus , @xmath85 fig .
[ average_height ] shows the empirical relation found for @xmath86 and the adequately scaled error function for a given interface evolution . again
, the theory proves to be consistent with our experimental data to a high degree of confidence .
versus time @xmath16 , the latter measured in lattice updating steps . a system of @xmath81 lattice units is used .
the guide line is obtained from fitting @xmath87 in a least - squares sense with @xmath88 and @xmath89 . ]
in this paper we point out a mapping between interfaces in two - dimensional systems , described by the negative-@xmath0 quenched kardar parisi
zhang equation and a one - dimensional island growth model .
an investigation of kpz interfaces reveals that pinning processes give rise to a morphology with valleys and facets that are formed .
the facets are found to have straight edges with vanishing deviations on a macroscopic scale , as well as sharp corners .
this establishes a connection with the png or kajm model .
the correspondence is supported by evidence based on comparing statistical quantities in both ensembles .
considering the kpz model from this point of view , it can be stated that pinning events are the manifestation of a local nucleation process that is uncorrelated both in time and space .
the pinning probability for an interface segment which advances covering an area of @xmath90 only depends on @xmath90 in an exponentially decaying fashion .
the work presented here gives a geometric description of the @xmath12-depinning transition , in analogy to the directed percolation depinning - picture of the opposite , @xmath11-case .
there it is possible to extend the correspondence of the geometrical construction and the pinned interfaces at @xmath2 to higher dimensions @xcite . in our case
this still awaits further work , but one should note that the kajm model of island growth becomes much more difficult to handle , analytically , in @xmath91 @xcite .
.comparison table of interface morphology for close - to - critical systems with different external driving forces ; @xmath92 .
the `` pinnings '' column contains the average number of valleys found in the systems , normalized to 1000 lattice units .
the `` gradient left '' and `` gradient right '' columns show the average slope on the left hand side and right hand side of the facets , respectively .
the number of interface samples averaged is 10000 in each of the cases .
the errors are measured by the standard deviation of the respective quantities . [ cols="^,>,>,<,^,^",options="header " , ]
|
we develop a phenomenological mapping between submonolayer polynuclear growth ( png ) and the interface dynamics at and below the depinning transition in the kardar parisi
zhang equation for a negative non - linearity @xmath0 .
this is possible since the phase transition is of first - order , with no diverging correlation length as the transition is approached from below .
the morphology of the still - active and pinned configurations and the interface velocity are compared to the png picture .
the interface mean height scales as @xmath1 . _
pacs # 05.70.np , 75.50.lk , 68.35.ct , 64.60.ht_ 2 [ ]
|
projectile fragmentation reaction is used extensively to study the reaction mechanisms in heavy ion collisions at intermediate and high energies .
this is also an important technique for the production of rare isotope beams and is used by many radioactive ion beam facilities around the world .
the fragment cross sections of projectile fragmentation reactions using primary beams of @xmath7 , @xmath8 , @xmath0 and @xmath2 at 140 mev / nucleon on @xmath3 and @xmath4 targets have been measured at the national superconducting cyclotron laboratory at michigan state university@xcite .
the canonical thermodynamical model(ctm)@xcite has been used to calculate some of these fragment cross sections@xcite . in the present work
, an evaporation code has been developed and has been coupled with the canonical thermodynamical model .
ctm coupled with this secondary decay code is then used to analyze the isoscaling data from the projectile fragmentation reactions .
the lighter fragments produced from these reactions exhibit the linear isoscaling@xcite phenomena and our model calculation also strongly supports this observation .
the secondary fragments ( produced after applying the evaporation code on the canonical model ) also exhibit isoscaling like the primary fragments from the fragmentation reaction@xcite but the temperature needed to reproduce the experimental data with the secondary fragments is lower than that required by the calculation with the primary fragments .
the isoscaling parameters @xmath5 and @xmath6 as obtained in the present work from the model calculations agree closely with those obtained from the experimental data .
these parameters as obtained from the secondary fragments are lower in magnitude than those obtained from the primary ones .
this effect is also seen in the dynamical models @xcite though the reduction is much more there .
the isoscaling behaviour displayed by the fragments produced in the projectile fragmentation reactions and the effect of evaporation on it in the framework of the hipse model has been discussed recently in @xcite .
this paper is structured as follows .
first we describe the canonical model briefly in sec.ii . in the same section we also present the main features of the evaporation code which is used to calculate the secondary fragments . in sec.iii
we present the results .
the effect of sequential decay on the distribution of the isotopic fragments is discussed .
the isoscaling phenomena as displayed by the primary as well as the secondary fragments is also described in this section and are compared with the experimental data . in sec .
iv we present the summary .
in models of statistical disassembly of a nuclear system formed by the collision of two heavy ions at intermediate energy one assumes that because of multiple nucleon - nucleon collisions a statistical equilibrium is reached .
consequently , the temperature rises .
the system expands from normal density and composites are formed on the way to disassembly .
as the system reaches between three to six times the normal volume , the interactions between composites become unimportant ( except for the long range coulomb interaction ) and one can do a statistical equilibrium calculation to obtain the yields of composites at a volume called the freeze - out volume .
the partitioning into available channels can be solved in the canonical ensemble where the number of particles in the nuclear system is finite ( as it would be in experiments ) . in the next subsection
we describe the canonical model . in this section
we describe briefly the canonical thermodynamical model .
assume that the system with @xmath9 nucleons and @xmath10 protons at temperature @xmath11 , has expanded to a higher than normal volume and the partitioning into different composites can be calculated according to the rules of equilibrium statistical mechanics . in a canonical model , the partitioning
is done such that all partitions have the correct @xmath12 ( equivalently @xmath13 ) .
details of the implementation of the canonical model can be found elsewhere @xcite ; here we give the essentials necessary to follow the present work .
the canonical partition function is given by @xmath14 here the sum is over all possible channels of break - up ( the number of such channels is enormous ) which satisfy @xmath15 and @xmath16 ; @xmath17 is the partition function of one composite with neutron number @xmath18 and proton number @xmath19 respectively and @xmath20 is the number of this composite in the given channel .
the one - body partition function @xmath17 is a product of two parts : one arising from the translational motion of the composite and another from the intrinsic partition function of the composite : @xmath21 here @xmath22 is the mass number of the composite and @xmath23 is the volume available for translational motion ; @xmath23 will be less than @xmath24 , the volume to which the system has expanded at break up .
we use @xmath25 , where @xmath26 is the normal volume of nucleus with @xmath10 protons and @xmath27 neutrons . in this calculation
we have used a fairly typical value @xmath28 .
the probability of a given channel @xmath29 is given by @xmath30 the average number of composites with @xmath18 neutrons and @xmath19 protons is seen easily from the above equation to be @xmath31 the constraints @xmath15 and @xmath16 can be used to obtain different looking but equivalent recursion relations for partition functions@xcite .
for example @xmath32 these recursion relations allow one to calculate @xmath33 we list now the properties of the composites used in this work .
the proton and the neutron are fundamental building blocks thus @xmath34 where 2 takes care of the spin degeneracy . for deuteron , triton ,
@xmath35he and @xmath36he we use @xmath37 where @xmath38 is the ground state energy of the composite and @xmath39 is the experimental spin degeneracy of the ground state .
excited states for these very low mass nuclei are not included . for mass number @xmath40 and greater
we use the liquid - drop formula . for nuclei in isolation , this reads ( @xmath22 ) @xmath41\end{aligned}\ ] ] the derivation of this equation is given in several places @xcite so we will not repeat the arguments here .
the expression includes the volume energy , the temperature dependent surface energy , the coulomb energy and the symmetry energy .
the term @xmath42 represents contribution from excited states since the composites are at a non - zero temperature .
we also have to state which nuclei are included in computing @xmath33 ( eq.(17 ) ) . for @xmath43 ,
( the neutron and the proton number ) we include a ridge along the line of stability .
the liquid - drop formula above also gives neutron and proton drip lines and the results shown here include all nuclei within the boundaries .
the long range coulomb interaction between different composites can be included in an approximation called the wigner - seitz approximation .
we incorporate this following the scheme set up in @xcite .
the statistical multifragmentation model described above calculates the properties of the collision averaged system that can be approximated by an equilibrium ensemble .
ideally , one would like to measure the properties of excited primary fragments after emission in order to extract information about the collisions and compare directly with the equilibrium predictions of the model .
however , the time scale of a nuclear reaction(@xmath44 ) is much shorter than the time scale for particle detection ( @xmath45 ) . before reaching the detectors , most fragments decay to stable isotopes in their ground states .
thus before any model simulations can be compared to experimental data , it is indispensable to have a model that simulates sequential decays .
a monte carlo technique is employed to follow all decay chains until the resulting products are unable to undergo further decay . for the purposes of the sequential decay calculations
the excited primary fragments generated by the statistical model calculations are taken as the compound nucleus input to the evaporation code .
hence , every primary fragment is decayed as a separate event .
we consider the deexcitation of a primary fragment of mass @xmath46 , charge @xmath47 and temperature @xmath11 . the succseesive particle emission from the hot primary fragments is assumed to be the basic deexcitation mechanism .
for each event of the primary breakup simulation , the entire chain of evaporation and secondary breakup events is monte carlo simulated .
the standard weisskopf evaporation scheme is used to take into account evaporation of nucleons , @xmath48 , @xmath49 , @xmath50 and @xmath5 .
the decays of particle stable excited states via gamma rays were also taken into account for the sequential decay process and for the calculation of the final ground state yields .
we have also considered fission as a deexcitation channel though for the nuclei of mass @xmath51 100 its role will be quite insignificant .
the process of light particle emission from a compound nucleus is governed by the emission width @xmath52 at which a particle of type @xmath53 is emitted . according to weisskopf s conventional evaporation theory @xcite , the partial decay width for emission of a light particle of type @xmath53
is given by @xmath54 here @xmath55 is the mass of the emitted particle , @xmath56 is its spin degeneracy .
@xmath57 is the particle separation energy which is calculated from the binding energies of the parent nucleus , daughter nucleus and the binding energy of the emitted particle and the liquid drop model is used to calculate the binding energies .
the subscript @xmath53 refers to the emitted particle , @xmath58 refers to the parent nuclei and @xmath59 refers to the residual(daughter ) nuclei .
@xmath60 @xmath1 @xmath61 are the level density parameters of the parent and residual nucleus respectively .
the level density parameter is given by @xmath62 and it connects the excitation energy @xmath63 and temperature @xmath11 through the following relations .
@xmath64 where @xmath65 @xmath1 @xmath66 are the temperatures of the emitting(parent ) and the final(residual ) nucleus respectively .
@xmath67 is the coulomb barrier which is zero for neutral particles and non - zero for charged particles . in order to calculate the coulomb barrier for charged particles of mass @xmath68 we use a touching sphere
approximation@xcite , @xmath69 where @xmath70 is taken as 1.44 m .
@xmath71 is the geometrical crosssection ( inverse cross section ) associated with the formation of the compound nucleus(parent ) from the emitted particle and the daughter nucleus and is given by @xmath72 where , @xmath73 where @xmath74 = 1.2 fm . for the emission of giant dipole @xmath75-quanta
we take the formula given by lynn@xcite @xmath76 with @xmath77 with @xmath78 , and @xmath79 and @xmath80 are the position and width of the giant dipole resonance . for the fission width
we have used the simplified formula of bohr - wheeler given by @xmath81 where @xmath82 is the fission barrier of the compound nucleus given by@xcite @xmath83 once the emission widths are known , it is required to establish the emission algorithm which decides whether a particle is being emitted from the compound nucleus .
this is done @xcite by first calculating the ratio @xmath84 where @xmath85 , @xmath86 and @xmath87 or fission and then performing monte - carlo sampling from a uniformly distributed set of random numbers . in the case that a particle is emitted , the type of the emitted particle is next decided by a monte carlo selection with the weights @xmath88 ( partial widths ) .
the energy of the emitted particle is then obtained by another monte carlo sampling of its energy spectrum .
the energy , mass and charge of the nucleus is adjusted after each emission .
this procedure is followed for each of the primary fragment produced at a fixed temperature and then repeated over a large ensemble and the observables are calculated from the ensemble averages . .
the number and type of particles emitted and the final decay product in each event is registered and are taken into account properly keeping in mind the overall charge and baryon number conservation .
first we will show our calculations for @xmath0 on @xmath3 reaction and @xmath2 on @xmath3 reaction . in the model
, the target imparts a certain amount of energy to the projectile transforming it to a projectile like fragment(plf ) with a temperature .
this excited plf will then expand and form composites during the expansion .
the partioning of the plf into different composites is done by the rules of equilibrium statistical mechanics in a freeze - out volume .
we consider production of different isotopes from the statistical breakup of the dissociating system .
if @xmath89 is the average number(multiplicity ) of composites with @xmath90 neutrons and @xmath91 protons , then the cross - section for this composite is @xmath92 , where c is a constant not calculable from the thermodynamic model .
it depends upon the dynamics that are outside the scope of this model . to be able to compute @xmath89 we need to know the mass and charge of the plf and its temperature .
the source sizes adopted for this calculation are zero order guesses
. it could be sometimes smaller or greater depending on the diffusion from the target . for @xmath2 or @xmath0 on @xmath3 which is a small target the choice of the mass and the charge of the plf is limited
. it can be slightly less than that of the projectile to as large as that of the projectile plus @xmath3 , the last being the case when the much larger projectile swallows the small target and drags it along retaining plf features .
similarly we have some limits on energy imparted(this fixes the temperature ) .
this energy can be small or upto the upper limit .
the upper limit is given by the case of projectile swallowing be and all the energy transforming into internal excitation(no part going into collective flow ) .
in the canonical calculation , the dissociating system is taken to be @xmath93 and for the other reaction the dissociating system is taken to be @xmath94.all composites between drip lines are included as detailed in sec.iia with the highest values of n , z terminating at @xmath27 , @xmath10 .
the temperature is taken to be 5.8 mev for both the reactions .
1 displays the isotopic distribution for z=12(magnesium ) produced from both the the reactions .
the dashed lines correspond to the distributions of the primary fragments while the solid lines correspond to the distributions after sequential decay . as expected , the more neutron rich system with @xmath95 ( right panel ) produces more neutron rich isotopes than the neutron deficient system with @xmath96 ( left panel ) . in all cases ,
the primary distributions are much wider and more neutron rich than the final distributions .
the peak positions of the isotopic distributions of both the primary and the secondary fragments coincide in case of the neutron deficient system as seen from the left panel of the figure . in case of the neutron rich system(right panel )
the peak of the distribution of the secondary fragments has shifted to the left with respect to that of the primary .
the experimental isotopic distributions(solid squares with error bars ) agree much more with the final results obtained after secondary decay than with the primary distributions .
the width and peak position of the isotopic distribution after the secondary decay agrees very well with the experimental data .
the model also successfully reproduces the rapid fall in crosssection for large neutron number .
we will now discuss the results about isoscaling .
it is observed from the experimental data@xcite that the light fragments emitted from the @xmath0 and @xmath2 systems exhibit the linear isoscaling behaviour represented by the equation @xmath97 where the isoscaling ratio @xmath98 is factored into two fugacity terms @xmath5 and @xmath6 , which contain the differences of the chemical potentials for neutrons and protons of the two reaction systems .
@xmath99 refers to the yield of fragment(n , z ) from system 2 which is usually taken to be the neutron - rich one and @xmath100 refers to the same from system 1 .
c is a normalization factor of the isoscaling ratio .
it is observed from our model that both the primary as well as the secondary fragments exhibit isoscaling .
2 shows the isoscaling results for ni on be system for the primary fragments . the ratio @xmath101 is plotted as function of the neutron number from z = 6 to z= 13 in the left panel whereas the right panel displays the ratio as function of the proton number z from n=8 to n=15 .
it is seen that the primary fragments exhibit very well the linear isoscaling behaviour for the lighter fragments over a wide range of isotopes and isotones .
the lines in the figures are the best fits of the calculated @xmath101 ratios(open triangles ) to eq.15 .
they are essentially linear and parallel on the semi log plot .
3 displays the isoscaling results for the secondary fragments .
the open triangles are the results obatined from our model while the solid squares with the error bars are the experimental ratios .
we have shown the isoscaling results for the even z and odd z isotopes in two separate panels for the sake of clarity .
while comparing with the results of the primary fragments in fig 2 , it is evident that the isoscaling is valid for a limited range of isotopes for the secondary fragments as compared to the primary ones .
when the isotopes away from the valley of stable nuclei are considered , the trends for the secondary fragments are not as clearly consistent with the isoscaling law as are the trends of the primary distribution .
one can conclude from this that isoscaling is approximately valid in the case of the secondary fragments .
the lines in the figures are the best fits of the calculated @xmath101 ratios which agree closely with the experimental data .
the temperature required to reproduce isoscaling data with the primary fragments is about 8 mev@xcite whereas that required for the secondary fragments is 5.8 mev .
this decrease was already predicted in one of our earlier papers@xcite in sec.9 . it has also been found out by ono et al.@xcite that the effect of secondary decay is to decrease the isoscaling parameter @xmath5 by about @xmath102 .
this is indeed what emerges from our calculations after including secondary decay code with the canonical model though the amount of reduction is less as compared to the dynamical model .
tha value of @xmath5 as obtained from the fits of the primary fragments ( left panel of fig .
2 ) is 0.713 whereas that obtained from the secondary fragments is 0.580 which is much closer to the experimentally obtained value for @xmath5@xcite equal to 0.566 . in fig .
4 we have also plotted the isoscaling ratios for different neutron number values as function of the proton number z and thereby calculated the other isoscaling parameter @xmath6 from them . the value of @xmath6 as obatined from the linear fits of the primary fragments ( right panel of fig .
2 ) is -0.849 . for the secondary fragments ,
the value of @xmath6 from our model is -0.634 whereas the experimentally obtained value is -0.621 .
the value of @xmath6 also decreases in the case of secondary from the primary ones as in the case of @xmath5 .
the results obtained after the sequential decay matches very well with the experimental ratios . as for the isotopes in fig 3 , for the isotones also it is seen that the linear isoscaling is valid for a limited range in case of secondary fragments as compared to the primary ones(right panel of fig.2 ) .
we now turn to the case where the target is @xmath4 .
one can consider the extreme limit which is target independence and by which we mean that @xmath27 , @xmath10 refers to simply to the case where just the projectile is the disintegrating system .
the more likely scenario where some matter has diffused to or from the target has many possibilities . in principle
, the target could shear away some material from the projectile leaving a plf which is a fraction of the projectile . for a peripheral collision
this is less likely than the alternative of the projectile picking up part of nuclear matter from the tail region of the much larger target .
the amount of nuclear matter curved out of ta will be small(for the disintegrating system to retain plf characteristics ) but other than that not much can be said and an integration over the different possibilities might be essential . keeping this limitation in mind , we compared with different possible scenarios and the case with projectile plus 10 neutrons and 8 protons from the target , i.e ,
projectile + @xmath103 yielded the best results .
it is also seen that target independence gives worse results than this case .
the agreement of our model calculations with experimental data is pretty good in this case .
the temperature used for this reaction is 6.2 mev .
fig . 5 and fig .
6 are similar to fig .
3 & fig . 4 except for the fact that they are for the case where the target is @xmath4 instead of @xmath3 . the theoretical slope @xmath5 as obtained from fig
. 5 from the fit of the secondary fragments is 0.459 while the experimentally obtained slope is 0.432 .
the straight line fit to the calculated points matches nicely with the experimental ratios .
the value of @xmath6 as obtained from our model from the slopes in fig .
6 is -0.489 whereas the experimentally obtained value is -0.487 .
thus we find that the values of the isoscaling parametrs as obtained from the secondary fragments for both the targets agree quite well with the experimental values as can be seen from table 1 . .
best fit values of the isoscaling parametrs @xmath5 and @xmath6 for the two targets @xmath3 and @xmath4 .
the values obtained from the slope of the primary and secondary fragments as well as the experimental values are tabulated . in the second column the range of @xmath47 or @xmath104 values used to calculate the parameters are indicated . [
cols="^,^,^,^,^,^",options="header " , ]
this work deals with the developing of the sequential decay code and successfully coupling it with the canonical thermodynamical model in order to compare the properties of the secondary fragments with the experimental data .
the width , peak position and rapid fall in cross - section of the isotopic distribution of the secondary fragments matches well with the experimental data .
the main purpose is to examine the effects of sequential decay on the phenomenon of isoscaling .
the secondary fragments also shows isoscsling and the the isoscaling parameters calculated from the secondary fragments matches closely with the experimental data .
the temperature required to reproduce the experimental data with the secondary fragments is less than needed by the primary ones .
we finally conclude that the canonical thermodynamical model can explain the isoscaling properties of the lighter fragments produced in the projectile fragmentation reaction .
the authors gratefully acknowledges important discussions with prof .
subal dasgupta .
they are thankful to prof .
m.b . tsang and dr .
valuable suggestions from dr .
santanu pal and jhilam sadhukhan is also acknowledged gratefully . ) of multiplicities of primary fragments of producing the nucleus @xmath107 where reaction 1 is @xmath108ni on @xmath105be and
reaction 2 is @xmath106ni on @xmath105be .
the left panel shows the ratios as function of neutron number @xmath104 for fixed @xmath47 values from 6 to 13 , while the right panel displays the ratios as function of proton number @xmath47 for fixed neutron numbers from @xmath104 = 8 to 15 .
the lines drawn through the theoretical points(open triangles ) are best fits of the calculated ratios .
the temperature used for both the reacions is 5.8 mev.,width=576,height=432 ] where reaction 1 is @xmath108ni on @xmath105be and reaction 2 is @xmath106ni on @xmath105be compred with the ratios of the experimental cross sections of the same two reactions .
the left panel shows the even @xmath47 isotopes while the right panel shows the results for the odd ones .
the lines drawn through the theoretical points(open triangles ) are best fits of the calculated ratios .
the experimental points are shown by solid squares with error bars.the temperature used for both the reacions is 5.8 mev .
, width=576,height=432 ]
|
the projectile fragmentation reactions using @xmath0 @xmath1 @xmath2 beams at 140 mev / n on targets @xmath3 @xmath1 @xmath4 are studied using the canonical thermodynamical model coupled with an evaporation code .
the isoscaling property of the fragments produced is studied using both the primary and the secondary fragments and it is observed that the secondary fragments also respect isoscaling though the isoscaling parameters @xmath5 and @xmath6 changes .
the temperature needed to reproduce experimental data with the secondary fragments is less than that needed with the primary ones .
the canonical model coupled with the evaporation code successfully explains the experimental data for isoscaling for the projectile fragmentation reactions .
|
one of the current theoretical challenges in quantum chromodynamics ( qcd ) is to describe high energy interactions with hadrons in terms of fundamental field theory .
it is observed that high - energy hadron - hadron scattering interactions become completely absorptive ( black ) at small impact parameters so that elastic scattering can be viewed essentially as a shadow of the inelastic cross section in the sense of babinet s principle .
if this regime occurs at most of the impact parameters which contribute to the inelastic cross section , then the elastic and inelastic cross sections are equal .
this limit is often referred to as the black body limit ( bbl ) in analogy with the quantum mechanical situation of scattering from an absorptive share of radius @xmath11 in which case the total cross section is equal to @xmath12 , ( see e.g. , problem 1 of section 131 in ref .@xcite ) .
this limit is also loosely referred to as the unitarity limit , although unitarity alone admits cross sections as large as @xmath13 provided there are no inelastic interactions ( see e.g. , problem 2 of section 132 in ref .
@xcite ) . in this paper we will assume , in line with experiment , that the amplitude is predominantly imaginary . in this case , the unitarity limit coincides with the bbl .
the black body limit in deep inelastic scattering ( dis ) is an interesting new regime in qcd where the coupling strength is small , but where the leading twist approximation breaks down and new small coupling methods are needed . in order to understand the transition to the bbl ,
it is important to construct models which accurately describe @xmath2 scattering over a wide range of kinematic variables . in particular
, such a model should interpolate smoothly between the nonperturbative domain and the region where interactions are accurately described by leading twist perturbative qcd ( pqcd ) . in a recent paper by munier _
@xcite , an estimate of the proximity to the bbl for the interaction of a color dipole with a proton was made using data from diffractive electroproduction of @xmath7-mesons @xcite .
the techniques used in @xcite are limited by the need to model the @xmath7-meson wave function , as well as by limited information on the @xmath1-dependence for @xmath7 production . within the impact parameter representation
, this means that their predictions are constrained to intermediate values of impact parameter ( @xmath14 fm ) @xcite .
( it will be important to keep in mind the distinction between impact parameter , @xmath15 , and dipole size , @xmath16 . )
we will find that our analysis is valid down to very small values of @xmath15 and @xmath16 .
we will make an improved estimate of the onset of the bbl by combining information from dis and @xmath0 production .
a great advantage of our technique is that , unlike models which are restricted to using meson production data , our model is valid down to very small dipole sizes because we utilize leading twist pqcd for calculations involving small size @xmath17 pairs . within our model of @xmath2 interactions
, we always assume that the virtual photon can be written as a linear combination of hadronic states .
furthermore , we relate the hardness of the interaction to the size of the hadronic state as is done in ref .
@xcite , and we take into account the dependence of the hadronic interaction on the virtuality of the original photon .
large size configurations constitute the soft component of the interaction whereas small size @xmath17 pairs constitute the hard component . modeling the @xmath2 interaction
requires making three principle observations .
the first is to recall that the total cross section for small size dipole configurations ( @xmath18 fm ) is known within the framework of perturbation theory down to @xmath6 .
therefore , we will have a complete picture of the interaction of small size configurations with nucleons if we can extract the @xmath1-dependence from experimental data . in sect .
[ sec : model ] , we discuss how the @xmath1-dependence of the small size @xmath17 pairs can be extracted from @xmath19 production data .
secondly , we note that the soft scattering of large size hadronic configurations is well understood phenomenologically in terms of effective pomeron exchange .
hence , our model should reproduce the pion - nucleon amplitude for hadronic configurations comparable to the size of the pion ( @xmath20 fm ) . finally , we must model the behavior of the amplitude for intermediate hadronic sizes ( @xmath21 fm @xmath22 fm ) .
this is an interesting and poorly understood region of kinematics , and our model will allow for readjustments in the transition region . in ref .
@xcite , a system was devised for relating the transverse size of a @xmath17 dipole to the virtuality , @xmath23 .
the pqcd result for the inelastic @xmath2 cross section was interpolated to large size hadronic configurations and matched to the total cross section for pion - nucleon scattering .
the hadronic size , @xmath16 , is used to interpolate between the hard and soft regions in this paper as it was in ref .
@xmath16 represents the transverse size of a quark - antiquark pair in the limit that @xmath16 is very small ( @xmath24 fm ) .
however , as @xmath16 grows large , the dipole picture becomes inappropriate since the hadronic components that correspond to soft interactions consist of large , complex hadronic states .
larger values of @xmath16 should be interpreted as transverse sizes of general hadronic components of the virtual photon wave function .
the cross section takes the form familiar from perturbation theory : @xmath25 here , @xmath26 is a universal scaling ansatz used to relate energy scales , @xmath4 , to transverse dipole sizes , @xmath16 ; @xmath27 is the light - cone fraction for the gluon attached to the @xmath17 loop .
one manifestation of the qcd factorization theorem is that the contribution of hadronic configurations within the photon to the longitudinal cross section , @xmath28 , is peaked around a narrow range of small dipole sizes ( see fig .
[ fig : dist ] ) .
the value of @xmath29 is chosen so that , within the perturbative region , @xmath30 is approximately the average dipole size contributing to @xmath28 . for large @xmath4 , @xmath29 takes on values of the order of 10 .
in fact , it is found that @xmath31 and @xmath32 depend very weakly on the value of @xmath29 within the perturbative region of @xmath16 @xcite .
changing the value of @xmath29 thereby provides a universal parameter for tuning the cross section within the transition region . in this paper , we use @xmath33 because it is found that @xmath33 best describes @xmath0 data over a wide range of kinematics ( both perturbative and nonperturbative ) .
figure [ fig : lambda ] compares the profile function for @xmath33 and @xmath34 .
the function used in ref .
@xcite to interpolate between the hard and soft regions matches smoothly to the pqcd result at small @xmath16 and to the pion - proton cross section at large @xmath16 .
furthermore , it takes into consideration the breakdown of the leading twist formula in the small @xmath3 limit . notice that the @xmath17 pair is not a fundamental object since it is always produced off - shell by a virtual photon
this is taken into account on the right side of eq .
( [ eq : pqcd ] ) through implicit dependence of @xmath35 , on @xmath23 and @xmath4 ( see ref . @xcite and eq .
( [ eq : xprime ] ) ) .
the dependence of the dipole cross section on the external photon virtuality , @xmath4 , is a feature that is absent in other models such as the one proposed by wsthoff and golec - biernat @xcite .
this point will be important for what we discuss later because it means that we can not speak unambiguously about the dipole cross section with referring to the interaction for which it is a sub - proccess . in this paper , we will model the @xmath1-dependence of the @xmath36 elastic scattering amplitude ( which , of course , can not be observed experimentally ) using data from @xmath37 photo(electro)-production in conjunction with the pion - proton elastic scattering amplitude .
as when we model the behavior of the total cross section in various kinematic regions , we model the @xmath1-dependence of the @xmath38 amplitude by considering three distinct steps .
first , we model the @xmath1-dependence of the small size @xmath39 configurations .
because of transverse squeezing in the @xmath37 wave function , data taken from @xmath37 production is appropriate for use in modeling the @xmath1-dependence of small size , hard scale @xmath39 configurations .
next , the @xmath1-dependence of soft , large size hadronic configurations is approximated by the pion structure function , the pion being a reasonable approximation to a large size hadronic component of the virtual photon wave function . for soft interactions , a factor for soft pomeron exchange
is included to account for the slow rise in cross section at small @xmath3 .
finally , we use the transverse size , @xmath16 , as a parameter to interpolate between hard and soft physics . in our analysis
, we will transform our expression for the amplitude into the impact parameter picture to look for the regions where the impact parameter space amplitude approaches the unitarity limit , and thus to estimate the values of the impact parameter where the bbl is attained .
it should be noted that our model has limitations which restrict how it can be applied .
it is important to keep in mind that eq .
( [ eq : pqcd ] ) is multiplied by a color factor of 9/4 when the hadronic configuration is a color singlet composed of octet representations of su(3 ) , as in a gluon dipole .
such interactions are expected to be abundant in the small @xmath3 limit , so the bbl will be reached at larger impact parameters than what is predicted by considering only the interactions of small size @xmath17 pairs . furthermore , at very small @xmath3 , effects from cross section fluctuations of the virtual photon become important , and taking into account only elastic dipole scattering becomes inappropriate .
in particular , the total cross section in hadron - hadron scattering has a significant contribution from inelastic diffraction . in ref .
@xcite , miettinen and pumplin write the contribution from inelastic diffraction in terms of fluctuations around hadronic eigenstates .
estimates of the contribution from hadronic fluctuations to the cross section show that it is not negligible ( see , e.g. ref .
in fact , with a decrease of the dipole size , the relative importance of inelastic diffraction increases .
the importance of inelastic diffraction for small @xmath39 sizes is discussed , for example , in ref .
the important point to note here is that inelastic diffraction will contribute to the breakdown of the leading twist approximation at low @xmath3 before the bbl for elastic scattering of the hadronic configurations in the photon wave function is reached . as such , we do not seek to place an absolute boundary on the region where corrections to the dglap evolution equation become relevant .
rather , we construct a model that puts a lower limit in impact parameter space on regions approaching the bbl .
furthermore , since the leading twist approximation is not accurate in the vicinity of the unitarity limit , the bbl establishes a lower limit in impact parameter space on regions where the dglap equation is applicable . in sect .
[ sec : model ] , we outline our model for the @xmath1-dependence of the hadronic configuration - nucleon amplitude . in sect .
[ sec : impact ] , we transform our expression for the amplitude into the impact parameter representation and study the proximity of the profile function to the unitarity limit as a function of impact parameter , @xmath15 .
we write the transverse and longitudinal cross sections , @xmath40 , as convolutions of the dipole cross section with the photon wave function in sect . [
sec : gamma ] .
section [ sec : gamma ] concludes with a calculation of the fraction of the @xmath38 cross section due to large values of the hadronic profile function . in sect .
[ sec : compare ] , we compare our results to an earlier study of the s - matrix @xmath1-dependence in impact parameter space . in sect .
[ sec : nuclear ] we perform the same calculation for the situation where the target is a @xmath8pb nucleus .
for the case of a nuclear target , we match the pqcd calculation at small @xmath16 to the glauber multiple scattering theory result at large @xmath16 .
finally , we summarize our observations in the conclusion .
starting with the expression for the total cross section in eq .
( [ eq : pqcd ] ) , we devise a model for the scattering amplitude by writing it in the form , @xmath41 where @xmath42 accounts for the @xmath1-dependence of the interaction , and @xmath43 is determined from the qcd improved dipole picture ( eq . ( [ eq : pqcd ] ) ) .
the `` hat '' on @xmath43 is to distinguish the total cross section for the scattering of one component of the photon wave function from the total @xmath38 cross section which we consider in sect .
[ sec : gamma ] . applying the optical theorem in the large @xmath44 limit reproduces @xmath45 .
for now we assume that the amplitude is purely imaginary .
we will return to the question of a real part of the amplitude at the end of sect .
[ sec : impact ] . in this section
, we will make an estimate of the form of @xmath42 which will take into account the nonzero size of the @xmath17 dipole and which will smoothly interpolate between perturbative and nonperturbative regimes using the hadronic size , @xmath16 , as a parameter .
we act in the spirit of ref .
@xcite by modeling the @xmath1-dependence of the amplitude in the soft and hard regimes and by using @xmath16 to build a smooth interpolation .
the three steps : building a model for the small dipole region , building a model for the large wave - packet region , and interpolating between the two regions are outlined in the next three paragraphs .
we start by writing the general structure of the amplitude .
the @xmath1-dependence , @xmath46 , is written as the product of three functions @xmath47 here and in the rest of this section , the dependence of @xmath48 upon @xmath44 is replaced by dependence upon @xmath3 and @xmath16 . @xmath49 describes the @xmath1-dependence of the nucleon target , @xmath50 describes the @xmath1-dependence of the hadronic projectile , and @xmath51 accounts for gribov diffusion .
this method of separating the @xmath1-dependence into three factors corresponding to different sources of @xmath1-dependence is similar to what is used in ref .
@xcite .
the next task is to model the small dipole size @xmath1-dependence .
both the soft pomeron exchange factor and the hadronic form factor must approach unity as the size , @xmath16 , shrinks to zero .
the qcd factorization theorem implies that the @xmath1-dependence of the small dipole - nucleon amplitude is universal .
hence , it can be extracted directly from @xmath0 photo(electro)-production since the @xmath0 wave function is known to be a small size wave - packet @xcite .
data from @xmath0 production reveal that the two - gluon form factor is , @xmath52 the subscript , 1 , labels the two - gluon form factor and @xmath53 is a measurable parameter in the two - gluon form factor .
the value , @xmath54 gev@xmath10 is extracted from data in refs .
@xcite . for a detailed discussion of the two - gluon form factor in @xmath37 production , see ref . @xcite . in @xcite ,
it was discussed in detail how , due to the transverse squeezing of the @xmath0 wave function , the @xmath1-dependence of the amplitude comes solely from the two - gluon form factor . in particular , it was found that the dipole form factor contributes only about @xmath55 gev@xmath56 to the slope of the @xmath1-dependence .
the assumption that only the gluon form factor is relevant for @xmath0 production has been successfully tested against data in refs .
hence , in the limit of small dipole sizes , @xmath57 next we construct a model for the large wave packet behavior .
when the hadronic state has a large size , the @xmath1-dependence receives contributions from sources other than the two - gluon form factor .
we rewrite eq .
( [ eq : largesize ] ) in the form , @xmath58 now , @xmath59 is the electromagnetic form factor of the nucleon which is known phenomenologically to take the form , @xmath60 where @xmath61 gev@xmath10 .
large size hadronic configurations can be reasonably expected to have @xmath1-dependence similar to the pion electromagnetic form factor .
thus , for the hadronic form factor we use the well - known form of the pion form factor , @xmath62 with @xmath63 gev@xmath10 . here , @xmath64 is the characteristic size of the pion , and takes on a value of approximately @xmath65 fm .
this value for the pion size is consistent with what is used in the matching ansatz of ref .
@xcite and agrees well with data for the @xmath66n cross section in ref .
for low-@xmath3 soft scattering there is also a factor that arises from gribov diffusion effects : @xmath67 the factor , @xmath51 , describes the exchange of a soft pomeron with regge slope @xmath68 gev@xmath56 and @xmath69 .
the value of @xmath70 is determined by the boundary of the region where gribov diffusion effects become significant .
finally , we must find a reasonable way to interpolate between the hard and the soft regions .
we use the @xmath1-dependence discussed in the previous two paragraphs to guess the following form for the hadronic configuration - nucleon amplitude : @xmath71 in order to give the variation with @xmath16 geometric behavior , we use @xmath72 as a parameter . to interpolate between the nucleon and the two - gluon form factors , we have defined the function ,
@xmath73 note that when @xmath16 equals @xmath64 , @xmath74 is the product of eqs .
( [ eq : nucleon ] ) , ( [ eq : pion ] ) , and ( [ eq : diff ] ) . in the small @xmath16 limit , the dipole form factor and the pomeron form factor approach unity , @xmath75 , and the limit in eq .
( [ eq : smallsize ] ) is recovered .
varying @xmath72 interpolates smoothly between the hard and soft regions .
note that we neglect a possible small @xmath3 dependence of @xmath49 at @xmath76 .
( see the discussion in ref .
) however , our model is adjusted to reproduce the observed @xmath3-dependence of the slope for photoproduction of @xmath0 mesons .
having obtained eq . ( [ eq : ourmodel ] ) , the next step is to transform to the impact parameter representation where the profile function is defined by the relation , @xmath77 where @xmath78 . the subscript , @xmath79 , indicates that we are considering the profile function for the scattering of a single hadronic component of the photon wave function from the proton .
we get the profile function by inverting eq .
( [ eq : amplitude ] ) , @xmath80 for an imaginary amplitude , the bbl is reached when @xmath81 and the elastic and inelastic cross sections are equal .
recall that if a singlet dipole consists of color octet representations of su(3 ) , eq .
( [ eq : pqcd ] ) has an extra factor of @xmath82 , so the bbl for the interaction of a hadronic configuration with the nucleon is certainly reached for @xmath83 .
therefore , whenever @xmath84 , we conclude that the interaction takes place near the bbl . in the rest of this section
, we will suppress explicit reference to the argument , @xmath44 , in the profile function .
we have plotted the function @xmath85 for different values of the dipole size and @xmath3 in fig .
[ fig : dipoleg ] .
we have used gluon distributions from cteq5l in the perturbative calculation of @xmath43 @xcite . recall from the introduction that our model requires that we specify the external photon virtuality .
since we are interested in the possibility of reaching the bbl at a few gev@xmath5 , we have set @xmath86 gev@xmath10 in fig .
[ fig : dipoleg ] .
a gaussian ansatz is commonly used in experiments to extrapolate the @xmath1-dependence to large values .
let us , therefore , briefly compare the behavior of our model to that of a simple gaussian .
we start with the form @xmath87 which is then transformed into impact parameter space giving , @xmath88 the slope of the gaussian , @xmath89 , is chosen so that it yields the same standard deviation in @xmath85 as our model .
one danger in using a gaussian model is that it neglects the importance of interactions in peripheral regions .
our model attempts to fix this problem by spreading out the distribution in @xmath1 .
note in fig .
[ fig : compgauss ] that our model fall off more slowly with @xmath15 .
now let us estimate the contribution of large values of @xmath85 to the total hadronic cross section .
the total cross section follows from the optical theorem , @xmath90 we have made a numerical evaluation of the fraction of the total hadronic configuration - nucleon cross section obtained by setting different upper limits on the @xmath15-integral in eq .
( [ eq : cross ] ) . in fig .
[ fig : fraction ] , one can see that no more that about thirty percent of the total hadronic cross section is due to values of @xmath84 .
moreover , contributions from large values of @xmath85 occur for hadronic sizes close to the pion size , @xmath20 fm .
averaging over the photon wave function will lead to a suppression of contributions from larger size hadronic configurations , so there will indeed be a small contribution to the total dis cross section due to large values of @xmath85 ( see fig .
[ fig : dist ] ) .
the goal of sect .
[ sec : gamma ] will be to determine whether the contribution to the @xmath38 cross section from large values of @xmath85 is significant enough that we may expect to see black body behavior within hera kinematics . to summarize , fig .
[ fig : dipoleg ] demonstrates that large values of @xmath85 are approached for hadronic configuration - nucleon scattering at central impact parameters , @xmath91 fm . in fig .
[ fig : dipoleg ] it is seen that this is particularly true for hadronic sizes around @xmath92 fm .
figure [ fig : fraction ] shows that for @xmath93 fm , a maximum of about 1/3 of the total hadron - nucleon cross section comes from values of @xmath85 that approach the black limit
. when @xmath94 fm , a very small fraction of the total hadronic configuration - nucleon cross section comes from large values of @xmath85 .
the only contribution from @xmath84 to the total cross section for @xmath95 fm occurs at very small @xmath3 ( @xmath96 ) .
most of the model dependence in this calculation comes from uncertainty in the large-@xmath1 behavior of the amplitude .
the different curves in fig . [ fig : testmodel ] demonstrate how our model changes if we remove contributions from large @xmath1 .
notice that simply removing the contribution from @xmath97 gev@xmath5 leads to an error of less than ten percent .
thus , we do not expect our uncertainty in the large @xmath1 behavior to have a drastic effect .
we should also remark that we have considered only the non - spin - flip interactions .
corrections which account for the spin - flip amplitude would result in a smaller non - spin - flip amplitude than what we consider here .
experimental results in ref .
@xcite demonstrate that the polarization , @xmath98 , is less than @xmath99 for the range of @xmath1 we are discussing . from the formula relating @xmath98 to the spin - flip amplitude , @xmath100 we find the fraction , @xmath101 . here
, @xmath102 represents the amplitude with no spin - flip whereas @xmath103 represents the amplitude with spin - flip .
we now return to the issue of a real part of the amplitude which we ignored in sect .
[ sec : model ] . in the considered kinematic region
( @xmath104 gev@xmath5 ) , the ratio of the real to imaginary part of the amplitude , @xmath105 , is rather small .
indeed , if we adopt power law behavior for the total cross section , @xmath106 , we can estimate the value of @xmath107 using the following formula which follows from the gribov - migdal result @xcite at high energies in the near forward direction , @xmath108 the amplitude can be rewritten as , @xmath109 where the function , @xmath42 , is assumed to be strictly real . for soft regions ,
the total cross section has the approximate s - behavior of the @xmath110 cross section as in ref .
@xcite , consistent with the behavior of a donnachie - landshoff soft pomeron @xcite . in that case ,
@xmath111 , and eq . ( [ eq : disp ] ) gives @xmath112 .
the second term in eq .
( [ eq : imaginamp ] ) appears squared in the calculation of the cross section , so the correction to the cross section is approximately one percent . for the high @xmath4 , low-@xmath3 region , the total cross section experiences rapid growth and @xmath113 , or , by eq .
( [ eq : disp ] ) , @xmath114 .
the correction to the squared amplitude is therefore approximately ten percent near the forward direction .
away from the forward direction , one must account for the small variation of @xmath105 with @xmath1 .
the effect can be estimated by considering the signature factor in the general form of the reggeon amplitude .
for @xmath115 gev@xmath10 , @xmath116 continues to contribute a negligible amount to the amplitude .
the elastic cross section associated with eq .
( [ eq : imaginamp ] ) is found by integrating the profile function over impact parameters , @xmath117 the total cross section is found using eq .
( [ eq : cross ] ) .
therefore , the inelastic cross section is , @xmath118 with the unitarity constraint , @xmath119 if the amplitude is purely imaginary , then @xmath120 and the unitarity constraint is that @xmath121 .
note that by considering only the imaginary part of the amplitude , we have considered only the real part of the profile function . if the amplitude is given a real part correction , then the profile function will obtain an imaginary part , the elastic cross section will increase , and the inelastic cross section will decrease .
the correction to the unitarity constraint on @xmath122 is , from eq .
( [ eq : unitconst ] ) , @xmath123 . in the region of large @xmath4 ,
the effect of a real part in the amplitude would clearly be noticeable . by eq .
( [ eq : unitconst ] ) , the unitarity limit on the real part of the profile function for @xmath114 would be , @xmath124 thus , the unitarity limit on the real part of the profile function may be less than unity by as much as ten percent . at this point , we should remark that both the contribution from the real part of the amplitude and the contribution from inelastic diffraction will tend to _ raise _ the boundary in impact parameter space where the bbl is reached .
we neglect both effects in our model . as a consequence , when our model predicts that the bbl has been reached below a certain impact parameter , we can be confident that the same would be true in a model that incorporates inelastic diffraction and the effects of a real component of the amplitude .
on the other hand , if our model predicts that the bbl has not been reached , we must keep in mind that corrections due to inelastic diffraction and a real part of the amplitude may be important .
in other words , the bbl may already be approached at larger values of @xmath15 than what our model predicts .
the uncertainty in the matching region is expressed by the uncertainty in the parameter , @xmath29 .
however , values of the order of 4 to 10 seem to work well and , as shown in fig .
[ fig : lambda ] , there is a variation of only about fifteen percent at small impact parameters when we vary @xmath29 from 10 to 4 .
note that this is done for a hadronic size of @xmath125 fm which is in the region where the dependence upon @xmath29 should be at its greatest .
however , there remains another subtlety related to the matching of kinematic regions .
first , recall the distinction between the energy scale , @xmath4 , denoting the virtuality of the photon in a particular scattering process , and the scale , @xmath23 , which is the energy scale related to the hadronic size , @xmath16 , through the scaling ansatz of ref . @xcite .
these two scales are nearly equal as long as we consider hadronic sizes in the vicinity of the average hadronic size for @xmath31 . in determining which value of @xmath3 ( called @xmath27 in eq .
( [ eq : pqcd ] ) ) should be used to calculate the hadronic cross section , the authors of ref .
@xcite chose to relate the value of @xmath27 to the value @xmath3 for a particular @xmath38 process in such a way that @xmath27 varies as @xmath126 and so that for typical hadronic sizes , @xmath127 . within the color dipole picture ,
this accounts for the dependence of the the light - cone fraction , @xmath27 , of the gluon attached to the @xmath39 pair on the mass , @xmath128 , of the dipole : @xmath129 the result is that we can not speak unambiguously of the hadronic cross section without referring to the virtuality of the probe which generated a component of given transverse size . from ref .
@xcite we have , neglecting the constituent quark mass , @xmath130 here we see that if we consider a fixed @xmath4 , the universality of the hadronic cross section fails for small hadronic sizes ( large @xmath23 ) , but is recovered for larger hadronic sizes .
the value of @xmath27 used in a calculation of the hadronic cross section will be significantly larger than @xmath3 for small hadronic sizes , leading to a suppression of the cross section in the small size region .
in particular , in fig . [
fig : dipoleg ] , the approach to the bbl at small @xmath16 is slowed due to the large values of @xmath3 needed to push the small size configuration on shell . in investigating the hadronic profile function , it may also be reasonable to determine @xmath4 by letting it equal @xmath23 so that the value of @xmath16 always corresponds to a typical component of the virtual photon .
we have done this in fig .
[ fig : profile_9_23 ] and we can see that at small @xmath16 , @xmath85 is substantially larger , especially at small @xmath3 . comparing figs .
[ fig : dipoleg ] and [ fig : profile_9_23 ] , we see that at @xmath131 fm this effect is significant while at intermediate hadronic sizes the effect is very small . for @xmath132 fm there is no discernible difference between the two cases .
the physical meaning of this effect is that the profile function for a small size configuration approaches the bbl more slowly if it is far off shell for a given @xmath4 .
note that once we begin to calculate the total cross section , an external value for @xmath4 is explicit , and we no longer have this ambiguity .
furthermore , there is some uncertainty in the gluon distribution used to calculate @xmath133 .
this is demonstrated in figs .
[ fig : f2vsx ] and [ fig : flvsx ] where we compare results for the structure functions using cteq5l @xcite , cteq6l @xcite , and mrst98 @xcite leading order gluon parton distributions .
the dependence upon the parton distribution is seen to be small , but we used cteq5l parton distributions for all other calculations because they seem to yield optimal consistency with data . as we mentioned in sect .
[ sec : model ] , the value of @xmath64 that we used is consistent with the slope of the @xmath66n cross section as measured in ref .
@xcite and with the matching ansatz used in ref .
@xcite . in the model of the @xmath1-dependence
, @xmath64 determines where soft pomeron behavior becomes important , and one may well ask whether a different value of @xmath64 is appropriate . for models with a larger value of @xmath64 ,
the suppression of the profile function due to the pomeron form factor , @xmath134 , does not occur until one considers larger hadronic configurations .
therefore , for intermediate hadronic sizes , the profile function rapidly approaches the unitarity limit as @xmath3 decreases when @xmath64 is large .
this can be seen in fig .
[ fig : profilewrong ] where we have repeated the calculation of fig .
[ fig : dipoleg ] , this time using @xmath135 fm .
note the large values of @xmath85 at small @xmath15 for @xmath136 fm .
therefore , by choosing a smaller value for @xmath64 we are making a conservative estimate of the approach to the bbl .
to properly study the proximity of dis to the unitarity limit , we must evaluate the degree to which the different hadronic components contribute to the @xmath2 cross section , @xmath137 . @xmath138 and @xmath139 refer , respectively , to the transverse and longitudinal cross sections . in the color dipole formalism
, the longitudinal and transverse cross sections can be factorized into the convolution of the perturbative light - cone wavefunction with a universal color dipole cross section , @xmath140 in this paper , eq .
( [ eq : totalcross ] ) applies also to cases where @xmath141 is the cross section for interactions of large size hadronic configurations with the nucleon .
@xmath142 is the quark momentum fraction , and @xmath143 is the longitudinal / transverse photon wave function calculated in qed . we have calculated @xmath144 using @xmath145 with @xmath1-dependence determined in sect .
[ sec : model ] .
plots of @xmath31 and @xmath32 are shown in fig .
[ fig : f2vsx ] and fig .
[ fig : flvsx ] .
the structure functions @xmath31 and @xmath32 are defined as , @xmath146 note that the @xmath1-dependence is not needed for calculations of @xmath31 and @xmath32 . for more plots of the total cross section calculated within
the qcd improved dipole model see ref .
figure [ fig : dist ] shows the distribution of the integrand in eq .
( [ eq : totalcross ] ) over total hadronic sizes and demonstrates the suppression of large size hadronic configurations .
we would like to study the contribution of intervals of @xmath147 to the @xmath38 cross section .
however , the profile function for @xmath2 scattering by itself is not useful because the photon wave function is not normalizable and because it depends on @xmath148 .
thus , in order to look for the proximity to the bbl , we have plotted the fraction of the total @xmath38 cross section due to different regions of the _ hadronic _ profile function .
plots with different values of @xmath3 are shown in fig .
[ fig : totalfrac ] .
the @xmath149-axis denotes the fraction of the longitudinal ( transverse ) cross section with contributions from @xmath85 greater than the corresponding value on the @xmath3-axis .
note that when @xmath86 gev@xmath10 and @xmath6 , about 1/5 of the longitudinal cross sections are due to dipole configurations corresponding to @xmath150 . at @xmath151 gev@xmath10 , fig .
[ fig : totalfrac ] demonstrates the recovery of the leading twist behavior , especially for @xmath28 at @xmath152 , where less than one - tenth of @xmath28 is due to hadronic configurations with @xmath150 .
it is clear from fig .
[ fig : totalfrac ] that , for central impact parameters and low enough @xmath4 , a significant portion of the total cross section is due to hadronic configuration - nucleon interactions that are close to the unitarity limit . in high energy @xmath153a scattering , where the effects of the bbl are enhanced , we may be able to use dis to probe the bbl .
this possibility is discussed within the context of the qcd improved dipole model in sect .
[ sec : nuclear ] .
the reasonableness of our model is demonstrated in fig . [ fig : f2vsx ] where the dipole model is seen to be consistent with recent hera data for @xmath31 at low values of @xmath4 and @xmath3 .
furthermore , in fig . [
fig : flvsx ] , our model is seen to be consistent with preliminary results from hera for @xmath32 @xcite .
other studies of the impact parameter picture of hadronic interactions with nucleons were done in @xcite , where @xmath7 production data was used to extract the @xmath1-dependence .
the analysis in @xcite used the s - matrix convention , @xmath154 , in place of the profile function . in fig .
[ fig : mueller ] we have plotted our prediction of the s - matrix profile for central impact parameters along with earlier result from munier _ et al . _ @xcite . in their analysis , the authors were restricted to using @xmath7 production data to model the @xmath1-dependence .
data for @xmath7 production is limited to kinematics where @xmath155 gev@xmath10 ( @xmath156 fm ) so the accuracy of their results is limited to moderate impact parameters .
further complication is introduced into their analysis by the need to model the @xmath7-meson wavefunction .
our model uses @xmath0 production data and is therefore valid at small values of @xmath15 . as mentioned in sect .
[ sec : model ] , production of @xmath0 depends only on the two - gluon form factor , and can therefore be extended down to very small impact parameters .
it is natural to compare our results for @xmath157 with the median value of the dipole size as it is was used in the calculation of the amplitude estimated in ref .
@xcite . in our case ,
the value of @xmath16 that corresponds to @xmath157 evaluated at @xmath158 gev@xmath10 is about @xmath159 fm @xcite . with this assumption ,
figure [ fig : mueller ] shows that our model has very good agreement with the results of ref .
@xcite at moderate values of @xmath15 while there is an expected deviation between the two models for low values of @xmath15 ( see footnote @xcite ) .
since our model will have @xmath160 behavior at small values of @xmath15 , then even in the small @xmath15 region , our model deviates from the results of ref .
@xcite by no more than about twenty - five percent .
reference @xcite used a simple exponential or power ansatz to interpolate to larger values of @xmath1 as indicated in the figure . in fig .
[ fig : mueller ] , it is seen that the model used in @xcite has a high degree of uncertainty at small values of @xmath15 because of the necessity to guess the form of the function that interpolates to large @xmath1 .
in contrast , our model uses information about @xmath37 production and dis to model the small @xmath15 behavior .
after preliminary results of our study were presented , there appeared an experimental analysis with improved data on inclusive cross sections and vector meson production at hera by henri kowalski and derek teaney @xcite .
they carry out an analysis similar to that used in @xcite .
therefore , it differs from our analysis in that it does not include information about large @xmath1 behavior of the two - gluon form factor .
future improvements on the dipole picture should make comparisons with this data .
it is interesting to examine how the profile function @xmath161 changes when the free proton target is substituted by a heavy nuclear target such as the nucleus of @xmath8pb . in the heavy nucleus case
, the procedure for obtaining @xmath161 differs from the one in the nucleon case and is outlined below .
first , for dipoles of small transverse sizes , @xmath162 fm , the inelastic scattering cross section at a given impact parameter @xmath15 is given by the perturbative qcd expression involving the impact parameter dependent nuclear gluon distribution , @xmath163 ( compare to eq .
( [ eq : pqcd ] ) ) @xmath164 where @xmath165 is given by eq .
( [ eq : xprime ] ) . the gluon distribution @xmath163 , normalized such that @xmath166 , was evaluated in @xcite using the theory of leading twist nuclear shadowing .
the profile function @xmath161 can be found from eq .
( [ eq : inelcross ] ) ( see see also @xcite ) , @xmath167 ignoring the small imaginary part of @xmath161 , which is even smaller in the heavy nucleus case than in the free proton case because of the effect of nuclear shadowing , eq .
( [ eq : pb2 ] ) gives @xmath168 which is valid for @xmath162 fm .
second , for dipoles of a larger size , @xmath169 fm , the profile function is found by interpolating between the pqcd expression of eq .
( [ eq : pb3 ] ) and the profile function calculated at @xmath170 @xcite @xmath171 the profile function @xmath172 is calculated using the glauber multiple scattering formalism @xcite @xmath173 where @xmath174 is the energy - dependent pion - nucleon total scattering cross section , @xmath175 mb , @xmath176 gev ; @xmath177 is the nuclear optical density normalized such that @xmath178 .
@xmath179 is the number of nucleons in the target .
third , for the dipoles with the size @xmath180 , the profile function is given by eq .
( [ eq : pbx ] ) , where the pion - nucleon cross section is allowed to slowly grow as @xmath181 the results for the profile function @xmath161 for the nucleus of @xmath8pb are presented in fig .
[ fig : pb ] by the two solid curves ( @xmath182 and @xmath183 ) , dashed ( @xmath184 ) and dot - dashed ( @xmath185 ) curves .
the profile function for the nuclear target shows some similarity with the profile function for the proton target .
the main differences are that the bbl is approached over a larger range of impact parameters than in the case of a proton target .
this is not surprising because of the larger thickness of the nuclear target .
the plots in fig .
[ fig : nuclearcase ] show the fraction of the hadronic cross section due to large values of @xmath186 .
large leading twist gluon shadowing tames the growth of the interaction of hadronic components of the photon with the nucleus so that the unitarity constraint is satisfied for @xmath187 while the bbl may be reached for a large range of impact parameters .
for smaller @xmath3 , unitarity starts to break down at central impact parameters . for large @xmath16 ,
unitarity is automatically satisfied since the glauber model for large total cross sections leads to a @xmath186 that approaches unity .
finally , we have included plots in fig .
[ fig : nuclearcase2 ] for the nuclear target showing the fraction of @xmath188 due to large values of @xmath186 analogous to those in fig .
[ fig : totalfrac ] . in fig .
[ fig : nuclearcase2 ] we see that the bbl is approached for nearly all values of @xmath3 at @xmath189 gev@xmath10 and @xmath190 gev@xmath10 .
( note the recovery of leading twist behavior for the longitudinal cross section at large @xmath3 and @xmath190 gev@xmath10 . )
notice also that the fraction of @xmath191 due to large values of @xmath186 for @xmath192 is actually larger in some cases than for the case , @xmath193 .
this effect can be explained qualitatively by inspection of fig .
[ fig : pb ] . for
the case of the nuclear target , the main contributions to @xmath191 come from smaller values of @xmath16 ( @xmath194 fm ) .
the growth of the profile function with decreasing @xmath3 at small @xmath16 is slower for smaller values of @xmath3 ( @xmath195 ) than for larger values .
thus , the tail of profile function at large impact parameter may become significant in these regions .
a general , well - known feature of @xmath38 and @xmath196 scattering is that the fraction of the interactions of hadronic components in the virtual photon wavefunction with the proton which take place at or near the bbl increases as @xmath3 and @xmath4 decrease as is exhibited explicitly in this paper in figs . [
fig : dipoleg ] , [ fig : fraction ] and [ fig : totalfrac ] .
we hope that one day we may be able to exploit the novel properties of interactions in the bbl to study a new phase of pqcd . in ref .
@xcite the signatures of the bbl for dis were discussed with the hopes that they would be seen in future experiments .
it remained to be determined , however , in which kinematical regions one can expect to see black body behavior .
having now constructed a model of the amplitude for the interaction of the hadronic components of the virtual photon wavefunction , we are in a position to make rough estimates of the fraction of the hadronic interactions that exhibit the characteristic behavior of black body interactions .
more precisely , since we know that the effects that we have ignored so far inelastic diffraction and the real part of @xmath74 , will tend to increase the proximity of the interactions to the unitarity limit , then we can place lower limits on the values of @xmath3 and @xmath15 where a significant fraction of the events will occur at or near the bbl .
our results show that , within available hera kinematics , a significant fraction of the total dis cross section is due to interactions of the hadronic components with the proton that occur near the bbl . in particular , fig .
[ fig : totalfrac ] shows that at @xmath197 gev@xmath10 and @xmath96 , about 1/5 of the longitudinal cross section is due to values of @xmath198 .
the agreement of our model with preliminary hera data and with previous models helps to strengthen this conclusion .
an improved model , with corrections for inelastic diffraction , will likely predict a more rapid approach to the bbl at small @xmath3 and central impact paramters .
the approach to the bbl as @xmath3 and @xmath15 decrease occurs much more rapidly for the case of a heavy nuclear target than for the case of a proton target .
this can be seen by comparing figs .
[ fig : fraction ] and [ fig : nuclearcase ] . for example , at @xmath199 fm and @xmath200 , fig .
[ fig : fraction ] shows that , for the proton , less than 1/2 of the total cross section is due to contributions from @xmath201 whereas with a @xmath8pb target , fig .
[ fig : nuclearcase ] shows that over seventy percent of the total cross section is due to contributions from @xmath201 .
this suggests that nuclear targets are ideal for studies of the bbl as a phase of qcd as has been discussed before in ref .
future work on this subject should incorporate inelastic effects .
also , a greater understanding of the large @xmath1 behavior would lead to greater accuracy in the model .
m. mcdermott , l. frankfurt , v. guzey and m. strikman , eur .
j. c * 16 * , 641 ( 2000 ) [ arxiv : hep - ph/9912547 ] .
k. golec - biernat and m. wusthoff , phys .
d * 60 * , 114023 ( 1999 ) [ arxiv : hep - ph/9903358 ] .
h. i. miettinen and j. pumplin , phys .
* 42 * , 204 ( 1979 ) .
b. blaettel , g. baym , l. l. frankfurt , h. heiselberg and m. strikman , phys .
d * 47 * , 2761 ( 1993 ) .
l. frankfurt , v. guzey , m. mcdermott and m. strikman , jhep * 0202 * , 027 ( 2002 ) [ arxiv : hep - ph/0201230 ] . b. povh and j. hufner , phys .
lett . * 58 * , 1612 ( 1987 ) .
l. frankfurt , a. radyushkin and m. strikman , phys .
d * 55 * , 98 ( 1997 ) [ arxiv : hep - ph/9610274 ] .
a. levy , phys .
b * 424 * , 191 ( 1998 ) [ arxiv : hep - ph/9712519 ] . c. adloff _ et al .
_ [ h1 collaboration ] , phys .
b * 483 * , 23 ( 2000 ) [ arxiv : hep - ex/0003020 ] .
s. chekanov _ et al . _
[ zeus collaboration ] , eur .
j. c * 24 * , 345 ( 2002 ) [ arxiv : hep - ex/0201043 ] .
l. frankfurt and m. strikman , phys .
d * 66 * , 031502 ( 2002 ) [ arxiv : hep - ph/0205223 ] .
m. binkley _ et al .
_ , phys .
lett . * 48 * , 73 ( 1982 ) .
u. camerini _ et al .
_ , phys .
lett . * 35 * , 483 ( 1975 ) .
b. gittelman , k. m. hanson , d. larson , e. loh , a. silverman and g. theodosiou , phys .
* 35 * , 1616 ( 1975 ) .
j. p. burq _ et al .
_ , phys .
b * 109 * , 111 ( 1982 ) .
m. strikman and c. weiss , arxiv : hep - ph/0308191 . h. l. lai _ et al . _
[ cteq collaboration ] , eur .
j. c * 12 * , 375 ( 2000 ) [ arxiv : hep - ph/9903282 ] .
i. p. auer _ et al .
_ , phys .
lett . * 39 * , 313 ( 1977 ) . v. n. gribov and a. a. migdal , sov .
j. nucl .
* 8 * , 583 ( 1969 ) [ yad .
* 8 * , 1002 ( 1968 ) ] .
a. donnachie and p. v. landshoff , phys .
b * 191 * , 309 ( 1987 ) [ erratum - ibid .
b * 198 * , 590 ( 1987 ) ] .
b * 296 * , 227 ( 1992 ) [ arxiv : hep - ph/9209205 ] .
j. pumplin , d. r. stump , j. huston , h. l. lai , p. nadolsky and w. k. tung , jhep * 0207 * , 012 ( 2002 ) [ arxiv : hep - ph/0201195 ] .
a. d. martin , r. g. roberts , w. j. stirling and r. s. thorne , eur .
j. c * 4 * , 463 ( 1998 ) [ arxiv : hep - ph/9803445 ] . h1prelim-03 - 043 e. lobodzinska f. p.
schilling [ h1 collaboration ] , arxiv : hep - ex/0210027 .
h. kowalski and d. teaney , arxiv : hep - ph/0304189 .
j. pumplin , phys .
scripta * 25 * , 191 ( 1982 ) .
r. j. glauber , phys .
* 100 * , 242 ( 1955 ) ; v. franco , phys .
rev . lett .
* 24 * , 1452 ( 1970 ) .
l. frankfurt , v. guzey , m. mcdermott and m. strikman , phys .
lett . * 87 * , 192301 ( 2001 ) [ arxiv : hep - ph/0104154 ] .
|
we use information about dis and @xmath0 production on hydrogen to model the @xmath1-dependence of the @xmath2 scattering amplitude .
we investigate the profile function for elastic scattering of hadronic components of the virtual photon off both a nucleon and heavy nuclear target , and we estimate the value of the impact parameter where the black body limit is reached .
we also estimate the fraction of the cross section that is due to hadronic configurations in the virtual photon wave function that approach the unitarity limit .
we extract , from these considerations , approximate lower limits on the values of @xmath3 where the leading twist approximation in dis is violated .
we observe that the black body limit may be approached within hera kinematics with @xmath4 equal to a few gev@xmath5 and @xmath6 .
comparisons are made with earlier predictions by munier _
et al . _ , and the longitudinal structure function
is compared with preliminary hera data .
the principle advantage of our method is that we do not rely solely on the @xmath1-dependence of @xmath7-meson production data .
this allows us to extend our analysis down to very small impact parameters and dipole sizes .
finally , we perform a similar calculation with a @xmath8pb target , and we demonstrate that the black body limit is already approached at @xmath9 gev@xmath10 and @xmath6 .
|
in a large part of the density - temperature plane , integral equation theories are a reliable tool for studying thermodynamic and structural properties of , among others , simple one - component fluids @xcite ; unfortunately , in the vicinity of a liquid - vapor critical point , integral equations are haunted by a host of difficulties , leading to a variety of shortcomings such as incorrect and non - matching branches of the binodal , classical values at best for the critical exponents , or other deviations from the correct behavior at the critical singularity @xcite .
asymptotically close to the critical point , on the other hand , renormalization group ( ) theory is the instrument of choice for describing the fluid ; in general , however , approaches do not allow one to derive non - universal quantities from microscopic information only , _
i. e. _ from knowledge of the forces acting between the fluid s particles alone .
one of the theories devised to bridge the conceptual gap between these complementary approaches is the _ hierarchical reference theory _ ( ) first put forward by parola and reatto @xcite : in this theory the introduction of a cut - off wavenumber @xmath2 inspired by momentum space theory and , for every value of @xmath2 , of a renormalized potential @xmath3 means that only non - critical systems have to be considered at any stage of the calculation ; consequently , integral equations may successfully be applied to every system with @xmath4 , and critical behavior characterized by non - classical critical exponents is recovered only in the limit @xmath5 .
while applicability of to a number of interesting systems , ranging from a lattice gas or ising model @xcite to various one - component fluids @xcite even including three - body interactions @xcite , internal degrees of freedom @xcite , or non - hard - core reference systems @xcite , was demonstrated early on , the main focus of research on has since shifted to the richer phase behavior of binary systems @xcite .
nevertheless , in the light of s high promise and low penetration into the liquid physics community , further study and critical assessment of this theory seem worthwhile , even and foremost in the case of simple one - component fluids : indeed , it is in this comparatively simple setting that we may gain important insights into the numerical side of the theory , and barring special mechanisms relevant to some specific model system only , any problems uncovered here must be expected to haunt more advanced applications of , too . in our work
we have found it convenient to restrict ourselves even further , implementing in its usual formulation @xcite for purely pairwise additive interactions _ via _ a potential @xmath6 obtained from the superposition of an infinitely repulsive hard sphere serving as reference system , @xmath7 , and a predominantly attractive tail @xmath8 , @xmath9 .
here we have made use of the notation introduced previously @xcite : superscripts always denote the system a quantitiy refers to ( here , `` ref '' and `` hs '' for the reference system and hard spheres , respectively ; similarly , `` @xmath10 '' for the system with cut - off @xmath2 ) , and a tilde indicates fourier transformation . in the present contribution
we apply our recent re - implementation of the theory @xcite to one of the simplest potentials exhibiting phase separation , _ viz . _
the square well potential @xmath11}$ ] ( cf .
sub - section iii b of ref . ):
@xmath12}(r ) = v^{{\rm hs}[\sigma]}(r)+w^{{\rm sw}[-\epsilon,\lambda,\sigma]}(r ) \\ v^{{\rm hs}[\sigma]}(r ) = \left\{\begin{array}{ccc } + \infty & : & r<\sigma \\ 0 & : & r>\sigma \end{array}\right . \\
w^{{\rm sw}[-\epsilon,\lambda,\sigma]}(r ) = \left\{\begin{array}{ccc } -\epsilon & : & r<\lambda\,\sigma \\ 0 & : & r>\lambda\,\sigma\ , .
\end{array}\right . \end{array}\ ] ] considering density - independent potentials only and chosing the hard core diameter @xmath13 and the well s depth @xmath14 as units of length and energy , respectively , the attractive well s range @xmath0 is the sole remaining parameter ; in this report we will study values of @xmath0 from slightly above unity up to 3.6 . with just one parameter , _ viz . _
@xmath0 , to vary , square wells obviously make for a convenient test case of and , indeed , of liquid state theories in general ; consequently , a great many simulational and theoretical efforts have been directed at this system , and studies of its phase behavior abound @xcite .
but square wells are also of interest in their own right , serving as albeit somewhat crude models of a wide variety of physical systems including , _
e. g. _ , @xmath15 , @xmath16 , @xmath17 , @xmath18 , @xmath19 , @xmath20 , @xmath21 , @xmath22-pentane and @xmath22-butane @xcite while current interest in this potential derives mainly from the finding that square wells capture the essential features of the interactions found in colloidal systems @xcite .
yet another motivation for this first application of to square wells comes from a recent , very accurate simulation study @xcite confirming and quantifying the presence in the system with @xmath23 of the yang - yang ( yy ) anomaly expected and experimentally found for asymmetric fluids @xcite . due to the extensive amount of data available in the literature the more recent of which
will shortly be presented later on , and in view of some of the limitations of in its current formulation we can not expect to gain new insight into the system at hand with a level of precision comparable to that of the more sophisticated simulation schemes .
instead , in the present contribution our focus of interest lies on some aspects of s numerical side , specifically on those that are sensitive to the potential s range : indeed , as stated already in ref .
, for a potential as pronouncedly short - ranged as square wells some of the numerical problems should show up much more prominently than in other systems like , _ e. g. _ , the hard - core yukawa fluid previously considered @xcite where they are , of course , in principle still present but do not manifest themselves as clearly . in accordance with the preceding remarks , another reason why application of to square wells might be worthwhile lies in the closure underlying seemingly all applications so far of to simple one - component fluids with hard sphere reference part : as the usual formulation of in these cases relies on an _ ansatz _ for the two - particle direct correlation function @xmath24 very much in the spirit of stell s _ lowest - order @xmath25-ordered approximation _ @xcite ( ) or the equivalent _ optimized random - phase approximation
_ @xcite ( ) by andersen and chandler , the direct correlation function can never extend to larger @xmath26 values than the potential @xmath6 itself .
in particular , for the square well potential @xmath11}(r)$ ] we necessarily have @xmath27 for @xmath28 so that all moments of @xmath24 , _ i. e. _
@xmath29 , @xmath30 , exist , which is obviously at variance with the correct behavior near the critical point @xcite ; furthermore , at intermediate @xmath2 the direct correlation function can hardly be considered satisfactory , especially @xcite close to @xmath31 .
while some earlier publications @xcite already blamed unsatisfactory aspects of results on this inadequacy of the closure , square wells should bring out related problems of with the usual /-style closure much more clearly , and the numerical procedure s response should provide us with a signature to be looked out for in other systems , too ; also , even within the /-style approximation the implementation of the core condition _
via_approximate ordinary differential equations ( s ) for the relevant expansion coefficients easily shown to be inadequate for very short - ranged potentials @xcite casts some doubt on the range of @xmath0 values amenable to an treatment in the current formulation of the theory .
determination of the admissible @xmath0-range , on the other hand , is particularly interesting in the light of refs .
and as well as in view of the global renormalization scheme @xcite originally developed by white and co - workers as an extension of wilson s phase - space cell method @xcite to the liquid state ; it is only by combining tests internal to the theory and comparison with data available by other means that we are able to answer this question . in this contribution , after a sketchy presentation of the underlying theory itself ( section [ sec : theory ] ) and the implementation used ( section [ sec : program ] ) , in section [ sec : application ] we turn to the results of applying to square well systems of variable range . after a short summary of the critical point s location as obtained from simulation - based and other purely theoretical studies of square wells for various values of @xmath0 ( sub - section [ sec : non - hrt ] ) we first look into the core condition s implementation , which provides us with a first hint regarding the range of @xmath0 values accessible to in its current formulation and once more highlights the decoupling assumption s _ rle _ ( sub - section [ sec : cc ] ) .
the latter is also implicated in the correct choice of the boundary condition imposed at high density as discussed , alongside the boundary condition s location s effect , in sub - section [ sec : boundary ] .
a particularly grave aspect of s numerical side is the stiffness of the partial differential equation ( ) for close - to - critical and sub - critical temperatures ( sub - section [ sec : stiff ] ) , the vestiges of which are evident in the results obtained for quasi - continually varying @xmath0 as presented in sub - section [ sec : interference ] .
a short summary of our findings and conclusions ends our contribution ( section [ sec : bye ] ) .
the definite resource on is the review @xcite by the theory s original authors , summarizing its formalism as developed in a series of earlier publications @xcite so that we here present only an overview of the theory used and of its formulation , recapitulating some of our earlier findings @xcite ; the notation employed here of course co - incides with that of our preceding contribution @xcite . as mentioned already in section [ sec : intro ] , s mainstay is the implementation of the suppression of long - wavelength fluctuations characteristic of methods by means of a renormalized potential @xmath32 .
thus , rather than directly going from a reference fluid the properties of which are assumed known to the fully interacting system , _
i. e. _ from pair potential @xmath33 to @xmath34 , proceeds _ via _ a succession of rather artificial @xcite intermediate potentials @xmath3 : for every value of the cut - off wave number @xmath2 , @xmath32 is given by @xmath35}(k ) = -4\pi\,\epsilon\ , { \sin\lambda\,\sigma\,k-\lambda\,\sigma\,k\,\cos\lambda\,\sigma\,k \over k^3}\ , , \end{array}\ ] ] where the last line specializes to the square well potential of eq .
( [ eq : def : potsw ] ) ; obviously , @xmath36 and @xmath37 are recovered in the limits @xmath38 and @xmath5 , @xmath39 allowing to gradually turn on fluctuations of ever increasing wavelength by lowering @xmath2 from @xmath40 to zero ( numerically @xcite , from @xmath41 to @xmath42 ) ; as mentioned before , criticality ( together with non - classical critical exponents ) and phase separation ( with isotherms rigorously flat in the two phase region ) are obtained in the limit @xmath5 . in this procedure
it is essential to maintain the differential picture implied by theory and to make sure that the transition from @xmath2 to infinitesimally smaller cut - off @xmath43 is continuous even in the limit @xmath5 .
the latter requirement necessitates replacing the usual free energy @xmath44 and two - particle direct correlation function @xmath45 of the hypothetical system with cut - off @xmath2 and potential @xmath3 , the `` @xmath2-system '' , by suitably modified quantities , _ viz .
_ @xmath46 where @xmath47 is the number density of the system at hand ; the higher order correlation functions @xmath48 , @xmath49 , are free from such problems .
( note that all the direct correlation functions including @xmath50 are taken to include the ideal gas terms @xcite . ) with this set of quantities continuous even in the limit @xmath5 , _ viz .
_ @xmath51 , @xmath52 , and the @xmath53 , @xmath49 , is derived as a non - terminating hierarchy of coupled s at fixed density @xmath47 , calculating the properties of the @xmath2-system by treating the system at infinitesimally higher cut - off @xmath54 as a reference system ; of these equations , usually only the evolution equation for @xmath51 , _ viz .
_ @xmath55 as well as the important compressibility sum - rule @xmath56 valid for any cut - off @xmath2 directly enter practical calculations . when combined with a closure on the two - particle level , eqs .
( [ eq : dq : a ] ) and ( [ eq : consistency ] ) define a in the @xmath57-plane ; it is this that we will concern ourselves with in the remainder of this text .
said closure , reminiscent of / but adding one free parameter to allow imposing thermodynamic consistency as embodied in eq .
( [ eq : consistency ] ) , is given , just as in our earlier contribution @xcite , by @xmath58 where we have generalized to density - dependent potentials .
basis function @xmath59 is chosen to coincide with @xcite @xmath60 , and the higher basis functions @xmath61 , @xmath62 , vanish outside the core ; for our specific choice of basis functions see appendix b of ref . .
in order to ensure that both the core condition , _ i. e. _ @xmath63 for @xmath64 where @xmath65 is the pair distribution function , and sum - rule ( [ eq : consistency ] ) are met it is necessary to choose the correct set of expansion coefficients @xmath66 , @xmath30 , at every cut - off @xmath2 and for every density @xmath47 ; assuming their validity for @xmath67 and adopting the short hand notations @xmath68 = \int_{{\bbb r}^3 } { { { \rm d}}^3k\over(2\pi)^3}\,{\psi(k,\rho)\over \left(\tilde c\q_2(k,\rho)\right)^2 } \end{array}\ ] ] ( here , @xmath69 is an arbitrary function of @xmath70 and @xmath47 ) , both relations can be combined to @xmath71 \,{\partial\g\q_n(\rho)\over\partial q } \\\qquad = \alpha\q(\rho)\ , \hat{\cal i}\q\left[\tilde u_j(k,\rho)\,\tilde u_0(k,\rho ) , \rho\right ] \\\qquad\quad + { q^2\over2\pi^2 } \,{\tilde\phi(q,\rho)\,\tilde u_j(q,\rho ) \over\tilde\c\q(q,\rho ) \,\left(\tilde\c\q(q,\rho)-\tilde\phi(q,\rho)\right)}\ , , \quad j\ge1\ , ; \end{array}\\\ ] ] this set of equations must , of course , be truncated to a finite number @xmath72 of basis functions , and it is also necessary to neglect non - local contributions to @xmath73 / \partial q$ ] to allow convenient evaluation at arbitrary @xmath2 .
both of these approximations have been discussed at length in our previous contribution @xcite , and while the value of @xmath74 was found to strongly influence the quality of the results obtained , determination of the @xmath75 from eq .
( [ eq : cc : cint ] ) and said approximation for the slowly converging @xmath76-integrals @xmath2-dependence leads to systematic deficiencies at small @xmath26 in @xmath77 as determined from the ornstein - zernike relation .
unfortunately , for numerical reasons @xcite it is necessary to also adopt the so - called decoupling assumption @xcite , _ viz .
_ @xmath78 ; as can easily be seen , this is not only mathematically incompatible with thermodynamic consistency but even suffices to decouple the implied by eqs .
( [ eq : dq : a ] ) and ( [ eq : consistency ] ) to a set of unrelated s at fixed density only lacking thermodynamic consistency and thus unable to predict clear phase boundaries @xcite ; furthermore , we can not rule out that decoupling may have a significant influence on the solution generated @xcite , which is particularly troublesome as the much longer range of @xmath79 when compared with the other basis functions was originally invoked as justification for setting @xmath78 : for square wells , this assumption is certainly even less justified than for the rather long - ranged hard - core yukawa system ( @xmath80 ) considered in ref . .
returning to the , for the numerical implementation s benefit we , too , adopted a re - formulation in terms of an auxiliary function @xmath81 simply related to the modified free energy s derivative with respect to @xmath2 .
the details of the procedure leading to a of the form @xmath82 + d_{01 } [ f , q,\rho]\,{\partial\over\partial\rho}f(q,\rho ) + d_{02 } [ f , q,\rho]\,{\partial^2\over\partial\rho^2}f(q,\rho)\ , , \\
f(q,\rho)\,\tilde u_0 ^ 2(q,\rho ) = \ln\left(1-{\tilde\phi(q,\rho)\over\tilde\c\q(q,\rho)}\right ) + { \tilde\phi(q,\rho)\over\tilde\k\q(q,\rho)}\ , , \end{array}\ ] ] and the coefficient functions @xmath83 themselves can be found in appendix a of ref . , _ q. v._ref . . the above formulation ( [ eq : pde : f ] ) of the problem , of course still coupled to the s implementing the core condition , obviously has to be amended by initial and boundary conditions ; while the former easily follow from @xmath84 , @xmath30 , ( which is sufficient to also determine @xmath85 ) , choice of appropriate boundary conditions is slightly more complicated : if , as is the case in most of the calculations reported here ( exceptions see below ) , the low density boundary is located at @xmath86 , we can make use of the divergence of the ideal gas term @xmath87 in @xmath88 to derive not only @xmath89 but also @xmath90 which alone is , in principle , sufficient to uniquely determine the solution up to arbitrarily high density ; for computational reasons , however , it is preferable to instead only impose vanishing @xmath91 at @xmath92 and to supply an approximate condition for calculating @xmath91 at @xmath93 . among the candidates for the constraint to be imposed upon the solution at @xmath93 in addition to the core condition
there are two we should mention here : starting with ref . , the so - called -condition , _ viz .
_ @xmath94 , has been used extensively ; it should , however , be noted that this condition is incompatible with both thermodynamic consistency and the decoupling assumption @xcite .
an alternative first considered in our previous report @xcite is the decoupling assumption @xmath95 itself ; of course , this condition is still incompatible with the compressibility sum - rule ( [ eq : consistency ] ) but this is less of a problem at a boundary where partial derivatives with respect to @xmath47 can not be evaluated anyway .
another option ( not pursued in this contribution ) is to give up the core condition altogether , retaining only the lowest basis function @xmath96 in the closure ( [ eq : closure ] ) and thus effectively replacing eq .
( [ eq : cc : cint ] ) by eq . ( [ eq : consistency ] ) ; this has the added advantage of mathematical consistency while still retaining the structure of a so important for obtaining clear phase boundaries @xcite , _ v. s. _ it is one of s main achievements to allow calculating a fluid s binodal ( coinciding with the spinodal in three dimensions @xcite ) without resorting to maxwell constructions @xcite , for subcritical temperatures yielding isotherms rigorously flat in density intervals the boundaries of which are readily identified with the coexisting densities @xmath97 and @xmath98 .
thus , as the isothermal compressibility @xmath99 of the fully interacting system , readily found to be proportional to @xmath100 ( cf .
appendix a of ref . ) , diverges in the two - phase region , so must the auxiliary function @xmath81 in the limit @xmath5 . as a direct consequence of this ,
the transition from the modified free energy @xmath101 to @xmath81 is not only computationally convenient but also allows us to follow the isothermal compressibility s build - up much more easily ; even more importantly , a simple analysis @xcite of the behavior of the s coefficients for large @xmath81 readily characterizes the as stiff : for any density @xmath102 $ ] and close to @xmath103 , the true solution @xmath81 oscillates rapidly on a @xmath2-scale of the order of @xmath104 , with both an upper bound on the oscillations amplitudes and @xmath91 s average slope growing like @xmath105 needless to say that this behavior can not be reproduced numerically ( _ v .
i._sub - section [ sec : stiff ] ; _
q. v. _ ref . ) .
note , however , that it is not an artifact of the re - writing of the in the form ( [ eq : pde : f ] ) but rather a problem inherent to itself in a formulation based upon eq .
( [ eq : closure ] ) @xcite .
the numerical study of for square well systems of varying range parameter @xmath0 in section [ sec : application ] has only become feasible due to our recent re - implementation@xcite of this theory , discussed at length in refs . and
; we will make use of results obtained with this program exclusively . from a practical point of view
, our software provides a means of solving a finite - difference approximation to the ( [ eq : pde : f ] ) in an iterated full - approximation scheme , imposing boundary conditions at densities @xmath92 and @xmath93 as well as initial conditions at @xmath106 , generating a solution for @xmath2 as low as @xmath42 while ensuring numerical soundness of every step by employing a number of criteria .
the pivotal parameter governing all of the numerics is a small quantity denoted @xmath107 characteristic of the maximum admissible relative error introduced in a single step in the @xmath108 direction ; due to the paramount importance of derivatives with respect to @xmath47 , @xmath107 is strictly related to the coarseness of the density grid .
the only exception to the general strategy of ensuring a numerical quality of @xmath107 at every step in the calculation is the choice of step sizes @xmath109 in the @xmath108 direction , at least for sub - critical and close - to - critical temperatures : indeed , in that part of the @xmath57-plane where the divergence of the isothermal compressibility builds up , the s stiffness ( _ v .
s. _ ) renders fixed - precision arithmetic and relative errors bounded by @xmath107 incompatible .
consequently , for the calculations reported below we resort to step sizes @xmath109 pre - determined in a way analogous to that employed in earlier applications @xcite ; still , monitoring and assessing suitable components of the solution vector in terms of @xmath107 as described in sub - section iii e of ref . may yield a wealth of information on the numerical process and its evolution .
most of the calculations reported here have been performed on an equispaced density grid of @xmath110 density intervals spanning the range from @xmath86 to @xmath111 , corresponding to a value of @xmath112 ; @xmath74 was usually set to 7 ; and the pre - determined step sizes started from @xmath113 at @xmath114 , plunging to a mere @xmath115 when approaching @xmath116 . when locating the binodal _
via_the divergence of the isothermal compressibility @xmath99 we did not require an actual overflow to occur but instead looked for a @xmath99-ratio at neighbouring densities exceeding @xmath117 , which is a rather reliable indicator for the binodal s location as @xmath99 typically jumps by far less than two or by at least some twenty orders of magnitude within one @xmath118 ; the reported values for @xmath97 and @xmath98 are the mid - points of the density intervals so found . in principle
this allows us to locate the coexisting densities and the critical temperature and density to arbitrary precision , even though the computational cost rises sharply with falling @xmath107 .
as mentioned before , much of the motivation for applying in the formulation outlined in section [ sec : theory ] to the simple square well model potential is based upon various observations indicating possible limitations of this approximate formulation of for short - ranged potentials . a case in point
is the recent work of caccamo _
et al._@xcite entirely devoted to several thermodynamically consistent theories ability to deal with narrow hard - core yukawa systems ; sure enough , in the case of the shortcomings of the /-style closure ( [ eq : closure ] ) and , presumably , of the accompanying decoupling assumption underlying the core condition s implementation _ via_eq .
( [ eq : cc : cint ] ) were manifest already in refs . and and
recently confirmed by us @xcite , _ q. v._ref . .
of course , any of the problems discussed below only relate to when implemented along the lines of sections [ sec : theory ] and [ sec : program ] and not to proper ; however , for reasons discussed in ref .
, alternative formulations almost certainly render the numerics far more demanding and open up a whole new suite of problems regarding the numerical implementation s soundness , especially when performing fourier transformations of cut - off affected functions @xcite . in the following sub - sections
we will complement the discussion of ref . by further investigation into the numerical nature of ; before that , however , it seems pertinent to re - iterate some of the points raised in that publication as far as they concern the reasoning to be put forward in the following . in particular , according to section iv of ref .
, for the numerical results to be meaningful the coexisting densities @xmath97 and @xmath98 must maintain a separation of at least several density grid spacings @xmath118 from the boundaries at @xmath92 and @xmath119 ; consequently , @xmath120 should never exceed some maximum value , @xmath121 , and for the systems considered here and in ref . and for the typical choices for @xmath92 and @xmath93 the binodal s proximity to the low density boundary renders @xmath122 largely density grid- and @xmath107-independent . not to be confused with @xmath122 is the lowest temperature @xmath123 numerically accessible to the program with pre - determined step sizes : this is the temperature below which the program of section [ sec : program ] never reaches @xmath124 or produces abnormal results ; note that @xmath125 may be larger or smaller than @xmath122 , depending on the chosen combination of physical potential , approximations in the formulation used ( the boundary conditions in particular ) , and the choice of parameters affecting the numerical work .
regarding the implementation of the core condition as sketched in section [ sec : theory ] , the main conclusion of ref .
was that a minimum of @xmath126 basis functions in addition to @xmath96 were necessary for acceptable results despite residual defects of @xmath77 close to the origin ; a short discussion of the core condition s slightly different _ rle _ for square wells will be presented below ( sub - section [ sec : cc ] ) .
the critical density @xmath127 predicted by , it should be noted , is virtually always in reasonable agreement with literature data as shortly presented in sub - section [ sec : non - hrt ] ; indeed , is even able to reproduce the marked rise in @xmath127 predicted by refs . , , and for @xmath128 as opposed to the rigorously constant value in ref . .
due to the satisfactory @xmath127-values obtained numerically we will henceforth exclude @xmath127 from the discussion ; for a demonstration of both @xmath127 s insensitivity to variation of parameters of the numerical procedure and the quantitative agreement with the data of sub - section [ sec : non - hrt ] see figures [ fig : rhomax ] and [ fig : crit : lambda ] .
in this context it may be of interest that the estimate for the critical density presents no difficulties for the hard - core yukawa fluid considered in ref .
, either , nor is there any mention of such difficulties in any of the other publications on this topic that we are aware of ; indeed , the theory s numerical problems primarily lie in the solution s small-@xmath2 behavior for close - to - critical and sub - critical temperatures on the one hand and the use of mutually incompatible assumptions prompted by the need to employ decoupling without giving up thermodynamic consistency on the other hand .
both of these aspects of pertain to different parts of the @xmath57-plane , located close to the high density boundary for the _ rle _ mathematical inconsistencies play and at not too large @xmath2 and @xmath129 for the pathological behavior related to coexistence ; they will be discussed in sub - sections [ sec : boundary ] and [ sec : stiff ] , respectively , and their vestiges will also be seen in the results of applying in the formulation of section [ sec : theory ] to square wells of quasi - continually varying range @xmath130 1,3.6]$ ] in sub - section [ sec : interference ] . for comparison purposes
we have collected in tables [ tb : ref.sim ] and [ tb : ref.theo ] the critical temperatures of various square well systems as obtained from simulations ( table [ tb : ref.sim ] ) or by purely theoretical means ( table [ tb : ref.theo ] ) ; the data included have been published within the last decade . of the simulation based results included in table [ tb : ref.sim ] , only those of ref . for @xmath131 have been obtained by molecular dynamics ( md ) ; most of the other simulation studies rely on one or the other variant of the monte carlo ( mc ) method : among these , the gibbs ensemble mc ( gemc ) calculations of ref . set out to determine critical exponents , @xmath120 in particular ; that work s finding of @xmath132 for @xmath133 as opposed to the expected @xmath134 found for @xmath0 up to 1.75 prompted re - examination of the square well fluid with @xmath133 by gemc augmented by finite - size scaling ( fss ) techniques @xcite , refuting the mean field value for the effective exponent . especially in the critical regime , grand canonical mc ( gcmc ) simulations incorporating histogram reweighting and fss offer some advantage over gemc due to the latter s restriction to fixed temperature ; such an approach has been applied to square wells with @xmath23 and 3 in ref . ; a more elaborate gcmc scheme not biased towards the ising universality class and taking into account the yy anomaly has recently been applied to @xcite @xmath23 , _
v. s. _ yet another method goes under the name of thermodynamic- or temperature - and - density - scaling mc ( tdsmc ) ; it was applied to the case of @xmath23 and analyzed in terms of an effective hamiltonian in refs . and .
also included in table [ tb : ref.sim ] are the results of ref . , employing an mc scheme modified to take advantage of a speed - up possible by combining simulation data with an analytical _ ansatz _ for the chemical potential ; the efficiency of this approach originally devised to study phase separation allows a large number of systems to be considered .
( the error bounds given for these `` modified mc '' results in table [ tb : ref.sim ] have been obtained from the different results displayed in ref . for different parameter settings .
) the theoretical predictions for the critical temperature listed in table [ tb : ref.theo ] comprise a second - order analytic perturbation theory @xcite ( apt2 ) applicable to @xmath135 and claimed accurate for @xmath136 as well as the hard - sphere van der waals ( hsvdw ) equation of state @xcite .
in addition , though not listed in table [ tb : ref.theo ] , we have utilized the non - square - well - specific okumura - yonezawa ( oy ) estimate @xcite for @xmath137 , primarily as a starting value when looking for the critical temperature in our calculations ; for @xmath11}$ ] , the oy prediction is @xmath138 .
ever since application of to continuous fluids started , the implementation of the core condition has been a major issue , probably motivating adoption of the closure ( [ eq : closure ] ) and variants thereof for non - hard - sphere reference systems @xcite despite its known deficiencies in the first place ; indeed , it is no coincidence that several studies @xcite primarily concerned with the aspect of the theory chose to completely eliminate the core condition . when applying as a regular liquid state theory , on the other hand , this is not an option : too great is the effect this may have on both correlation functions and phase behavior @xcite . from table
[ tb : ncc ] where we compile the critical temperature @xmath139 for various square well potentials as functions of the number @xmath140 of basis functions in the closure ( [ eq : closure ] ) , just as in ref .
we find virtually constant critical temperatures for @xmath141 ; on the other hand , the amount of variation seen upon further increasing @xmath74 strongly depends on @xmath0 , which immediately carries over to the pair distribution function @xmath142 and its compatibility with the core condition : for @xmath143 , the longest ranged potential considered in table [ tb : ncc ] , @xmath144 , @xmath145 , holds reasonably well except very close to @xmath146 even for @xmath147 ; when increasing the number of basis functions all the way to @xmath148 , the pair distribution function has to be corrected for very small @xmath26 only , yielding a @xmath149 that remains bounded by some @xmath150 of the contact value @xmath151 for all @xmath145 ; the corresponding small change in @xmath142 and @xmath152 is reflected in the near - constant predictions for @xmath137 evident from table [ tb : ncc ] .
similarly , for @xmath23 and @xmath133 and within the @xmath74-range considered , the implementation of the core condition does not convincingly improve except for supercritical temperatures and intermediate densities ; this time , however , the pair distribution functions remain far from compatible with the core condition even for @xmath148 , and neither @xmath137 nor @xmath142 itself nor , for that matter , the final values of the / expansion coefficients @xmath153 indicate that the expansion ( [ eq : closure ] ) for @xmath154 might be close to convergence .
but if the quality of @xmath142 improves only little if at all , the remaining deficiencies are probably to be blamed on the approximation for the poorly convergent integrals derivative with respect to @xmath2 mentioned earlier ( cf .
( 12 ) of ref . ) rather than on an insufficient number of basis functions ; on the other hand , even though the decoupling assumption can not directly affect the pair distribution function s compliance with the core condition , the approximation of neglecting the non - local term in @xmath155 / \partial q$ ] is on the same level as that of setting @xmath78 , as was stressed by the authors of ref . upon jointly introducing these two assumptions .
thus , combining the above findings regarding the core condition with the analogous analysis of sub - section iv of ref . and with that contribution s investigation into the decoupling assumption s possible effect ( cf .
fig . 2 of ref . )
we are led to the conclusion that decoupling poses certainly no less a problem here than for the hard - core yukawa potential studied there . numerically , there are two ways for the implementation of section [ sec : program ] to fail to reach @xmath156 , both , of course , easily detected by the `` monitoring '' variant of our code ( cf .
sub - section iii eof ref . ): due to the solution s pathological behavior wherever @xmath81 is large ( cf .
sub - section [ sec : stiff ] ) , or because of inappropriate boundary conditions at high density . as for the latter
an issue intimately linked to the decoupling assumption , the immediate reason for the program s failure is a near - discontinuity in the numerical solution close to the boundary : for the moment setting aside the decoupling assumption and other approximations , in the application of with the closure ( [ eq : closure ] ) at any point @xmath57 in the interior of the s domain the core condition uniquely determines the @xmath66 , @xmath62 , for given @xmath157 ; this expansion coefficient is then determined by imposing thermodynamic consistency as embodied in the compressibility sum - rule ( [ eq : consistency ] ) . at a boundary , _
i. e. _ for @xmath158 ,
however , the second density derivative can not be evaluated reliably so that some other condition must be imposed ; in the calculations reported here ( with the obvious exception of those for fig .
[ fig : c : interference ] ) we always choose @xmath159 so that the divergence of the ideal gas term in @xmath160 provides @xmath161 as a convenient and unproblematic boundary condition .
for @xmath162 , on the other hand , we are in principle free to use any suitable approximation for the structural and thermodynamic properties of the @xmath2-system and to calculate @xmath163 from said approximation , thereby providing the necessary boundary condition for the ( [ eq : pde : f ] ) ; but for practical reasons it is desirable to use the same /-form for the @xmath2-system s direct correlation function at @xmath93 as in the rest of the problem s domain so that , in particular , the / prescription @xmath94 is a natural choice of boundary condition . in general , however , due to the s diffusion - like character any condition imposed at @xmath93 that is incompatible with the solution for @xmath164 by necessity induces a corresponding near - discontinuity in @xmath81 close to the boundary ; within the framework of a finite difference scheme this is reflected in a mismatch of @xmath163 and the solution at densities close by , _
i. e. _
@xmath165 for small @xmath166 , and the mismatch s severity may serve as a direct measure for the inappropriateness of the boundary condition at @xmath93 in relation to the approximations applied at densities in @xmath167{\rho_{\rm min } } , { \rho_{\rm max}}\right[$ ] . on the other hand , the numerics become intractable unless we adopt the decoupling assumption , and the only way to consistently use @xmath78 without abandoning the core condition is to decouple the - to a set of s at fixed density only @xcite ; this , unfortunately , removes all traces of thermodynamic consistency from the equations and thereby precludes obtaining clear phase boundaries @xcite .
it is therefore necessary to restrict decoupling to the implementation of the core condition only while retaining the structure of a together with the compressibility sum rule ( [ eq : consistency ] ) despite the latter s incompatibility with decoupling .
thus , for @xmath168 , both @xmath169 and @xmath78 are used for different parts of the problem ; at @xmath93 , however , again any approximation allowing calculation of @xmath163 may be used , so that it is tempting to once again resort to the /-condition of vanishing @xmath170 or variants thereof .
but due to the decoupling assumption s possibly large effect , any boundary condition that does not incorporate @xmath95 and bear in mind that @xmath170 and @xmath171 can not both vanish at the same time for generic cut - off @xmath2 will once again incur a fatally large mismatch ; if , however , we must resort to decoupling anyway , it seems preferable to consistently apply it for the boundary condition rather than to inconsistently combine it with a condition alien to the theory ; also , though the mismatches magnitudes from imposing @xmath95 alone or from mixing it with the / condition @xmath94 generally do not differ much as long as the s stiffness does not play a _
g. _ , for @xmath23 , as long as we restrict ourselves to @xmath172 or higher , or to @xmath173 ) , the former approach fares better than the other one more often than not .
it is only in this sense , _
i. e. _ presupposing a /-like _ ansatz _ even at @xmath93 and application of decoupling in the implementation of the core condition according to eq .
( [ eq : cc : cint ] ) at all @xmath47 , that the results are largely independent of the choice of boundary condition as claimed , for @xmath174 , in ref . . in the numerical work we find that such a mismatch is present whenever
the calculation proceeds _ via _ mathematically inconsistent or conflicting approximations ; in the case of square wells with their comparatively short potential range , however , the problems are much more severe than in other systems so that @xmath125 is rather small and even drops below @xmath137 for most of the @xmath0 interval from 1 to 2 ( cf .
sub - section [ sec : interference ] ) . restricting ourselves to @xmath175 and @xmath156 , the mismatch is typically reflected in an increase by one order of magnitude in the three - point finite - difference estimate of , _
e. g. _ , @xmath176 right at the boundary over the near - constant values at slightly lower densities ; apart from a positive correlation with @xmath107 , the mismatch s severity is qualitatively unaffected by a change in parameters of the numerical procedure or the choice and location of the boundary condition ( with the above provisions ) .
another effect worth mentioning in connection with the boundaries is the influence their locations , _ viz .
_ @xmath92 and @xmath93 , may have . the basic mechanism and its implications for the coexisting densities
were already mentioned in the opening remarks of this section ; here we only want to point out that the non - criticality enforced by the boundary conditions not only may unduely distort the binodal predicted by as demonstrated in fig . [ fig : rhomax ] , very small @xmath93 may also allow one to reach @xmath156 at higher @xmath120 , thus effectively raising @xmath125 while lowering @xmath122 .
sometimes , however , the expectation of the binodal keeping a separation from the boundary of several @xmath118 at least does not hold , and a preposterous two - phase region appears very close to @xmath93 or , very rarely , close to @xmath92 ; _
e. g. _ , for @xmath177 and @xmath178 the equations can be solved all the way down to @xmath179 , predicting an unrealistic two - phase region extending from @xmath180 to @xmath181 .
this behavior turns out to come from the interplay of the mismatch at @xmath93 and the numerical treatment of the stiffness of the . for subcritical temperatures
the - s true solution s erratic behavior in that part of the @xmath57-plane where @xmath81 is large and the isothermal compressibility s divergence is built up ( cf .
section [ sec : theory ] , _ q. v. _ ref . ) obviously eludes reliable numerical realization ; in particular , while @xmath107 still characterizes the level of accuracy in auxiliary calculations , the same can no longer be true for the accuracy of the s discretization as this would require step sizes @xmath109 so small as to cause floating point underflow upon evaluating , _
e. g. _ , @xmath182 , thus rendering finite differences numerically insignificant .
consequently , in this respect we have to give up our strategy of controlling the numerical procedure so as to locally ensure a quality of @xmath107 at least , turning to pre - determined step sizes @xcite @xmath109 in addition to fixed @xmath118 , to which similar concerns apply @xcite ; on such a coarse mesh of @xmath57-points underlying the finite difference scheme , however , the true solution can not even be represented adequately , and the numerical approximation for @xmath81 obtained from the s discretization with these far too large step sizes can not be trusted to faithfully represent even the average behavior of @xmath81 .
this inadequacy of the step sizes is reflected in various peculiarities of the solution vector obtained in the numerical procedure ; indeed , when monitoring the evolution of @xmath81 and the core condition coefficients @xmath75 , our code readily detects the plummeting step sizes necessary and signals the incompatibility of the behavior seen with the assumption of smoothness underlying finite difference schemes .
another telltale sign is iterated corrector steps failure to converge when @xmath81 is large : even though implicit schemes like the one we employ @xcite are the standard treatment for stiff systems , the rapid growth of the oscillations amplitudes renders the finite difference equations themselves unstable under iteration ; only when resigning on any control of the numerical error and refraining from iterations of the corrector step do the step sizes @xmath109 chosen allow one to force advancing @xmath2 all the way to @xmath42 in remarkably many cases .
also , comparison of @xmath183 as obtained with different sets of step sizes @xmath109 reveals that , for @xmath184 , the evolution of @xmath81 seen numerically is driven by the number and size of @xmath2-steps only and certainly does not correspond to an average over oscillations @xcite ; the same mechanism is also responsible for a small @xmath109-dependence of the critical temperature @xmath137 . by the same token , due to the @xmath185- and @xmath186-terms in eq .
( [ eq : pde : f ] ) , the s stiffness and the related problems have a direct bearing on the solution outside the coexistence region even if the numerical predictions there turn out rather insensitive to variation of parameters of the numerical procedure ; in particular , we expect a gradual but non - negligible distortion ( in addition to the effects of numerical differentiation close to the near discontinuity ) of the binodal , increasing with falling temperature . in the light of the preceding exposition as well as of the discussion in ref .
it may at first seem surprising that in the formulation of section [ sec : theory ] has a record of being well applicable to a variety of systems ( cf .
section [ sec : intro ] ) ; also , as we shall see in a moment , even for square wells , a system expected to be particularly vulnerable to the problems just outlined , we find reasonable estimates of the critical points locations for a wide range of @xmath0-values .
still , the mechanisms sketched in subsections [ sec : boundary ] and [ sec : stiff ] as well as the difficulties presented in ref .
remain and manifest themselves numerically in a number ways . for a first orientation ,
let us look at the results summarized in figs .
[ fig : crit : lambda ] and [ fig : crit : lambda : details ] where the critical temperature @xmath187 and density @xmath127 are shown as functions of @xmath0 ; the underlying calculations have been obtained with @xmath112 , imposing decoupling in a consistent way at @xmath111 and with @xmath188 basis functions in the expansion ( [ eq : closure ] ) of the /-function @xmath189 . with the exception of some spurious results at @xmath190 ,
whereever @xmath191 the critical temperature in general compares quite favorably with the data of tables [ tb : ref.sim ] and [ tb : ref.theo ] ; from the calculations we have performed for a large number of systems in the range @xmath192 and ignoring some isolated results , a critical point is found for @xmath193 , for @xmath194 , and for @xmath195 ; calculations with @xmath196 yield analogous results @xcite , with @xmath191 in a somewhat larger part of the parameter range , _ viz .
_ for @xmath197 and for @xmath198 , but will not be considered in the following in view of the considerations of sub - section [ sec : cc ] and of other defects that turn out to be larger than for @xmath126 . for the moment setting aside the data for @xmath199 , s predictions for the critical temperature
are generally found to be in satisfactory agreement with the @xmath200-curve expected from the simulation - based and theoretical results presented in sub - section [ sec : non - hrt ] .
embedded into this regular overall behavior of @xmath137 as a function of @xmath0 , however , we find a number of depressions and elevations of @xmath137 , some of which can not be seen on the scale of the plot [ fig : crit : lambda ] but from the numeric results only @xcite ; others , however , are so strong as to render the critical temperature a non - monotonic function of @xmath0 , which is certainly not expected on the grounds of the literature presented in sub - section [ sec : non - hrt ] , the data of refs . in particular . in the light of sub - section
[ sec : stiff ] it is of course tempting to simply attribute this behavior to the difficulties previously discussed , especially since the critical point is located in the region of large @xmath183 by definition ; the peculiar distribution of @xmath0-values affected , however , suggests that these problems of the numerical procedure are triggered by a special mechanism .
indeed , a closer look at the core condition function @xmath201 for fixed density @xmath47 reveals , for every single one of the @xmath0 values implicated that we checked , that the combination of terms pertaining to @xmath8 or @xmath202 alone ( of ranges @xmath203 and @xmath13 , respectively ) regularly and quite frequently reduces the amplitude of this function s swings about the ideal gas value of @xmath87 ; the same happens only occasionally for @xmath0-values removed from these irregularities so that it is , in fact , possible to quite reliably determine whether or not a given @xmath0 is affected from a plot of @xmath201 for @xmath129 alone as illustrated in fig .
[ fig : c : interference ]
. it will come as no surprise that most of the irregularities occur when @xmath0 , the ratio of the two characteristic lengths present in the model , is close to a simple fraction : among the shifts in @xmath187 most obvious are those where @xmath0 is close to @xmath204 ( cf .
[ fig : crit : lambda : details ] ) , @xmath205 , @xmath206 , @xmath207 and @xmath208 , and in retrospect it seems justified to also include the small parameter range around @xmath209 in this list , _ v. i. _ ; the effect is less obvious from fig .
[ fig : crit : lambda ] but still discernible at @xmath210 , @xmath211 and @xmath212 , whereas for @xmath205 and @xmath213 it is so small as to make the plot of @xmath200 appear smooth while the irregularities are still evident from the numerical values ; also note that , once again , @xmath127 is hardly affected .
all these observations seem to indicate that indeed it is the interplay of the two different length scales and the resulting partial oppression of a significant portion of the oscillations of @xmath201 that cause the discrepancy of and literature results for the critical temperature around certain @xmath0 values even though a smooth interpolation of s predictions from @xmath0 values nearby is well compatible with the data presented in sub - section [ sec : non - hrt ] . even though we currently can not pinpoint the precise mechanism by which this unphysical behavior of arises and , in particular , can not distinguish between the closure s inadequacy and the s stiffness as the main culprit though the latter is certainly implicated to some degree , two conclusions may be drawn quite safely : for one , as long as we stay clear of values of @xmath1 that are close to simple fractions , or restrict ourselves to @xmath214 where the effects are rather small , we can probably trust the numerical results with the _ caveats _ of ref . and sub - sections [ sec : boundary ] and [ sec : stiff ] to the same degree of confidence as those obtained for the hard - core yukawa system in ref . .
and secondly , it is only in the presence of discontinuities in the potential that certain lengths feature prominently in the relevant functions fourier transforms and can so give rise to problems of the kind outlined above ; consequently , as long as we confine ourselves to continuous @xmath215 , which still includes most of the potentials popular in liquid state physics , the unphysical shifts in @xmath137 seen for certain parameter combinations are likely not an issue , whereas the same problems are expected to resurface , _
e. g. _ , for the multi - step potential also defined in sub - section iii b of ref . .
another lesson to be drawn from the findings presented here as well as in ref .
is that , as a general rule , conclusions should never be drawn from isolated results alone ; it is only through the combination and meticulous scrutiny of a set of related calculations that meaningful information can be extracted from calculations : due to the problems related to the implementation of the core condition , to the nature and location of the boundary conditions , and to the s stiffness , any single calculation must be considered as of uncertain standing . as an example @xcite , the analogue of fig . [
fig : rhomax ] for @xmath23 shows a considerably larger variation in the binodal and the critical point s location , which is consistent with the above conclusions regarding the reason for @xmath125 rising above @xmath137 in a narrow region around this @xmath0 value , whereas any one of the phase boundaries found in itself is a perfectly plausible candidate for the `` true '' binodal .
in conjunction with the findings of ref . , the discussion of section [ sec : application ] provides quite coherent a picture of s numerical side as well as of some peculiarities encountered for square wells .
most prominently , we see a marked dependence of the quality of the results on the potential s range , confirming the trend of decreasing accuracy for narrower potentials reported @xcite for the hard - core yukawa fluid ; it has long been accepted @xcite that the simplistic but computationally convenient @xcite closure ( [ eq : closure ] ) has a part in this , and an improved closure has recently been proposed @xcite .
still , as far as numerical application of is concerned the closure can not be discussed without reference to the decoupling assumption and to the approximate implementation of the core condition _ via _
s coupled to the - ; and while the former has been found problematic both for square wells ( present contribution ) and for the hard - core yukawa fluid considered in ref . and should probably not be trusted easily for any system , the severity of the difficulties brought about by the simplified treatment of the core condition sensitively depends on the potential type and parameters chosen : for the continuous and rather long ranged yukawa potential with @xmath80 , @xmath142 can be made sufficiently small within the core , and the square well fluid with @xmath143 fares equally well at least ; from the discussion of sub - section [ sec : cc ] and the data of table [ tb : ncc ] , however , it becomes apparent that smaller @xmath0 we have looked at @xmath23 and @xmath133 in particular incurs substantial problems , with residual defects in the pair distribution functions attributed to the ill - justified approximation of neglecting some slowly converging integrals @xcite in the core condition .
but table [ tb : ncc ] demonstrates not only the @xmath0-dependence of the results sensitivity to the number @xmath216 of basis functions retained in the truncated eq .
( [ eq : cc : cint ] ) when varying @xmath217 in the range @xmath218 : while the virtually constant critical temperature predicted for @xmath143 seems trustworthy and is , indeed , well compatible with simulation results ( cf.table [ tb : ref.sim ] ) , the amount of variation in @xmath137 for @xmath23 and , to a much lesser degree , for @xmath133 precludes accurate determination of the critical temperature ; this is a first indication that the theory might be able to handle square wells with @xmath143 quite reliably whereas problems can not be denied for @xmath133 , and @xmath23 seems largely out of reach for in the present formulation .
this is confirmed by the results obtained by quasi - continuous variation of @xmath0 in the range @xmath192 as shown , for @xmath126 , in fig .
[ fig : crit : lambda ] : the critical point is accessible only in part of this parameter range , and not only the critical temperature at fixed @xmath0 but also the boundaries of the @xmath0-intervals where is able to reach temperatures as low as @xmath187 strongly depend on @xmath74 ( cf.sub-section [ sec : interference ] ) .
applicability of to only a restricted @xmath0-range is , of course , again related to the pronounced short - rangedness of the square well potential ; to explain it , however , we have to invoke not only the /-style closure and the approximate implementation of the core condition but also the other difficulties encountered in the numerical procedure as highlighted in this and our preceding contribution , _ viz . _ the decoupling assumption ( ref . ) , inappropriate boundary conditions ( sub - section [ sec : boundary ] ) and , most importantly , the s stiffness for thermodynamic states of high compressibility ( sub - section [ sec : stiff ] ) .
all of these are , in principle , always present to some degree in numerical applications of ; it may prove valuable that sub - sections [ sec : boundary ] and [ sec : stiff ] provide distinct signatures readily detected by the implementation of section [ sec : program ] that might be looked out for in more advanced applications of the theory , too .
related to these difficulties is a peculiar effect specific to square wells ( sub - section [ sec : interference ] ) : close to certain @xmath0 values , simple fractions in particular , we see shifts in the critical temperature that render s predictions much less compatible with simulations and other theoretical descriptions of the square well fluid than would be expected from the results obtained for @xmath0 values close by ; the mechanism for triggering these local distortions of the @xmath200-curve is illuminated at least to the point of linking it to the presence of a discontinuity in the potential s perturbational part .
all in all , the numerical evidence as well as comparison with literature data suggest that the formulation of sketched in section [ sec : theory ] is well able to deal with square wells and to locate their critical points to reasonable accuracy for @xmath1 as long as certain values are avoided , or else for @xmath214 .
the authors gratefully acknowledge support from _ sterreichischer forschungsfonds _ under project number p13062-tph .
.the critical temperature @xmath187 of square well systems for various values of @xmath0 as predicted by simulations and simulation - based theoretical analyses , and the corresponding references .
the acronyms used for labeling the method employed in obtaining these results are given in sub - section [ sec : non - hrt ] of the text . [ cols="^,^,<,^",options="header " , ]
|
continuing our investigation into the numerical properties of the _ hierarchical reference theory _ , we study the square well fluid of range @xmath0 from slightly above unity up to 3.6 .
after briefly touching upon the core condition and the related decoupling assumption necessary for numerical calculations , we shed some light on the way an inappropriate choice of the boundary condition imposed at high density may adversely affect the numerical results ; we also discuss the problem of the partial differential equation becoming stiff for close - to - critical and sub - critical temperatures . while agreement of the theory s predictions with simulational and purely theoretical studies of the square well system is generally satisfactory for @xmath1 , the combination of stiffness and the closure chosen is found to render the critical point numerically inaccessible in the current formulation of the theory for most of the systems with narrower wells .
the mechanism responsible for some deficiencies is illuminated at least partially and allows us to conclude that the specific difficulties encountered for square wells are not likely to resurface for continuous potentials .
# 1^(#1 ) c_2
|
it is well known that free groups are ubiquitous in the automorphism group of an infinite rooted spherically homogeneous tree , see for instance @xcite , though explicit examples ( especially ones generated by finite automata ) were not so easy to construct , see @xcite . on the other hand , many famous groups , which are defined by their action on rooted trees do not have free subgroups .
absence of free subgroups is proved in different ways . in some cases it follows from torsion or sub - exponential growth ( as in the grigorchuk groups @xcite and gupta - sidki groups @xcite ) . in other cases
it is proved using some contraction arguments ( see , for instance @xcite ) .
s. sidki has proved in @xcite absence of free groups generated by `` automata of polynomial growth '' , which covers many examples .
an important class of groups acting on rooted trees are the _ contracting self - similar groups_. they appear naturally as iterated monodromy groups of expanding dynamical systems ( see @xcite ) . there are no known examples of contracting self - similar groups with free subgroups and it was a folklore conjecture that they do not exist .
the intuition behind this conjecture and the theorems mentioned above is that there is no sufficient `` room '' for free subgroups .
the action of the elements of the groups in all examples are concentrated on small portions of the tree and the graphs of the action of the groups on the boundary of the rooted tree are also small in some sense .
for instance , the graphs of the action of contracting groups have polynomial growth .
we prove in our paper the following theorem formalizing this intuition .
* theorem [ th : mainlemma ] .
* _ let @xmath0 be a group acting faithfully on a locally finite rooted tree @xmath1 .
then one of the following holds .
@xmath0 has no free non - abelian subgroups , 2 .
there is a free non - abelian subgroup @xmath5 and a point @xmath4 such that the stabilizer @xmath6 is trivial , 3 .
there is a point @xmath4 such that the group of @xmath0-germs @xmath7 has a free non - abelian subgroup . here
the group of @xmath0-germs @xmath7 is the quotient of the stabilizer @xmath8 by the subgroup of automorphisms @xmath9 of the tree @xmath1 acting trivially on a neighborhood @xmath10 of @xmath2 .
theorem [ th : mainlemma ] was inspired by a result of m. abrt implying that if @xmath11 is a free group acting faithfully and level transitively on a rooted tree , then there exists a point of the boundary of the tree having trivial stabilizer in @xmath11 , see @xcite . even though the theorem itself is not very complicated
, it gives a way to find simple proofs of absence of free subgroups in many groups acting on rooted trees .
one has to show that the graphs of the action of the group on the boundary are so small that a free action of a free subgroup is not possible , and then to analyze the action of the elements of the group on neighborhoods of fixed points in order to show that the groups of germs of the action are also small and have no free subgroups .
in particular , we confirm the conjecture on contracting groups
. * theorem [ th : contracting ] . * _ contracting groups have no free subgroups . _ this theorem implies , for instance , that the iterated monodromy groups of post - critically finite rational functions and other expanding dynamical systems ( see @xcite ) have no free subgroups .
it is an interesting open question if all contracting groups are amenable .
we also generalize ( in theorem [ th : bounded ] ) the fact that there are no free groups generated by bounded automorphisms of a rooted tree . for the notion of bounded automorphisms see definition [ def : bdd ] of our paper and the articles @xcite .
it is known that groups generated by bounded automorphisms _ defined by finite automata _ have no free subgroups @xcite and that they are even amenable @xcite .
absence of free subgroups for the general case of bounded automorphisms is proved here for the first time .
it is not known if they are all amenable .
we also give a shorter proof of the theorem of s. sidki @xcite about automata of polynomial growth ( only for the case of finite alphabets ) .
a very intriguing open question now is to see if all the groups covered by these theorems are amenable and to prove a theorem on amenability of groups acting on rooted trees similar to theorem [ th : mainlemma ] . the first more or less general result in this direction is the proof of amenability of groups generated by bounded automata in @xcite
here we recall the basic notions related to rooted trees and fix notation .
the reader can find more on this in @xcite .
a _ rooted tree _ is a tree with a fixed vertex called the _ root _ of the tree .
we consider only locally finite trees in our paper , i.e. , trees in which every vertex belongs to a finite number of edges .
we say that a vertex @xmath12 of a tree @xmath1 is _ below _ a vertex @xmath13 if the path from the root of @xmath1 to @xmath12 goes through @xmath13 .
we denote by @xmath14 the subtree of all vertices which are below @xmath13 together with @xmath13 serving as a root of @xmath14 .
see figure [ fig : t ] .
+ the _ boundary _ @xmath3 of the tree @xmath1 is the set of simple infinite paths starting at the root of @xmath1 .
the boundary @xmath15 of a subtree @xmath14 consists then of the paths going through the vertex @xmath13 .
the collection of subsets @xmath15 of @xmath3 is a basis of topology on @xmath3 .
the topological space @xmath3 is compact and totally disconnected .
if @xmath16 is a subtree of @xmath1 , then @xmath17 is a closed subset of @xmath3 in the natural way .
an automorphism of the rooted tree @xmath1 is an automorphism of the tree @xmath1 fixing the root vertex .
every automorphism of @xmath1 fixes also the _ levels _ of the tree as sets . here a _ level number @xmath18 _ of the rooted tree @xmath1 is the set @xmath19 of the vertices on distance @xmath18 from the root . a rooted tree @xmath1 is said to be _ spherically homogeneous _
if the automorphism group of @xmath1 is transitive on the levels .
if @xmath1 is spherically homogeneous , then @xmath3 is equipped with a natural probability measure @xmath20 defined by the condition that measure of @xmath15 is equal to @xmath21 , where @xmath19 is the level of the vertex @xmath13 .
this is the unique probability measure invariant under the action of the automorphism group of @xmath1 .
let @xmath22 be a finite set , called _ alphabet _ and let @xmath23 be the free monoid generated by @xmath22 , i.e. , the set of finite words @xmath24 over the alphabet @xmath22 , including the empty word @xmath25 .
the set @xmath23 has a natural structure of a rooted tree , where a vertex @xmath26 is connected to the vertices of the form @xmath27 and the empty word is the root .
we denote by @xmath28 the sub - tree of words starting by @xmath13 , i.e. , the sub - tree @xmath29 of vertices below the vertex @xmath13 .
the boundary @xmath30 is naturally homeomorphic to the set of infinite sequences @xmath31 with the product topology ( where @xmath22 is discrete ) . more generally ,
if @xmath32 is a sequence of finite sets , then we denote @xmath33 where @xmath34 for @xmath35 and @xmath36 .
we denote by @xmath37 the length of a word @xmath38 , i.e. , number such that @xmath39 .
the set @xmath40 is a rooted tree with the root @xmath25 in which an element @xmath41 is connected to all elements of the form @xmath27 for @xmath42 .
the boundary of the tree @xmath40 is homeomorphic to the direct product @xmath43 of discrete sets .
the invariant measure on @xmath44 coincides in this case with the _ uniform bernoulli _ measure defined as the direct product of the uniform distributions on @xmath45 .
we denote by @xmath46 the sequence @xmath47 if @xmath9 is an automorphsm of the tree @xmath40 , then for every @xmath38 there exists a unique automorphism @xmath48 of the tree @xmath49 such that @xmath50 for all @xmath51 .
it is easy to see that the following properties of this operation hold @xmath52 let @xmath9 be an automorphism of the tree @xmath40 .
a sequence @xmath53 is _
@xmath9-regular _ if there exists a beginning @xmath38 of @xmath2 such that @xmath48 is trivial .
we say that @xmath53 is _ @xmath9-singular _ if it is not @xmath9-regular .
the set of @xmath9-singular points of @xmath44 can be measured using the _ growth _ of the automorphism @xmath9 .
[ def : growthfunction ] let @xmath9 be an automorphism of the spherically homogeneous rooted tree @xmath40 .
then its _ growth function _
is @xmath54 if @xmath55 is a _ rooted _ subtree of @xmath40 , i.e. , a sub - tree containing the root of @xmath40 , then the _ relative growth function _ is @xmath56 and @xmath9 is _ almost finitary on @xmath1 _ if @xmath57 where @xmath58 is the @xmath18th level of the tree @xmath1 .
[ pr : measureregular ] let @xmath1 be a spherically homogeneous rooted subtree of @xmath40 and let @xmath9 be an automorphism of @xmath40 .
then @xmath59 where @xmath19 is the @xmath18th level of the tree @xmath1 and @xmath60 is the set of @xmath9-singular points @xmath61 .
it is easy to see that the set of @xmath9-regular points @xmath53 is open , hence the set @xmath60 is closed in @xmath3 .
it follows from by definitions of @xmath62 and the measure @xmath20 that the number @xmath63 is equal to @xmath64 , where @xmath65 the sequence of the sets @xmath66 for @xmath67 is decreasing and their intersection is @xmath60 , since @xmath60 is closed .
this finishes the proof , by continuity of the measure .
recall that if we have a group @xmath0 generated by a finite symmetric set @xmath68 and acting on a set @xmath69 , then the corresponding schreier graph is the graph with the set of vertices @xmath69 and the set of edges @xmath70 , where an edge @xmath71 starts in @xmath72 and ends in @xmath73 .
a locally finite graph is said to be _ amenable _ if for every @xmath74 there exists a finite set @xmath75 such that @xmath76 where @xmath77 is the set of edges beginning in @xmath75 and ending outside of @xmath75 . a regular tree , in particular the cayley graph of a free non - abelian group , is an example of a _ non_-amenable graph . [ pr : griamenable ]
let @xmath0 be a finitely generated automorphism group of @xmath40 and let @xmath55 be a @xmath0-invariant rooted subtree on which @xmath0 acts level transitively .
if all elements of @xmath0 are almost finitary on @xmath1 , then all components of the schreier graph of the action of @xmath0 on @xmath78 are amenable .
the proposition follows directly from the theorem of @xcite ( which in turn is a corollary of a result of v. kaimanovich @xcite ) .
but we prefer to give here a direct proof constructing the sets @xmath75 ( called _ flner sets _ ) .
fix a finite symmetric generating set @xmath68 of @xmath0 and consider the schreier graph @xmath79 of the action of @xmath0 on the @xmath18th level @xmath19 of the tree @xmath1 .
let @xmath80 be the subgraph of @xmath79 which consists only of the edges @xmath81 such that @xmath82 is trivial .
note that if @xmath81 belongs to @xmath80 , then the inverse edge @xmath83 also belongs to @xmath80 .
the number of edges in the difference @xmath84 is equal to @xmath85 .
let @xmath86 be the connected components of @xmath80 ( as sets of vertices ) .
then in @xmath79 we have @xmath87 hence there exists a component @xmath88 such that @xmath89 consider an orbit @xmath90 of the action of @xmath0 on @xmath3 .
since the action of @xmath0 on @xmath1 is level - transtive , there exists @xmath91 such that the beginning @xmath92 of length @xmath18 of @xmath2 belongs to @xmath88 .
let @xmath93 be such that @xmath94 . by the definition of @xmath88 , for every @xmath95 there exists an element @xmath96 such that @xmath97 and @xmath98 ( see the first two properties of in subsection [ ss : words ] ) .
then @xmath99 , i.e. , the point @xmath100 belongs to the orbit @xmath90 .
let @xmath101 if the edge @xmath81 from @xmath95 to @xmath102 belongs to @xmath80 , then @xmath103 , hence @xmath104 belongs to @xmath88 .
consequently , in the schreier graph of the action of @xmath0 on @xmath90 we have @xmath105 as @xmath106 .
this means that @xmath107 are flner sets , i.e. , that the schreier graph of @xmath90 is amenable .
we will use the following simple lemmata .
their proofs are known ( see , for instance proposition 1 of @xcite for lemma [ lemma2 ] ) , but we provide them for completeness . [ lemma1 ] let @xmath11 be a free non - abelian group and let @xmath108 be a cyclic subgroup .
then there exists a free non - abelian subgroup @xmath109 such that @xmath110 is trivial .
we can find a subgroup of @xmath11 freely generated by the generator @xmath111 of @xmath112 and two other elements @xmath113 of @xmath11 ( take , for instance , any element @xmath9 which does not commute with @xmath111 and then take the index 2 subgroup of the free group @xmath114 generated by @xmath115 and @xmath116 ) .
then we can take @xmath117 .
[ lemma2 ] let @xmath11 be a free non - abelian group and let @xmath118 be a homomorphism into a finite direct product of groups .
if every composition @xmath119 of @xmath120 with the projection @xmath121 has a non - trivial kernel , then @xmath120 has a non - trivial kernel .
it is sufficient to prove the lemma for @xmath122 .
the general statement will follow then by induction .
let @xmath123 and @xmath124 be non - trivial elements of the kernels of @xmath125 and @xmath126 , respectively . if @xmath123 and @xmath124 belong to one cyclic subgroup of @xmath11 , then there exist @xmath127 such that @xmath128 is a non - trivial element of the kernel of @xmath120 .
if they do not belong to a common cyclic subgroup , then they do not commute and @xmath129 $ ] is a non - trivial element of the kernel of @xmath120
. let @xmath0 be a group acting on a topological space @xmath130 .
the _ group of @xmath0-germs _ of a point @xmath131 is the quotient of the stabilizer of @xmath72 in @xmath0 by the subgroup of elements acting trivially on a neighborhood of @xmath72 .
we denote it by @xmath132 . informally speaking
, @xmath132 describes the action of the stabilizer @xmath133 locally on neighborhoods of @xmath72 .
see an application of groups of germs to growth of groups in @xcite .
[ th : mainlemma ] let @xmath0 be a group acting faithfully on a locally finite rooted tree @xmath1 .
then one of the following holds . 1 .
@xmath0 has no free non - abelian subgroups , 2 .
there is a free non - abelian subgroup @xmath5 and a point @xmath4 such that the stabilizer @xmath6 is trivial , 3 .
there is a point @xmath4 such that the group of @xmath0-germs @xmath7 has a free non - abelian subgroup .
suppose that on the contrary , @xmath0 has a free subgroup @xmath11 , the groups of @xmath0-germs @xmath7 for all @xmath4 have no free subgroups and there is no free subgroup @xmath134 and a point @xmath4 such that the stabilizer @xmath135 is trivial .
for every point @xmath4 the stabilizer @xmath6 contains a free non - abelian subgroup , since otherwise @xmath6 is cyclic or trivial , and then we can find by lemma [ lemma1 ] a free non - abelian subgroup @xmath136 such that @xmath137 is trivial .
but the group @xmath7 has no free subgroups , consequently the natural homomorphism @xmath138 has a non - trivial kernel .
this means that there exists a vertex @xmath139 on the path @xmath2 and a non - trivial element of @xmath11 acting trivially on the sub - tree @xmath140 .
we get a covering of @xmath3 by open subsets @xmath141 .
the boundary of the tree is compact , hence there exists a finite sub - covering .
this means that there exists a finite set of vertices @xmath142 such that @xmath143 and for every @xmath144 the group of the elements of @xmath11 acting trivially on @xmath14 is non - trivial , hence infinite
. denote by @xmath145 the intersection of the stabilizers of the vertices of @xmath142 in @xmath11 .
it is a subgroup of finite index in @xmath11 .
consider the homomorphism @xmath146 mapping an automorphism @xmath147 to the automorphisms of the trees @xmath14 that it induces .
composition of @xmath120 with the projection onto @xmath148 has non - trivial kernel for every @xmath149 , since @xmath145 has finite index in @xmath11 and the group of elements of @xmath11 acting trivially on @xmath14 is infinite for each @xmath144 .
consequently , by lemma [ lemma2 ] , the kernel of the homomorphism @xmath120 is non - trivial .
but this is not possible , since we assume that @xmath11 acts faithfully on @xmath3 and @xmath150 .
let @xmath11 be a free group .
then there exists a descending series of finite index subgroups @xmath151 with trivial intersection @xmath152 .
every such a series defines an action of @xmath11 on the _ coset tree_. it is the rooted tree with the set of vertices equal to the union @xmath153 of the sets of cosets .
the set @xmath154 is the @xmath18th level of the tree and a vertex @xmath155 is connected by an edge with a vertex @xmath156 if and only if @xmath157 . the group @xmath11 acts then on the coset tree by the natural action on the cosets : @xmath158 the action is level transitive and the stabilizer of the path @xmath159 is equal to the intersection of the subgroups @xmath160 , i.e. , is trivial .
it is not hard to prove ( see , for instance , proposition 4.4 of @xcite ) that every level transitive action of @xmath11 on a rooted tree @xmath1 for which there exists a path @xmath4 with trivial stabilizer is conjugate to the action on a coset tree .
let @xmath1 be a rooted tree and @xmath161 be an infinite simple path starting in the root , where @xmath162 are the vertices that it goes through .
then the stabilizer of @xmath2 in the @xmath163 is isomorphic to the cartesian product of the automorphism groups of the sub - trees @xmath164 `` hanging '' from the path @xmath2 . for any free group @xmath11 consider a sequence of homomorphisms @xmath165 such that the intersection of kernels @xmath166 is trivial .
then we get a faithful action of @xmath11 on the tree @xmath1 for which @xmath2 is a fixed point .
if the intersection @xmath167 is trivial for every @xmath18 , then the image of @xmath11 in the group of germs @xmath168 is faithful .
a self - similar group @xmath0 of automorphisms of @xmath40 is called _ contracting _ if there exists a finite set @xmath175 such that for every @xmath171 there exists @xmath18 such that @xmath176 for all words @xmath13 of length more than @xmath18 .
the smallest set @xmath177 satisfying this condition is called the _ nucleus _ of the contracting group .
it is proved in @xcite proposition 2.13.8 that if a finitely generated group @xmath0 is contracting then the growth of the components of the schreier graph of the action of @xmath0 on @xmath44 is polynomial .
let @xmath178 be a subset such that such that @xmath179 , where @xmath177 is the nucleus of the group .
there is @xmath18 such that @xmath176 for all @xmath180 and all @xmath41 . in particular , there exist @xmath181 such that @xmath182 , where @xmath12 is the beginning of the length @xmath18 of the word @xmath2 .
it means that @xmath183 acts trivially on the subtree @xmath184 , i.e. , that the images of @xmath9 and @xmath111 in @xmath7 are equal .
consequently , @xmath7 has no finite subsets of size more than @xmath185 , i.e. , @xmath186 .
we have to eliminate the possibilities ( 2 ) and ( 3 ) of theorem [ th : mainlemma ] .
note that it is sufficient to prove this theorem for finitely generated contracting groups , since every finitely generated subgroup @xmath187 of a contracting group is a subgroup of a finitely generated contracting group .
one has to take the group generated by all elements of the form @xmath82 for @xmath188 and @xmath38 .
iterated monodromy groups ( see definitions in @xcite ) of expanding self - coverings of orbispaces , in particular , iterated monodromy groups of post - critically finite rational functions , are contracting hence have no free subgroups . in some cases ( like for the polynomial @xmath189 ,
@xcite ) the iterated monodromy groups have sub - exponential growth , which obviously implies that they have no free subgroups . in some other cases absence of free groups
was proved separately using some contraction arguments .
see for instance a proof of absence of free subgroups in the iterated monodromy group of @xmath190 in @xcite .
we say that an automorphisms @xmath9 of @xmath40 is _ finitary _ if there exists @xmath18 such that @xmath48 is trivial for all @xmath41 .
( recall that @xmath191 . ) the smallest @xmath18 with this property is called the _ depth _ of @xmath9 . [
def : bdd ] an automorphism @xmath9 of @xmath40 is _ bounded _ if there exists a finite set of sequences @xmath193 and a number @xmath18 such that if @xmath38 is not a beginning of any sequence @xmath194 then @xmath48 is finitary of depth at most @xmath18 . the number @xmath18 is called the _ finitary depth _ of @xmath9 and the set @xmath195 is called the set of _ directions _ of @xmath9 .
suppose that the group generated by bounded automorphisms contains a finitely generated free non - abelian group .
then by theorem [ th : mainlemma ] either there exists a free group @xmath11 generated by bounded automorphisms and having trivial stabilizer of a sequence @xmath53 , or there exists a free group @xmath11 fixing a point @xmath53 such that @xmath11 acts faithfully on every neighborhood @xmath202 of @xmath2 .
suppose that the first case holds .
let @xmath1 be the sub - tree of @xmath40 such that @xmath4 and the free group @xmath11 acts level transitively on @xmath1 .
( the tree @xmath1 is the union of the @xmath11-orbits of beginnings of @xmath2 . )
the boundary @xmath3 has to be infinite , otherwise the stabilizer of @xmath4 is of finite index . but for every @xmath203 the set of @xmath9-singular points of @xmath3 is finite , hence of zero measure .
consequently , by proposition [ pr : measureregular ] and [ pr : griamenable ] , the components of the schreier graph of the action of @xmath11 on @xmath3 are amenable , which contradicts with triviality of the stabilizer of @xmath2 .
suppose now that there exists @xmath2 such that a free 2-generated group @xmath204 fixes @xmath2 and acts faithfully on neighborhoods of @xmath2 .
let @xmath205 and @xmath206 be the sets of directions of @xmath9 and @xmath111 , respectively
. then @xmath207 , since otherwise there would exist a neighborhood @xmath202 of @xmath2 such that @xmath48 or @xmath208 is trivial .
there exists a beginning @xmath12 of @xmath2 such that all elements of @xmath205 and @xmath206 except for @xmath2 have beginning of length @xmath209 different from @xmath12
. then @xmath210 and @xmath211 are bounded automorphisms of @xmath212 with one direction @xmath213 , where @xmath214 .
they generate a free group , since @xmath11 acts faithfully on @xmath215 .
let @xmath216 .
let @xmath217 be a number greater than the finitary depths of @xmath218 and @xmath219 .
then @xmath220 and @xmath221 may change in a word @xmath222 , for @xmath223 , only the letters @xmath224 and can not change @xmath225 to @xmath226 . since the size of the alphabets @xmath227 are uniformly bounded , this implies that the automorphisms @xmath220 and @xmath221 have finite order , which is a contradiction .
it is proved in @xcite that groups generated by _ finite - state _
( see definition [ def : finitestate ] ) bounded automorphisms of a rooted tree @xmath40 are amenable , which implies that they have no free subgroups .
some examples of non - finite - state groups of bounded automorphisms of @xmath40 are studied in @xcite and @xcite .
the set of all finite state automorphisms of the tree @xmath40 is a countable group .
it is proved in @xcite that if @xmath9 is finite state then the growth function @xmath197 ( as given in definition [ def : growthfunction ] ) has either exponential or polynomial growth .
the set @xmath229 of all finite state automorphisms of @xmath40 for which @xmath197 is bounded by a polynomial of degree @xmath230 is a group .
a version of the growth function @xmath197 was defined and studied by s. sidki in @xcite .
later in @xcite s. sidki proved that the groups @xmath229 have no free subgroups ( actually , his theorem is more general , since it also covers some cases of infinite alphabet @xmath22 ) .
let us show how his theorem ( for the case of finite alphabet ) follows from theorem [ th : mainlemma ] .
the following inductive description of elements of @xmath229 is given in @xcite .
we denote by @xmath231 the group of finitary automorphisms of @xmath40 , i.e. , the automorphisms @xmath9 for which the sequence @xmath197 is eventually zero .
[ pr : polydescription ] let @xmath230 be a non - negative integer .
a finite - state automorphism @xmath9 of @xmath40 belongs to @xmath229 if and only if there exists a finite number of almost periodic words @xmath232 such that the restrictions @xmath233 do not depend on @xmath18 , and @xmath234 if @xmath13 is not a beginning of any @xmath235 .
it seems that the schreier graphs of subgroups of @xmath229 acting on @xmath44 have sub - exponential growth .
that would of course imply their amenability .
see , for instance the graphs considered in @xcite , which are schreier graphs of a sub - group of @xmath237 ( see example [ ex : tullio ] ) .
let us show that for every @xmath238 the set of @xmath9-singular points of @xmath44 is at most countable .
we argue by induction on @xmath230 .
if @xmath239 , i.e. , if @xmath9 is finitary , then the set of @xmath9-singular points is empty .
suppose that for every @xmath240 the set of @xmath9-singular points is at most countable .
take arbitrary @xmath238 . then by proposition
[ pr : polydescription ] , there is a finite set @xmath241 such that @xmath234 for every @xmath38 which is not a beginning of an element of @xmath195 .
then the set @xmath60 of @xmath9-singular points of @xmath44 is a subset of @xmath242 the sets @xmath243 are at most countable by the inductive hypothesis .
hence @xmath60 is at most countable .
let @xmath0 be a group generated by a finite set @xmath244 .
let @xmath1 be a @xmath0-invariant rooted subtree of @xmath40 on which @xmath0 acts level transitively .
then the measure space @xmath245 is either a finite set , or it is isomorphic to the standard lebesgue space . in the first case
the schreier graph of the action of @xmath0 on @xmath3 is finite . in the second case
the components of the schreier graph of the action of @xmath0 on @xmath78 are amenable by propositions [ pr : measureregular ] and [ pr : griamenable ] , since the sets of @xmath9-singular points of @xmath3 for @xmath246 are at most countable , hence have zero measure .
let us investigate now the groups of germs in @xmath229 .
assume that we have proved that @xmath248 has no free subgroups .
suppose that elements @xmath249 and @xmath250 of @xmath229 generate a free subgroup of the group of @xmath229-germs of @xmath53 .
it follows from proposition [ pr : polydescription ] that @xmath2 can be represented in the form @xmath251 for some finite words @xmath13 and @xmath12 so that @xmath252 , @xmath253 and @xmath254 and @xmath255 do not move @xmath12 .
moreover , @xmath256 for @xmath257 which are not beginnings of the word @xmath258 .
since @xmath249 and @xmath250 generate a free group of germs , their restrictions @xmath259 and @xmath260 onto @xmath261 generate a free group . since @xmath262 and @xmath263 , restrictions of the action of @xmath264 and @xmath265 onto @xmath266 generate a free group @xmath11 ( everything is periodic along the path @xmath267 ) .
let @xmath268 and let @xmath269 be the set of words of the form @xmath270 , where @xmath271 and @xmath272 , i.e. , the set of vertices adjacent to the path from the root to @xmath12 . denote by @xmath273 the intersection of the stabilizers in @xmath11 of the elements of @xmath274 .
see figure [ fig : u ] , where the elements of @xmath274 are circled . then @xmath273 is a subgroup of finite index in @xmath11 and we get a monomorphism @xmath275 mapping @xmath9 to @xmath276 .
the homomorphism @xmath120 takes values in @xmath277 , hence it has a non - trivial kernel on each coordinate .
then , by lemma [ lemma2 ] , the homomorphism @xmath120 has a non - trivial kernel , which is a contradiction .
[ ex : tullio ] consider the permutations @xmath249 and @xmath250 of the set of integers given by @xmath278 and @xmath279 for @xmath280 and @xmath281 .
let @xmath0 be the group generated by @xmath249 and @xmath250 .
it is easy to see ( using the binary numeration system on @xmath282 ) that @xmath0 is isomorphic to the group of automorphisms of the binary rooted tree @xmath283 given by the recurrent rules @xmath284 and @xmath285 a direct check ( for instance using proposition [ pr : polydescription ] ) shows that @xmath286 and @xmath287 , hence the group @xmath0 has no free subgroups by theorem of s. sidki .
the schreier graphs of the action of @xmath0 on @xmath44 coincide with the graphs considered by t. ceccherini - silberstein , f. fiorenzi , f. scarabotti @xcite and i. benjamini , c. hoffman @xcite and have intermediate growth .
laurent bartholdi , rostislav grigorchuk , and volodymyr nekrashevych , _ from fractal groups to fractal sets _ , fractals in graz 2001 .
dynamics geometry
stochastics ( peter grabner and wolfgang woess , eds . ) , birkhuser verlag , basel , boston , berlin , 2003 , pp
. 25118 . hyman bass , maria victoria otero - espinar , daniel rockmore , and charles tresser , _ cyclic renormalization and automorphism groups of rooted trees _ ,
lecture notes in mathematics , vol . 1621 , springer - verlag , berlin , 1996 .
kai - uwe bux and rodrigo prez , _ on the growth of iterated monodromy groups _ ,
topological and asymptotic aspects of group theory , contemp . math .
394 , amer .
soc . , providence , ri , 2006 , pp .
6176 .
t. ceccherini - silberstein , f. fiorenzi , and f. scarabotti , _ garden of eden theorems for cellular automata and for symbolic dynamical system _ , proceedings of the workshop `` random walks and geometry '' held at erwin schroedinger institute for mathematical physics , vienna , austria , 18.0613.07 2001 ( vadim a. kaimanovich , klaus schmidt , and wolfgang woess , eds . ) , de gruyter , berlin , 2004 , pp .
73108 .
vadim a. kaimanovich , _ equivalence relations with amenable leaves need not be amenable _ , topology , ergodic theory , real algebraic geometry .
rokhlin s memorial , amer .
( 2 ) , vol . 202 , 2001 ,
. 151166 .
|
we show that if a group @xmath0 acting faithfully on a rooted tree @xmath1 has a free subgroup , then either there exists a point @xmath2 of the boundary @xmath3 and a free subgroup of @xmath0 with trivial stabilizer of @xmath2 , or there exists @xmath4 and a free subgroup of @xmath0 fixing @xmath2 and acting faithfully on arbitrarily small neighborhoods of @xmath2 .
this can be used to prove absence of free subgroups for different known classes of groups .
for instance , we prove that iterated monodromy groups of expanding coverings have no free subgroups and give another proof of a theorem by s. sidki .
|
in this work , we introduce the technique of transformation inverse design , which combines the elegance of transformation optics @xcite ( to ) with the power of large - scale optimization ( inverse design ) , enabling automatic discovery of the best possible transformation for given design criteria and material constraints .
we illustrate our technique by designing multimode waveguide bends @xcite and mode squeezers @xcite , then measuring their performance with finite element method ( fem ) simulations .
most designs in transformation optics use either hand - chosen transformations @xcite ( which often require nearly unattainable anisotropic materials ) , or quasiconformal and conformal maps @xcite which can automatically generate nearly - isotropic transformations ( either by solving partial differential equations or by using grid generation techniques ) but still require _ a priori _ specification of the entire boundary shape of the transformation .
further , neither technique can directly incorporate refractive - index bounds . on the other hand ,
most inverse design in photonics involves repeatedly solving computationally expensive maxwell equations for different designs @xcite .
transformation inverse design combines elements of both transformation optics and inverse design while overcoming their limitations .
first , the use of optimization allows us to incorporate arbitrary fabrication constraints while at the same time searching the correct space of transformations without unnecessarily underconstraining or overconstraining the problem .
second , instead of solving maxwell s equations , we require only simple derivatives to be computed at each optimization step . this is because transformation optics works by using a coordinate transformation @xmath1 that warps light in a desired way ( e.g. mapping a straight waveguide to a bend , or mapping an object to a point or the ground for cloaking applications @xcite ) and then employing transformed materials which are given in terms of the jacobian @xmath2 to mathematically mimic the effect of the coordinate transformation .
this transforms all solutions of maxwell s equations in the same way ( as opposed to non - to multimode devices which often have limited bandwidth and/or do not preserve relative phase between modes @xcite ) , and is therefore particularly attractive for designing multimode optical devices @xcite ( such as mode squeezers , expanders , splitters , couplers , and multimode bends ) with no intermodal scattering .
( before to was first used for cloaking , electrostatic cloaking using anisotropic conductivities @xcite had been proposed ) .
examples of such transformations are shown in fig .
[ fig : three - possible - applications ] . for a more comprehensive treatment of the physical and mathematical ideas of transformation media used in this paper , see @xcite . ]
one major difficulty with transformation optics is that most functions @xmath1 yield highly anisotropic and magnetic materials . in principle
, these transformed designs can be fabricated with anisotropic microstructures @xcite or naturally birefringent materials @xcite . however , in the infrared regime ( where metals are lossy ) it is far easier to instead fabricate effectively isotropic dielectric materials , provided that the refractive index falls within the given bounds @xmath3 and @xmath4 of the fabrication process ( for example , subwavelength nanostructures @xcite or waveguides with variable thickness @xcite ) .
this requirement means that we would prefer to consider the subset of transformations that can be mapped to approximately isotropic dielectric materials .
the theory of transformation optics with nearly isotropic materials is intimately connected to the subjects of conformal maps ( which are isotropic by definition @xcite ) and _ quasiconformal maps _ [ which in mathematical analysis are defined as _ any _ orientation - preserving transformation with bounded anisotropy ( as quantified in sec .
[ sub : nearly - isotropic - transformations ] ) ] .
however , in transformation optics the term `` quasiconformal '' has become confusingly associated with only a single choice of quasiconformal map suggested by li and pendry @xcite . in that work , li and pendry
proposed minimizing a mean anisotropy with `` slipping '' boundary conditions ( defined in sec .
[ sub : nearly - isotropic - transformations ] ) , which turns out to yield a transformation that is essentially conformal up to a _ constant stretching _ ( and thus anisotropy ) everywhere .
this map , which also happens to minimize the peak anisotropy given the slipping boundary conditions @xcite , is sometimes confusingly called `` the quasiconformal map '' @xcite . however
, we point out in sec .
[ sub : nearly - isotropic - transformations ] that slipping boundary conditions are not the correct choice if one wishes to ensure a reflectionless interface between transformed and untransformed regions .
instead , for interfaces to be reflectionless requires at least continuity of the transformation @xmath5 at the interface @xcite and , as we show in sec .
[ sub : conformal - maps ] for the case of isotropic dielectric media , continuity of the jacobian @xmath6 as well .
if one fixes the transformation on part or all of the boundary ( instead of just the corners ) and minimizes the peak anisotropy , the result is called ( in analysis ) an _ extremal _ quasiconformal map @xcite .
we point out in sec .
[ sub : isotropic - transformations ] that this extremal quasiconformal map can never be conformal except in trivial cases .
additionally , previous work in quasiconformal transformation optics underconstrained the space of transformations in one way but overconstrained it in another .
li and pendry s method , along with other work on extremal quasiconformal maps in mathematical analysis , assumed that the entire boundary _ shape _ of the transformed domain is specified _ a priori _ ( even if the _ value _ of the transformation at the boundary is not specified ) .
in contrast , transformation inverse design allows parts of the boundary shape to be freely chosen by the optimization , only fixing aspects of the boundary that are determined by the underlying problem ( e.g. the input / output facets of the boundary in fig .
[ fig : three - possible - applications ] ) as explained in secs .
[ sub : generalized - bend - transformation ] , allowing a much larger space of transformations to be searched .
also , for such stricter boundary conditions , minimizing the mean anisotropy is _ not _ equivalent to minimizing the peak anisotropy @xcite , and we argue below that the peak anisotropy is a better figure of merit for transformation optics in general .
we solve all of these problems by using large - scale numerical optimization to find the transformation with minimal peak anisotropy that exactly obeys continuity conditions at the boundary with untransformed regions .
this allows the input / output interfaces to transition smoothly and continuously into untransformed devices while also satisfying fabrication constraints ( e.g. bounds on the attainable refractive indices and bend radiii ) .
a large space of arbitrary smoothly varying transformations ( that satisfy the continuity conditions and fabrication constraints ) is explored quickly and efficiently by parametrizing in a `` spectral '' basis @xcite of fourier harmonics and chebyshev polynomials . the optimized transformation is then scalarized ( as in the case of previous work on quasiconformal transformation optics ) into an isotropic dielectric material that guides modes with minimal intermodal scattering and loss . in the case of a multimode bend ,
for which our design was recently fabricated and characterized @xcite , we achieve intermodal scattering at least an order of magnitude smaller than a conventional non - to bend . in sec .
[ sec : transformation - optics ] , we review the equations of transformation optics . in sec . [ sub : isotropic - transformations ] , we describe situations where the transformation - designed material can be mapped to isotropic media . in sec .
[ sub : conformal - maps ] , we point out that such isotropic transformations , due to their analyticity , always have undesirable interface discontinuities when coupled into untransformed regions . in secs . [ sub : nearly - isotropic - transformations ] and [ sub : scalarization - errors - for ] , we review the techniques of quasiconformal mapping ( as used in both the transformation optics and mathematical analysis literature ) and scalarization of nearly isotropic transformations . we show that the inherent restrictions of quasiconformal mapping can be circumvented by directly optimizing the map using transformation inverse design . in secs .
[ sub : simple - circular - bend ] and [ sub : generalized - bend - transformation ] , we design a nearly isotropic transformation for a @xmath7-bend by perturbing from the highly anisotropic circular bend transformation . in secs .
[ sub : setup - of - optimization ] and [ sub : spectral - parameterization ] , we set up the bend optimization problem and the spectral parameterization . in sec .
[ sub : results ] , we present the optimized structure , which reduces anisotropy by several orders of magnitude compared to the circular to bend . in sec . [ sub : peak - minimized - structure ] , we present finite element simulation results comparing our optimized design to the conventional non - to bend and the circular to bend . in secs .
[ sub : minmax ] we show that minimizing the mean anisotropy can lead to pockets of high anisotropy ( which in turn leads to greater intermodal scattering ) while minimizing the peak does not . in secs .
[ sub : tradeoff - between - anisotropy ] , we discuss the tradeoff between the bend radius and the optimized anisotropy . in sec .
[ sec : mode - squeezer ] we briefly present methods and results for applying transformation inverse design to optimize mode squeezers .
the frequency domain maxwell equations ( fields @xmath8 ) , without sources or currents , in linear isotropic dielectric media [ @xmath9 are @xmath10 consider a coordinate transformation @xmath11 with jacobian @xmath12 .
we define the _ primed _ gradient vector as @xmath13 and the primed fields as @xmath14 and @xmath15 .
one can then rewrite eq .
( [ eq : maxwell ] ) , after some rearrangement @xcite , as @xmath16 where the effects of the coordinate transformation have been mapped to the equivalent tensor materials @xmath17 this equivalence has become known as _ transformation optics _ ( to ) .
it is actually the specific case of a much more general result from general relativity @xcite . for a further discussion of space
time transformations and connections to negative refraction , see @xcite and @xcite .
most useful applications of to require that the transformation be coupled to untransformed regions ( e.g. the input and output straight waveguides in the case of a bend transformation , or the surrounding air region for the case of a ground - plane cloaking transformation ) .
however , in order for to to guarantee that the interface between transformed and untransformed regions be reflectionless , the transformation must be equivalent to a _ continuous _ transformation of all space that is the identity @xmath18 in the `` untransformed '' regions , as depicted in fig .
[ fig : jboundary ] .
more generally , the untransformed regions can be simple rotations or translations , but when examining a particular interface , we can always choose the coordinates to be @xmath19 at that interface .
it is clear by construction that continuous @xmath5 is sufficient for reflectionless interfaces @xcite , and this is in fact a necessary condition as well @xcite .
although a general anisotropic transformation need only have @xmath1 continuous at the interface , we show below that an _ isotropic _ transformation will also have a continuous @xmath20 at the interface .
these boundary conditions are essential for designing useful transformations without interface reflections .
continuous in order for there not to be any interface reflections.[fig : jboundary ] ] for the vast majority of transformations , the materials in eq .
( [ eq : transformed_materials ] ) are anisotropic tensors . however , for certain transformations , the tensors are effectively scalar .
suppose that the transformation @xmath1 is 2d ( @xmath21 and @xmath22 ) , making @xmath20 block - diagonal ( with the @xmath23 element independent of the @xmath24 block ) .
then , the @xmath24 block of @xmath25 is isotropic if and only if the diagonal elements are equal and the off - diagonal elements vanish : @xmath26 in this case , the @xmath20 part of eq .
( [ eq : transformed_materials ] ) becomes @xmath27 this isotropy has different implications for transverse - magnetic ( tm ) polarized modes in 2d ( which have @xmath28 and @xmath29 ) versus transverse - electric ( te ) polarized modes ( which have @xmath30 and @xmath31 ) . for tm - polarized modes , the fields @xmath32 in the primed coordinate system are also tm - polarized and eq .
( [ eq : primed_maxwell ] ) becomes @xmath33 hence , for tm modes , an isotropic transformation can be exactly mapped to an isotropic dielectric material .
similarly , for te - polarized modes the equivalent material is isotropic and _
magnetic_. however , if @xmath34 varies slowly compared to the wavelengths of the fields , then the transformation can _ approximately _ still be mapped to an isotropic dielectric material by making an eikonal approximation ( as in @xcite ch .
8.10 ) and commuting @xmath35 with one of the curls in the maxwell equations . in particular , eq . ( [ eq : primed_maxwell ] ) can be written : @xmath36 where the last term can be neglected for slowly varying transformations . because the tm case is conceptually simpler and
does not require this extra approximation , we work with it exclusively for the rest of this paper . also , because the non - trivial aspects of the transformation occur in the @xmath24 plane , we hereafter use @xmath6 to denote the @xmath24 block .
if a transformation has an isotropic * @xmath6 * , then the transformation preserves angles in the @xmath24 plane .
additionally , if @xmath37 , then the transformation also preserves handness and orientation . the combination of these two properties is called a conformal map @xcite , and is the only case where the situation in sec .
[ sub : isotropic - transformations ] can be realized .
we only consider transformations with @xmath37 in order to restrict ourselves to dielectric materials .
also , a @xmath37 transformation coupled continuously to an untransformed ( @xmath38 ) region would require singularities ( @xmath39 ) at some points .
conformal maps are described by analytic functions , which are of the form @xmath40 ( where @xmath41 is the untransformed complex coordinate ) and whose real and imaginary parts satisfy the cauchy riemann equations of complex analysis @xcite . however
, true conformal maps can not directly be used for transformation optics in typical applications , because of the impossibility of coupling them to untransformed regions with the boundary conditions discussed in sec . [
sec : transformation - optics ] .
in particular , the uniqueness theorem of analytic functions ( * ? ? ? * thm . 10.39 )
tells us that if @xmath42 in some region , then @xmath42 everywhere ( similarly for a simple rotation or translation in some regions ) . as a corollary , in the limit where a transformation becomes more and more isotropic in the neighborhood of an interface
, it must have a continuous @xmath6 , not just a continuous @xmath43 .
it is easy to see this explicitly in the example of fig . [
fig : jboundary ] : continuity of @xmath43 at the interface requires that @xmath44 on both sides of the interface , which determines the first row of @xmath6 .
the isotropy of @xmath25 then forces @xmath45 .
therefore , in the sections that follow ( where we search for _ approximately _ isotropic maps ) , we will impose the condition of continuous @xmath6 as a boundary condition on our transformations .
the resulting transformations are nearly isotropic in the interior and exactly isotropic on the interfaces .
this condition , discussed at the end of sec .
[ sub : scalarization - errors - for ] , also has the useful consequence of producing a continuous refractive index @xmath46 .
because true conformal maps can not be used , one widely used alternative is to search for a nearly _ _ isotropic transformation , which can be _ approximated _ by an isotropic material at the cost of some scattering corrections to the exactly transformed modes of the nearly isotropic material . to do this
, one must first quantify the measure of anisotropy that is to be minimized .
the isotropy condition of eq .
( [ eq : isotropic equations ] ) is equivalent to @xmath47 , where @xmath48 are the two eigenvalues of @xmath25 .
while @xmath49 works as a measure of anisotropy , it is convenient for optimization purposes to define _ differentiable _ measures that can be expressed directly in terms of the trace and determinant of @xmath6 , and precisely such quantities have been developed in the literature on quasiconformal maps @xcite .
a general transformation is an arbitrary function of @xmath50 and @xmath51 or , equivalently , an arbitrary function @xmath52 , of @xmath53 and @xmath54 , which may not be analytic in @xmath53 .
the anisotropy can be related to the _ beltrami coefficient _
@xcite @xmath55 the term _ quasiconformal map _ refers to _ any _ map that has bounded @xmath56 , which includes all non - singular sense preserving ( @xmath37 ) transformations .
it can be shown that the _ linear distortion _
@xmath57 @xcite satisfies @xmath58 various other measures of anisotropy have been defined in the grid generation literature @xcite , including the _ winslow _ and _ modified liao _ functionals , which are given by @xmath59 and @xmath60 , respectively . however , for the rest of this work we refer to the quantity @xmath61 as the `` anisotropy '' , where @xmath62 is the _ distortion function _ @xcite , defined as @xmath63 the tensor @xmath64 is known as the _ distortion tensor _ @xcite .
as mentioned in the introduction , an _ extremal _
quasiconformal map is one that minimizes the peak anisotropy , given the shape of the transformed region and the values of the transformation on some or all of the boundary @xcite . because @xmath62 , @xmath65 , @xmath57 , and @xmath66 are all monotonic functions of one another , they are equivalent for the purpose of finding an extremal quasiconformal map . however , @xmath62 is numerically convenient because it is a differentiable function of the entries of @xmath6 .
these quantities are _ not _ generally equivalent for minimizing the _ mean _ anisotropy @xcite , and we argue in sec .
[ sub : minmax ] that the peak anisotropy is a better figure of merit .
however , in the special case where the value of the transformation is only fixed at the _ corners _ of the domain and is allowed to vary freely in between ( a `` slipping'' boundary condition ) , li and pendry showed that it is equivalent to minimize the mean ( either winslow or modified - liao ) or the peak anisotropy , and that these yield a constant - anisotropy map ( a uniform scaling of a conformal map ) @xcite that they and other authors have used for transformation optics @xcite .
however , the slipping boundary will generally lead to reflections at the interface between the transformed and untransformed regions because of the resulting discontinuity in the transformation , which can only be reduced by making the transformation domain very large in cases ( e.g. cloaking ) with localized deformations . in order to design compact transformation - optics devices , especially for applications such as bends where the deformation is nonlocalized
, we will instead impose continuity of the transformation and/or its jacobian on the input / output facets of the domain , while at the same time allowing the shape of some or all of the boundary to vary ( unlike all previous work on quasiconformal maps , to our knowledge ) .
the minimum - anisotropy quasiconformal map is then _ scalarized _ ( as in @xcite ) by approximating it with an isotropic dielectric material . as shown in sec .
[ sub : isotropic - transformations ] , a perfectly isotropic 2d transformation of a geometry with an isotropic dielectric material that guides tm modes @xmath67 can be mapped to a transformed material and geometry that is also isotropic dielectric and guides tm modes @xmath68 .
this is exact for @xmath69 , but for a _ nearly _ isotropic transformation with @xmath70 , the equivalent permeability is @xmath71 , where the anisotropic part @xmath72 is proportional to @xmath65 to lowest order . while @xmath73 can not be fabricated using dielectric gradient index processes , one can neglect this small correction so that the actual fabricated material has permeability @xmath74 . in practice ,
we absorb any @xmath75 into @xmath76 by multiplying @xmath76 by the average eigenvalue of @xmath77 @xmath78 but this does not change the @xmath79 error . a born approximation @xcite tells us that , given an exact transformation with no scattering , any small error of @xmath80 and @xmath72 will generically lead to scattered fields with magnitudes of @xmath81 and scattered power of @xmath82 .
the modes of the approximate scalarized material @xmath83 are then the exact guided modes plus scattered power corrections of @xmath84 $ ] .
a similar analysis explains why we must explicitly impose continuity of @xmath6 at the input / output facets of the domain . as explained in sec .
[ sub : conformal - maps ] , a purely isotropic transformation in the neighborhood of the interface , along with a continuity of @xmath5 , would automatically yield continuous @xmath6 , so one might hope that minimizing anisotropy would suffice to obtain a nearly continuous @xmath6 .
unfortunately , as we show in the appendix , the resulting discontinuity in @xmath85 ( and hence the discontinuity in the refractive index ) is of order @xmath86 , which would lead to @xmath87 power loss due to reflections , much larger than the @xmath88 $ ] power scattering from anisotropy in the interior .
this would make it pointless to minimize the anisotropy in the interior , since the boundary reflections would dominate .
in fact , our initial implementation of the bend optimization in sec .
[ sub : spectral - parameterization ] did not enforce continuity of @xmath6 , and we obtained a large @xmath89 index discontinuity at the endfacets for @xmath90 . therefore , in sec .
[ sub : spectral - parameterization ] we impose continuity of @xmath6 explicitly . in this paper
, we directly minimize @xmath62 using large - scale numerical optimization while keeping track of constraints on the transformation @xmath5 and its jacobian @xmath6 , as well as the engineering fabrication bounds @xmath3 and @xmath4 . by using numerical optimization
, we can in principle achieve both a lower mean anisotropy and a lower peak anisotropy than by traditional quasiconformal mapping , since the optimization is also free to vary the boundary shape ( with at most the input / output interfaces fixed , although in some cases their locations and shapes are allowed to vary as well ) .
the minimization problem can be written , for example , as @xmath91 where @xmath92 is a _
functional _ norm taken over the domain of @xmath1 .
we consider two possible norms : the @xmath93 norm ( the mean @xmath94 ) , and the @xmath95 norm ( @xmath96 ) .
we show in sec .
[ sub : minmax ] that minimizing the mean can lead to pockets of high anisotropy which can cause increased scattering .
directly optimizing the peak anisotropy on the other hand , avoids such pockets while simultaneously keeping the mean nearly as low .
the continuity of @xmath5 and @xmath6 at the input / output interfaces , as well as other constraints on the interface locations , are imposed implicitly by the parametrization of @xmath1 ( as explained in sec .
[ sub : spectral - parameterization ] ) .
in this section , we design a bend transformation ( depicted in fig . [ fig : schematic ] ) using general methods to ( locally ) solve the optimization problem of eq .
( [ eq : general_min ] ) .
in contrast , previous work on to bend design either utilized materials that were either anisotropic or consisted of multiple stacked isotropic layers @xcite or employed slipping boundary conditions @xcite ( which result in endfacet reflections when coupled to untransformed waveguide ) .
first , we consider a simple circular bend transformation ( which we refer to hereafter as the circular to bend ) that maps a rectangular segment of length @xmath97 and width unity ( in arbitrary distance units to be determined later ) into a bend with inner radius @xmath98 and outer radius @xmath99 ( as shown in fig . [ fig : schematic ] ) . for convenience ,
we choose the untransformed coordinates to be @xmath100 and @xmath101 , with the untransformed segment length @xmath102 equal to the inner arclength of the bend .
the transformation @xmath1 can be written as @xmath103 where @xmath104 and @xmath105 . while @xmath5 is continuous at the input / output interfaces @xmath106 ,
one issue is that @xmath6 is not continuous there , which can be seen from @xmath107 .
another issue is that @xmath108 is highly anisotropic .
the anisotropy for this transformation is @xmath109 , which has a peak value @xmath110 for @xmath111 at the outer radius @xmath112 .
note that one can instead choose @xmath113 , which gives the _ conformal _ bend @xmath114 $ ] . as explained in sec .
[ sub : conformal - maps ] , this map has zero anisotropy , but neither @xmath5 nor @xmath20 are continuous at the input / output interfaces , leading to large reflections there . in order to address the problems of the circular to bend , we look for minimum anisotropy _ and _ continuous - interface transformations of the form of eq .
( [ eq : polar ] ) , where the intermediate polar coordinates are now arbitrary functions @xmath115 and @xmath116 .
the ratio @xmath117 is now an optimization parameter .
the jacobian then satisfies @xmath118 we find that the optimization always seems to prefer a symmetric bend ( and if the optimum is unique , it must be symmetric ) , so we impose a mirror symmetry in order to halve our search space : @xmath119 we also require interface continuity of @xmath5 and @xmath6 ( as discussed in sec .
[ sub : conformal - maps ] ) , which give the conditions at @xmath106 : @xmath120 besides minimizing the objective function @xmath62 , the optimization must keep track of several constraints . first , any fabrication method will bound the overall refractive index @xmath121 to lie between some values @xmath3 and @xmath4 .
we choose units so that the width of the transformed region is unity ( @xmath122 ) , and consider transforming a straight waveguide of width @xmath123 .
@xmath124 should be small enough so that the exponential tails of the waveguide modes are negligible outside the transformed region . in the straight waveguide segment to be transformed ( as well as the straight waveguides to be coupled into the input and output interfaces of the bend )
, @xmath125 is high in the core @xmath126 and low in the cladding @xmath127 . for convenience ,
we write this refractive index as a product @xmath128 of an overall refractive index @xmath129 and a normalized profile @xmath130 that is unity in the cladding and some value greater than unity in the core ( determined by the ratio of the high and low index regions of the straight waveguide ) .
the transformed refractive index is given by @xmath131 where the average eigenvalue @xmath132 of the magnetic permeability eq .
( [ eq : average_eigenvalue ] ) has been absorbed into the dielectric index .
the overall refractive - index scaling @xmath129 is then allowed to freely vary as a parameter in the optimization .
second , like the circular to bend , the optimum to bend is expected to have a tradeoff between the bend radius and anisotropy .
because of this expected tradeoff , we can choose to either minimize @xmath98 while keeping @xmath62 fixed , or minimize @xmath62 while keeping @xmath98 fixed .
we focus on the latter choice , since the bend radius is the more intuitive target quantity to know beforehand .
also , we find empirically that optimizing @xmath62 converges much faster than optimizing @xmath98 while yielding the same local minima . with these constraints
, there are several ways to set up the optimization problem , depending on which norm we are minimizing .
one method is to minimize the peak anisotropy @xmath96 with @xmath133 for some grid @xmath134 of some points to be defined in sec .
[ sub : spectral - parameterization ] .
however , the peak ( the @xmath95 norm ) is not a differentiable function of the design parameters , so it should not be directly used as the objective function .
instead , we perform a standard transformation @xcite : we introduce a dummy variable @xmath135 and indirectly minimize the peak @xmath62 using a differentiable inequality constraint between @xmath135 and @xmath136 at all @xmath133 : @xmath137 for comparison , we explain in sec . [ sub : minmax ] why the @xmath95 norm is better to minimize than the @xmath93 norm ( the mean anisotropy ) . the minimization of the @xmath93 norm , @xmath138 [ which _ is _ differentiable in terms of the parameters @xmath139 , @xmath140 , @xmath129 , and @xmath97 ] is implemented as @xmath142 we use the circular bend @xmath143 as a starting guess , and search the space of general transformations @xmath144 by perturbing from this base case .
( we only perform _ local _ optimization ; not global optimization , but comment in sec .
[ sub : tradeoff - between - anisotropy ] on a simple technique to avoid being trapped in poor local minima . ) as explained in sec .
[ sub : spectral - parameterization ] , the perturbations will be parameterized such that the symmetry and continuity constraints are satisfied automatically . figure [ fig : schematic ] shows a schematic of the bend transformation optimization process .
first , the straight region is mapped to a circular bend .
then , the intermediate polar coordinates @xmath145 and @xmath146 for every point @xmath147 are perturbed , using an optimization algorithm described at the end of sec .
[ sub : spectral - parameterization ] , and the desired norm ( either @xmath93 and @xmath95 ) of the anisotropy is computed .
this process is repeated at each optimization step until the structure converges to a local minimum in @xmath148 . ] to faciliate efficient computation of the objective and constraints , the functions @xmath145 and @xmath146 can be written as the circular bend transformation plus perturbations parametrized in the spectral basis @xcite : @xmath149 where the coordinate @xmath150 has been centered appropriately for the domain @xmath151 $ ] of degree-@xmath152 chebyshev polynomials @xmath153 . the sines and cosines
have been chosen to satisfy the mirror - symmetry conditions of eq .
( [ eq : symmetry ] ) .
the sine series also automatically satisfies the second continuity condition of eq .
( [ eq : continuity ] ) . in order to satisfy the rest of the conditions ,
the following constraints are also imposed : @xmath154 these equations are solved to simply eliminate the @xmath155 coefficients before optimization .
this spectral parametrization has several advantages over finite - element discretizations such as the piecewise - linear parameterization of @xcite .
first , the spectral basis converges exponentially for smooth functions @xcite .
we found that only a small number ( @xmath156 ) of spectral coefficients @xmath157 are needed to achieve very low - anisotropy ( @xmath158 ) transformations .
second , if the fabrication process favors slowly varying transformations ( or if these are needed to make the eikonal approximation for the te polarization , as in sec .
[ sub : isotropic - transformations ] ) , this constraint may be imposed simply by using smaller @xmath159 and @xmath160 . with this spectral parameterization , the formulation of the optimization problem eq .
( [ eq : min_t method ] ) becomes @xmath161 the local optimization was performed using the derivative - free cobyla non - linear optimization algorithm @xcite in the nlopt package @xcite . in principle
, we can make the optimization faster by analytically computing the derivatives of the objective and constraints with respect to the design parameters and using a gradient - based optimization algorithm , but that is not necessary because both @xmath162 and @xmath85 , which determine all the non - trivial objective and constraint functions in this optimization problem , are so computationally inexpensive to evaluate that the convergence rate is not a practical concern .
a @xmath163 design is shown in fig . [ fig : mreduced ] , along with the scalarized circular to bend for comparison .
the bend radius was @xmath164 and the number of spectral coefficients was @xmath165 .
the objective and constraints were evaluated on a @xmath166 grid @xmath134 in @xmath147 ( chebyshev points in the @xmath50 direction and a uniform grid in the @xmath51 direction ) .
this design had @xmath167 and mean @xmath168 . in comparison ,
the circular to bend of the same radius has @xmath169 and @xmath170 . , while dramatically improving the scattered - power matrix .
[ fig : mreduced ] ] the @xmath164 optimized design structure was compared in finite - element maxwell simulations ( using the fenics code @xcite ) , with the conventional non - to bend [ simply bending the waveguide profile around a circular arc with @xmath171 and the scalarized circular to bend .
the four lowest - frequency modes of a multimode straight waveguide were injected at the input interface @xmath172 , and the scattered - power matrix @xmath173 was computed using the measured fields at the output interface @xmath174 .
the scattered - power matrix is defined as @xmath175 where @xmath176 is the propagation direction of the guided modes , @xmath177 is the normalized electric field of the @xmath178th exactly guided mode of the non - scalarized material @xmath179 , and @xmath180 is the actual magnetic field of the approximate scalarized material at the interface after injecting a normalized mode @xmath181 at the input interface .
this makes @xmath182 equal to the power scattered into the @xmath178th output mode from the @xmath183th input mode . for a straight waveguide , which has no intermodal scattering , @xmath184 .
figure [ fig : mreduced ] shows a dramatically improved @xmath173 for the scalarized and optimized to bend compared to the scalarized circular to bend .
[ the rows and columns of @xmath173 for the circular bend add up to less than one because some power has either been scattered out of the waveguide entirely , or some power has been scattered into fifth or higher - order modes .
the rows and columns of @xmath173 for the optimized bend add up to nearly 1 , with the small deficiency due to the @xmath79 out - of - bend and higher - order intermodal scattering as well as mesh - descretization error . ]
the electric - field profiles for the fundamental mode , displayed in fig .
[ fig : finite - element - simulations ] , show a dramatic difference in the performance of the optimized structure versus the other structures .
both the conventional and circular to bend show heavy intermodal scattering in the bend region , while the optimized transformation displays very little scattering . ]
we found a clear difference between minimizing the peak anisotropy versus minimizing the mean .
the results of an optimization run with @xmath185 , @xmath186 , and @xmath187 are shown in fig .
[ fig : mean_max ] .
both structures had very low mean anisotropy @xmath188 .
the mean - minimized structure , at @xmath189 , had a slightly lower mean than the peak - minimized structure which had @xmath190 .
however , in terms of the peak anisotropy , the peak - minimized structure is the clear winner by a factor of @xmath191 , with @xmath192 as opposed to @xmath193 for the mean - optimized structure .
both structures were scalarized and tested in finite - element maxwell simulations of the four lowest - frequency modes of the straight waveguide .
the scattered - power matrix shows that the difference in @xmath96 resulted in an order of magnitude reduction in the intermodal scattering ( as shown in the off - diagonal elements ) and noticeably improved transmission , ( especially in the element @xmath194 for the fourth mode ) . , @xmath186 , and @xmath187 .
[ fig : mean_max ] ] in optimized structures , we found that @xmath195 for the optimized bend , similar to the circular to bend , decreases monotonically with @xmath98 ( as shown in fig .
[ fig : tradeoff ] ) . unlike the circular bend
, however , this tradeoff seems asymptotically _ exponential _ rather than @xmath196 . in particular , there are two clearly different regimes for this tradeoff : a power law @xmath197 at small @xmath198 and an exponential decay @xmath199 , at larger @xmath98 .
the second regime was only attained after using successive optimization , because with only one independent optimization run the algorithm tended to get stuck in local minima . for successive optimization
, the optimum structure is used as a starting guess for the next run , and the initial step size is set large enough so that the algorithm can reach better local minima than the previous one .
, @xmath200 results in a power law decaying tradeoff @xmath201 at low @xmath98 and an exponentially decaying tradeoff at higher @xmath98 . for comparison , the unoptimized anisotropy for the circular to bend is shown above .
[ fig : tradeoff ] ] for @xmath198 , we found that there are multiple local minima and that independent optimizations for different @xmath98 tend to be trapped in suboptimal local minima , as shown by the open dots in fig .
[ fig : tradeoff ] . to avoid this problem
, we used a `` successive optimization '' technique in which the optimal structure for smaller @xmath98 is rescaled as the starting guess for local optima at a larger @xmath98 , in order to stay along the exponential - tradeoff curve .
( another possible heuristic is `` successive refinement '' @xcite , in which optima for smaller @xmath202 are used as starting points for optimizing using larger @xmath202 . )
we also applied transformation inverse design to another interesting geometry : a mode squeezer that concentrates modes and their power in a small region in space , again with minimal intermodal scattering ( quite unlike a conventional lens , which is intrinsically angle / mode - dependent ) , similar to the problem considered in @xcite ( which did not construct isotropic designs ) .
we choose the untransformed region to be @xmath203 and @xmath204 .
the goal of this transformation @xmath1 is to focus the beam by minimizing the _ mid_-_beam width _ @xmath205 as in sec .
[ sub : spectral - parameterization ] , the transformation is written as a perturbation from the identity transformation ( which was used as the starting guess ) and parameterized in the spectral basis @xmath206 the sine series automatically satisfies mirror symmetry about @xmath172 and continuity of @xmath5 at the input / output interfaces @xmath207 .
however , we found that constraining the coefficients @xmath208 to enforce continuity of @xmath6 ( as in sec .
[ sub : spectral - parameterization ] ) was not necessary ( although it might give a better result ) since the optimization algorithm only squeezed the center region while leaving the interfaces and the regions around them relatively untouched . in this problem
, we could either minimize @xmath62 for a fixed @xmath209 or minimize @xmath209 for a fixed @xmath62 , and we happened to choose the latter . ] finite - element maxwell simulations , shown in fig . [
fig : squeezer ] , demonstrate that the optimized design is greatly superior to a simple gaussian taper transformation designed by hand .
the gaussian transformation was given by @xmath210 $ ] , where @xmath211 and @xmath212 .
superficially , the design seems similar to an `` adiabatic '' taper between a wide low - index waveguide and a narrow high - index waveguide , and it is known that any sufficiently gradual taper of this form would have low scattering due to the adiabatic theorem@xcite .
however , the optimized to design is much too short to be in this adiabatic regime .
if it were in the adiabatic regime , then taking the same design and simply stretching the index profile to be more gradual ( a taper twice as long ) would reduce the scattering , but in fig .
[ fig : squeezer ] we perform precisely this experiment and find that the stretched design _ increases _ the scattering .
the analytical simplicity of to design no maxwell equations need be solved in order to warp light in a prescribed way paradoxically makes the application of computational techniques more attractive in order to discover the best transformation by rapidly searching a large space of possibilities .
previous work on to design used optimization to some extent , but overconstrained the transformation by fixing the boundary shape while underconstraining the boundary conditions required for reflectionless interfaces .
in fact , even our present work imposes more constraints than are strictly necessary as long as we require continuous @xmath6 at the input / output interfaces , there is no conceptual reason why those interfaces need be flat . a better bend , for example , might be designed by constraining the location of only two corners ( to fix the bend radius ) and constraining only @xmath6 on other parts of the endfacets .
however , we already achieve an exponential tradeoff between radius and anisotropy , so we suspect that further relaxing the constraints would only gain a small constant factor rather than yielding an asymptotically faster tradeoff . in the case of the mode squeezer , one could certainly achieve better results by imposing the proper @xmath6 constraints at the endfacets
. it would also be interesting to apply similar techniques to ground - plane cloaking .
all to techniques suffer from some limitations that should be kept in mind .
first , to seems poorly suited for optical devices in which one _ wants _ to discriminate between modes ( e.g. a modal filter ) or to scatter light between modes ( e.g. a mode transformer ) .
to is ideal for devices in which it is desirable that all modes be transported equally , with no scattering .
even for the latter case ( such as our multimode bend ) , however , to designs almost certainly trade off computational convenience for optimality , because they impose a stronger constraint than is strictly required : to is restricted to designs where the solutions at _ all points _ in the design are coordinate transformations of the original system , whereas most devices are only concerned with the solutions at the endfacets . for example , it is conceivable that a more compact multimode bend could be designed by allowing intermodal scattering _ within _ the bend as long as the modes scatter back to their original configurations by the endfacet ; the interior of the bend might not even be a waveguide , and instead might be a resonant cavity of some sort@xcite .
however , optimizing over such structures seems to require solving maxwell s equations in some form at each optimization step , which is far more computationally expensive than the to design and , unlike the to design , must be repeated for different wavelengths and waveguide designs .
in this appendix , we briefly derive the fact , mentioned in sec .
[ sub : scalarization - errors - for ] , that the endfacet discontinuity scales much worse with anisotropy than the scalarization errors in the transformation interior , which leads us to impose an explicit continuity constraint on the jacobian @xmath6 .
in particular , we examine the jacobian @xmath6 for nearly isotropic transformations ( @xmath213 ) that also have @xmath19 explicitly constrained at the interfaces .
( the following analysis can also be straightfowardly extended to situations where @xmath5 is a simple rotation of @xmath147 on the interface , or where the interface has an arbitrary shape . ) in this case , the jacobian is @xmath214 where @xmath215 and @xmath216 are small quantities ( @xmath217 ) if @xmath25 is nearly isotropic . the anisotropy eq .
( [ eq : kminusone ] ) is then : @xmath218 the determinant then satisfies @xmath219 this square - root dependence is also reflected in the refractive index @xmath46 and leads to @xmath87 power loss due to interface reflections that overwhelm the @xmath88 $ ] corrections to scattered power due to the scalarization of nearly isotropic transformations ( as explained in sec .
[ sub : scalarization - errors - for ] ) .
hence , it becomes necessary to explicitly constrain @xmath45 _ in addition _ to @xmath19 .
this work was supported in part by the afosr muri for complex and robust on - chip nanophotonics ( dr . gernot pomrenke ) , grant number fa9550 - 09 - 1 - 0704 .
|
we present a new technique for the design of transformation - optics devices based on large - scale optimization to achieve the optimal effective isotropic dielectric materials within prescribed index bounds , which is computationally cheap because transformation optics circumvents the need to solve maxwell s equations at each step .
we apply this technique to the design of multimode waveguide bends ( realized experimentally in a previous paper ) and mode squeezers , in which all modes are transported equally without scattering .
in addition to the optimization , a key point is the identification of the correct boundary conditions to ensure reflectionless coupling to untransformed regions while allowing maximum flexibility in the optimization .
many previous authors in transformation optics used a certain kind of quasiconformal map which overconstrained the problem by requiring that the entire boundary shape be specified _ a priori _ while at the same time underconstraining the problem by employing `` slipping '' boundary conditions that permit unwanted interface reflections .
100 a. j. ward and j. b. pendry .
`` refraction and geometry in maxwell s equations , '' j. mod
. opt .
* 43*(4):773793 ( 1996 ) .
u. leonhardt .
`` optical conformal mapping , '' science * 312*:17771780 ( 2006 ) .
j. b. pendry , d. schurig , and d. r. smith .
`` controlling electromagnetic fields , '' science * 312*:17801782 ( 2006 ) .
h. chen , c. t. chan , and p. sheng .
`` transformation optics and metamaterials , '' nat . mater . * 9*:387396 ( 2010 ) .
y. liu and x. zhang .
`` recent advances in transformation optics , '' nanoscale * 4*(17):52775292 ( 2012 ) .
n. i. landy and w. j. padilla .
`` guiding light with conformal transformations , '' opt .
express * 17*(17):1487214879 ( 2009 ) .
m. heiblum and j. h. harris .
`` analysis of curved optical waveguides by conformal transformation , '' j. quantum electron . * 11*(2):7583 ( 1975 ) .
y.g .
ma , n. wang , and c. k. ong .
`` application of inverse , strict conformal transformation to design waveguide devices , '' j. opt .
soc .
am . a * 27*:968972 ( 2010 ) .
s. han , y. xiong , d. genov , z. liu , g. bartel , and x. zhang .
`` ray optics at a deep - subwavelength scale : a transformation optics approach , '' nano lett . * 8*(12):42434247 ( 2008 ) .
k. yao and x. jiang .
`` designing feasible optical devices via conformal mapping , '' j. opt .
soc . am .
b * 28*(5):10371042 ( 2011 ) .
t. han , c. qiu , j. dong , x. tang , and s. zouhdi .
`` homogeneous and isotropic bends to tunnel waves through multiple different / equal waveguides along arbitrary directions , '' opt .
express * 19*(14):1302013030 ( 2011 ) .
d. a. roberts , m. rahm , j. b. pendry , and d. r. smith .
`` transformation - optical design of sharp waveguide bends and corners , '' appl .
. lett . * 93*(251111 ) ( 2008 ) .
j. huangfu , s. xi , f. kong , j. zhang , h. chen , d. wang , b. wu , l. ran , and j. a. kong .
`` application of coordinate transformation in bent waveguides , '' j. appl .
phys . * 104*(014502 ) ( 2008 ) .
h. xu , b. zhang , y. yu , g. barbastathis , and h. sun .
`` dielectric waveguide bending adapter with ideal transmission , '' opt .
express * 29*(6):12871290 ( 2012 ) .
z. l. mei and t. j. cui .
`` experimental realization of a broadband bend structure using gradient index metamaterials , '' opt .
express * 17*(20):1835418363 ( 2009 ) .
z. l. mei and t. j. cui .
`` arbitrary bending of electromagnetic waves using isotropic materials , '' j. appl .
phys .
* 105*(104913 ) ( 2009 ) .
m. rahm , d. a. roberts , j. b. pendry , and d. r. smith .
`` transformation - optical design of adaptive beam bends and beam expanders , '' opt .
express * 16*(15):1155511567 ( 2008 ) .
b. vasi , g. isi , r. gaji , and k. hingerl .
`` coordinate transformation based design of confined metamaterial structures , '' phys . rev .
b * 79*(085103 ) ( 2009 ) . c. garca - meca , m. m. tung , j. v. galn , r. ortuo , f. j. rodrguez - fortuo , j. mart , and a. martnez . `` squeezing and expanding light without reflections via transformation optics , '' opt .
express * 19*(4):35623757 ( 2011 ) .
x. zhang , h. chen , x. luo , and h. ma .
`` transformation media that turn a narrow slit into a large window , '' opt . express * 16*(16):1176411768 ( 2008 ) .
o. ozgun and m. kuzuoglu .
`` utilization of anisotropic metamaterial layers in waveguide miniaturization and transitions , '' ieee microw .
wirel .
. lett .
* 17*(754 ) ( 2007 ) .
b. zhang , y. luo , x. liu , and g. barbastathis .
`` macroscopic invisibility cloak for visible light , '' phys .
rev .
lett .
* 106*(033901 ) ( 2011 ) .
z. liang , x. jiang , f. miao , s. guenneau , and j. li .
`` transformation media with variable optical axes , '' new j. phys .
* 14*(103042 ) ( 2012 ) .
z. l. mei , y. s. liu , f. yang , and t. j. cui .
`` a dc carpet cloak based on resistor networks , '' opt .
express * 20*(23):2575825764 ( 2012 ) .
s. wang and s. liu .
`` controlling electromagnetic scattering of a cavity by transformation media , '' opt .
express * 20*(6):67776785 ( 2012 ) .
d. schurig , j. j. mock , b. j. justice , s. a. cummer , j. b. pendry , a. f. starr , and d. r. smith .
`` metamaterial electromagnetic cloak at microwave frequencies , '' science * 314*(5801):977980 ( 2006 ) .
x. chen , y. luo , j. zhang , k. jiang , j. b. pendry , and s. zhang .
`` macroscopic invisibility cloaking of visible light , '' nat . commun .
* 2*(176 ) ( 2011 ) .
j. mei , q. wu , and k. zhang .
`` multimultifunctional complementary cloak with homogeneous anisotropic material parameters , '' j. opt . soc .
am . a * 29*(10):20672073 ( 2012 )
. a. v. novitsky .
`` inverse problem in transformation optics , '' j. opt .
* 13*(035104 ) ( 2011 ) .
j. li and j. b. pendry .
`` hiding under the carpet : a new strategy for cloaking , '' phys .
rev .
lett *
101*(203901 ) ( 2008 ) .
q. wu , j. p. turpin , and d. h. werner .
`` integrated photonic systems based on transformation optics enabled gradient index devices , '' light : science and applications * 1*(e38 ) ( 2012 ) . c. garcia - meca , a. martinez , and u. leonhardt
`` engineering antenna radiation patterns via quasi - conformal mappings , '' opt .
express * 19*(24):2374323750 ( 2011 ) .
h. f. ma and t. j. cui .
`` three - dimensional broadband and broad - angle transformation - optics lens , '' nat .
commun .
* 1*(124 ) ( 2010 ) .
n. i. landy , n. kundtz , and d. r. smith .
`` designing three - dimensional transformation optical media using quasiconformal coordinate transformations , '' phys .
rev .
lett .
* 105*(193902 ) ( 2010 ) .
z. l. mei , j. bai , and t. j. cui .
`` illusion devices with quasi - conformal mapping , '' j. electromagn . waves
app . *
24*(17):25612573 ( 2010 ) .
n. kundtz and d. r. smith .
`` extreme - angle broadband metamaterial lens , '' nat .
mater . * 9*:129132 ( 2010 ) .
r. liu , r. c. ji , j. j. mock , j. y. chin , t. j. cui , and d. r. smith .
`` broadband ground - plane cloak , '' science * 323*(5912):366369 ( 2009 ) .
j. valentine , j. li , t. zentgraf , g. bartal , and x. zhang .
`` an optical cloak made of dielectrics , '' nat .
mater . * 8*:568571 ( 2009 ) .
l. gabrielli , j. cardenas , c. b. poitras , and m. lipson .
`` silicon nanostructure cloak operating at optical frequencies , '' nat .
photonics * 3*:461463 ( 2009 ) .
t. ergin , n. stenger , p. brenner , j. b. pendry , and m. wegener .
`` three - dimensional invisibility cloak at optical wavelengths , '' science * 328*(5976):337339 ( 2010 ) .
m. yin , x. y. tian , h. x. han , and d. c. li .
`` free - space carpet - cloak based on gradient index photonic crystals in metamaterial regime , '' appl .
phys .
lett .
* 100*(124101 ) ( 2012 ) .
z. chang , x. zhou , j. hu , and g. hu .
`` design method for quasi - isotropic transformation materials based on inverse laplace s equation with sliding boundaries , '' opt . express * 18*(6):60896096 ( 2010 ) .
p. markov , j. g. valentine , and s. m. weiss .
`` fiber - to - chip coupler designed using an optical transformation , '' opt .
express * 20*(13):1470514712 ( 2012 ) .
l. h. gabrielli and m. lipson .
`` transformation optics on a silicon platform , '' j. opt .
* 13*(024010 ) ( 2011 ) .
z. l. mei , j. bai , and t. j. cui .
`` experimental verification of a broadband planar focusing antenna based on transformation optics , '' new j. phys .
* 13*(063028 ) ( 2011 ) .
l. tang , j. yin , g. yuan , j. du , h. gao , x. dong , y. lu , and c. du .
`` general conformal transformation method based on schwarz christoffel approach , '' opt . express * 19*(16):1511915126 ( 2011 ) .
j. p. turpin , a. t. massoud , z. h. jiang , p. l. werner , and d. h. wener .
`` conformal mappings to achieve simple material parameters for transformation optics devices , '' opt .
express * 18*(1):244252 ( 2010 ) .
d. r. smith , y. urzhumov , n. b. kundtz , and n. i. landy .
`` enhancing imaging systems using transformation optics , '' opt . express * 18*(20):2123821251 ( 2010 ) .
w. r. frei , h. t. johnson , and k. d. choquette .
`` optimization of a single defect photonic crystal laser cavity , '' j. appl . phys .
* 103*(033102 ) ( 2008 ) . c. y. kao and f. santosa .
`` maximization of the quality factor of an optical resonator , '' wave motion * 45*(4):412427 ( 2008 ) .
d. c. dobson and f. santosa .
`` optimal localization of eigenfunctions in an inhomogeneous medium , '' siam j. appl .
math * 64*(3):762774 ( 2004 ) .
j. vuckovic , m. loncar , h. mabuchi , and a. scherer . `` design of photonic crystal microcavities for cavity qed '' phys
. rev .
e * 65*(016608 ) ( 2002 ) .
j. lu and j. vuckovic .
`` inverse design of nanophotonic structures using complementary convex optimization , '' opt . express * 18*(4):37933804 ( 2010 ) .
j. s. jensen and o. sigmund .
`` topology optimization for nano - photonics , '' laser photon . rev . * 5*(2):308321 ( 2011 ) .
l. frandsen , a. harpth , p. borel , m. kristensen , j. jensen , and o. sigmund .
`` broadband photonic crystal waveguide 60 bend obtained utilizing topology optimization , '' opt .
express * 12*(24):59165921 ( 2004 ) .
w. r. frei , h. t. johnson , and d. a. tortorelli .
`` optimization of photonic nanostructures , '' comput . method .
appl .
m. * 197*(41):34103416 ( 2008 ) .
j. s. jensen and o. sigmund .
`` systematic design of photonic crystal structures using topology optimization : low - loss waveguide bends , '' appl .
phys .
lett . * 84*(12):20222024 ( 2004 ) .
j. andkjaer and o. sigmund .
`` topology optimized low - contrast all - dielectric optical cloak , '' appl .
. lett . * 98*(021112 ) ( 2011 ) .
p. borel , a. harpth , l. frandsen , m. kristensen , p. shi , j. jensen , and o. sigmund .
`` topology optimization and fabrication of photonic crystal structures , '' opt .
express * 12*(9):1996 ( 2004 ) .
s. j. cox and d. c. dobson .
`` maximizing band gaps in two - dimensional photonic crystals , '' siam j. appl . math * 59*(6):21082120 ( 1999 ) . j. s. jensen and o. sigmund .
`` topology optimization of photonic crystal structures : a high - bandwidth low - loss t - junction waveguide , '' j. opt .
soc . am .
b * 22*(6):11911198 ( 2005 ) .
j. riishede and o. sigmund .
`` inverse design of dispersion compensating optical fiber using topology optimization , '' j. opt .
soc .
am .
b * 25*(1):8897 ( 2008 ) . c. y. yao , s. osher , and e. yablonovitch . `` maximizing band gaps in two - dimensional photonic crystals by using level set methods , '' appl .
phys .
b * 81*(2):235244 ( 2005 ) .
w. r. frei , d. a. tortorelli , and h. t. johnson .
`` topology optimization of a photonic crystal waveguide termination to maximize directional emission , '' appl .
. lett .
* 86*(111114 ) ( 2005 ) .
y. tsuji and k. hirayama .
`` design of optical circuit devices using topology optimization method with function - expansionbased refractive index distribution , '' ieee phot .
tech .
lett . * 20*(12):982984 ( 2008 ) . c. y. kao and s. osher .
`` incorporating topological derivatives into shape derivatives based level set methods , '' j. comp .
phys . * 225*(1):891909 ( 2007 ) .
p. seliger , m. mahvash , c. wang , and a. f. j. levi .
`` optimization of aperiodic dielectric structures , '' j. appl .
phys .
* 100*(034310 ) ( 2006 ) .
y. watanabe , n. ikeda , y. sugimoto , y. takata , y. kitagawa , a. mizutani , n. ozaki , and k. asakawa .
`` topology optimization of waveguide bends with wide , flat bandwidth in air bridge - type photonic crystal slabs , '' j. appl
. phys .
* 101*(113108 ) ( 2007 ) .
j. h. lee , j. blair , v. a. tamma , q. wu , s. j. rhee , c. j. summers , and w. park .
`` direct visualization of optical frequency invisibility cloak based on silicon nanorod array , '' opt .
express * 17*(15):1292212928 ( 2009 ) .
u. leonhardt and t. tyc .
`` broadband invisibility by non - euclidian cloaking , '' science * 323*(5910):110112 ( 2009 ) .
a. greenleaf , m. lassas , and g. uhlmann .
`` anisotropic conductivities that can not be detected by eit , '' physiol
. meas . * 24*(413):413419 ( 2003 ) .
a. greenleaf , m. lassas , and g. uhlmann .
`` on nonuniqueness for calder ' on s inverse problem , '' math .
res .
letters * 10*(5):685693 ( 2003 ) .
u. leonhardt and t. philbin .
( dover , 2010 ) .
v. liu and s. fan .
`` compact bends for multi - mode photonic crystal waveguides with high transmission and suppressed modal crosstalk , '' opt .
express * 21*(7):80698075 ( 2013 ) .
a. mekis , j. c. chen , i. kurland , s. fan , p. r. villeneuve , and j. d. joannopoulos .
`` high transmission through sharp bends in photonic crystal waveguides , '' phys .
. lett . * 77*:37873790 ( 1996 ) .
c. ma , q. zhang , and e. v. keuren .
`` right - angle slot waveguide bends with high bending efficiency , '' opt .
express * 16*(19):14330 ( 2008 ) .
v. liu and s. fan .
`` compact bends for multi - mode photonic crystal waveguides with high transmission and suppressed modal crosstalk , '' opt . express * 21*(7):80698075 ( 2013 ) .
a. chutinan , m. okano , and s. noda .
`` wider bandwidth with high transmission through waveguide bends in two - dimensional photonic crystal slabs , '' appl .
. lett . * 80*(10):16981699 ( 2002 ) .
a. chutinan and s. noda .
`` highly confined waveguides and waveguide bends in three - dimensional photonic crystal , '' appl . phys
. lett . * 75*(24):37393741 ( 1999 ) .
b. chen , t. tang , and h. chen .
`` study on a compact flexible photonic crystal waveguide and its bends , '' opt .
express * 17*(7):50335038 ( 2009 ) .
y. zhang and b. li .
`` photonic crystal - based bending waveguides for optical interconnections , '' opt .
express * 14*(12):57235732 ( 2006 ) .
j. smajic , c. hafner , and d. erni .
`` design and optimization of an achromatic photonic crystal bend , '' opt .
express * 11*(12):13781384 ( 2003 ) .
m. schmiele , v. s. varma , c. rockstuhl , and f. lederer .
`` designing optical elements from isotropic materials by using transformation optics , '' phys .
rev .
a * 81*(033837 ) ( 2010 ) .
d. schurig , j. b. pendry , and d. r. smith .
`` transformation - designed optical elements , '' opt .
express * 15*(22):1477214782 ( 2007 ) .
m. rahm , s. a. cummer , d. schurig , j. b. pendry , and d. r. smith . `` optical design of reflectionless complex media by finite embedded coordinate transformations , '' phys . rev .
lett *
100*(063903 ) ( 2008 ) .
d. smith , j. mock , a. starr , and d. schurig .
`` gradient index metamaterials , '' phys . rev .
e * 71*(036609 ) ( 2005 ) .
d. r. smith , j. b. pendry , and m. c. k. wiltshire .
`` metamaterials and negative refractive index , '' science * 305*(5685):788792 ( 2004 ) .
f. xu , r. c. tyan , p. c. sun , y. fainman , c. c. cheng , and a. scherer .
`` fabrication , modeling , and characterization of form - birefringent nanostructures , '' opt . lett . *
20*(24):24572459 ( 1995 ) . h. kurt and d. s. citrin .
`` graded index photonic crystals , '' opt . express * 15*(3):12401253 ( 2007 ) .
b. vasi , r. gaji , and k. hingerl .
`` graded photonic crystals for implementation of gradient refractive index media , '' j. nanophotonics * 5*(051806 ) ( 2011 ) . u. levy , m. abashin , k. ikeda , a. krishnamoorthy , j. cunningham , and y. fainman . `` inhomogenous dielectric metamaterials with space - variant polarizability , '' phys .
rev .
lett .
* 98*(243901 ) ( 2007 .
l. gabrielli and m. lipson .
`` integrated luneburg lens via ultra - strong index gradient on silicon , '' opt .
express * 19*(21):2012220127 ( 2011 ) .
y. wang , c. sheng , h. liu , y.j .
zheng , c. zhu , s. m. wang , and s. n. zhu . `` transformation bending device emulated by graded - index waveguide , '' opt .
express * 20*(12):1300613013 ( 2012 ) .
r. ulrich and r. j. martin .
`` geometric optics in thin film light guides , '' appl . opt . * 10*(9):20772085 ( 1971 ) .
f. zernike .
`` luneburg lens for optical waveguide use , '' opt .
commun . * 12*(4):379381 ( 1974 ) .
b. u. chen , e. marom , and a. lee .
`` geodesic lenses in single - mode linbo@xmath0 waveguides , '' appl . phys
. lett . * 31*(4):263 ( 1977 ) .
j. brazas , g. kohnke , and j. mcmullen .
`` mode - index waveguide lens with novel gradient boundaries developed for application to optical recording , '' appl . opt . * 31*(18):34203428 ( 1992 ) .
w. rudin . .
( mcgraw hill , 1986 ) .
g. e. shilov . .
( mit , 1973 ) .
j. li and j. b. pendry . private communication ( 2013 ) .
w. yan , m. yan , and m. qiu .
`` necessary and sufficient conditions for reflectionless transformation media in an isotropic and homogenous background , '' arxiv:0806.3231 ( 2008 ) .
l. bergamin .
`` electromagnetic fields and boundary conditions at the interface of generalized transformation media , '' phys .
rev .
a * 80*(063835 ) ( 2009 ) .
w. yan , m. yan , z. ruan , and m. qiu .
`` coordinate transformations make perfect invisibility cloaks with arbitrary shape , '' new j. phys .
* 10*(043040 ) ( 2008 ) .
o. weber , a. myles , and d. zorin .
`` computing extremal quasiconformal maps , '' symp . geom . process .
* 31*(5):16791689 ( 2012 ) .
k. astala , t. iwaniec , and g. martin . .
( princeton university , 2008 ) .
a. papadopoulos , editor .
, volume 1 .
( european mathematical society , 2007 ) .
r. khnau , editor . , volume 2 .
( elsevier b.v . , 2005 ) .
l. bers .
`` an extremal problem for quasi - conformal mappings and a problem of thurston , '' acta math . 7398 ( 1978 ) .
l. v. ahlfors . .
( american mathematical society , 1966 ) .
z. balogh , k. fssler , and i. platis .
`` modulus of curve families and extremality of spiral - stretch maps , '' j. anal . math . * 113*(1):265291 ( 2011 ) .
k. astala , t. iwaniec , and g. martin .
`` deformations of annuli with smallest mean distortion , '' arch . rational mech .
anal . *
195*:899921 ( 2010 ) .
z. balogh , k. fssler , and i. platis .
`` modulus method and radial stretch map in the heisenberg group , '' ann .
acad .
sci .
fenn . * 38*(1):149180 ( 2013 ) .
j. p. boyd . .
( dover , 2001 ) .
s. boyd and l. vandenberghe . .
( cambridge university , 2004 ) .
l. gabrielli , d. liu , s. g. johnson , and m. lipson .
`` on - chip transformation optics for multimode waveguide bends , '' nat .
commun . * 3*(1217 ) ( 2012 ) .
u. leonhardt and t. g. philbin .
`` transformation optics and the geometry of light , '' prog . opt . * 53*:69152 ( 2009 ) .
j. d. jackson . .
( wiley , 1998 ) . j. plebanski .
`` electromagnetic waves in gravitational fields , '' phys . rev . *
118*(5):13961408 ( 1960 ) .
u. leonhardt and t. g. philbin .
`` general relativity in electrical engineering , '' new j. phys .
* 8*(10 ) ( 2006 ) .
l. ahlfors .
on quasiconformal mappings
. j. anal .
math . * 3*(1):158 ( 1953 ) .
w. zeng , f. luo , s. t. yau , and x. d. gu .
`` surface quasi - conformal mapping by solving beltrami equations , '' proc .
math . surfaces xiii 391408 ( 2009 ) .
j. f. thompson , b. k. soni , and n. p. weatherill . .
( crc , 1999 ) .
p.
knupp and s. steinberg . .
( crc , 1994 ) .
s. g. johnson , m. l. povinelli , m. soljai , a. karalis , s. jacobs , and j. d. joannopoulos .
`` roughness losses and volume - current methods in photonic - crystal waveguides , '' appl . phys .
b * 81*(283293 ) ( 2005 ) . a. w. snyder and j. d. love . .
( chapman and hall , 1983 ) .
w. c. chew . .
( ieee , 1995 ) .
m. j. d. powell .
`` a direct search optimization method that models the objective and constraint functions by linear interpolation , '' adv .
optim .
numer .
anal .
( 1994 )
. m. j. d. powell .
`` direct search algorithms for optimization calculations , '' acta numer . * 7*:287336 ( 1998 ) .
s. g. johnson .
the nlopt nonlinear - optimization package ( http://ab-initio.mit.edu/nlopt ) ( 2007 ) .
a. logg , k. a. mardal , and g. n. wells . .
( springer , 2012 ) .
a. oskooi , a. mutapcic , s. noda , j. d. joannopoulos , s. p. boyd , and s. g. johnson .
`` robust optimization of adiabatic tapers for coupling to slow - light photonic - crystal waveguides , '' opt .
express * 20*(19):2155821575 ( 2012 ) .
a. mutapcica , s. boyd , a. farjadpour , s. g. johnson , and y. avniel .
robust design of slow - light tapers in periodic waveguides .
. optim . * 41*(4):365384 ( 2009 ) .
t. f. chan , j. cong , t. kong , and j. r. shinnerl .
`` multilevel optimization for large - scale circuit placement , '' ieee icad 171176 ( 2000 ) .
k. w. chun and j. ra .
`` fast block - matching algorithm by successive refinement of matching criterion , '' proc .
spie , vis .
commun . image process .
, * 1818*:552560 ( 1992 ) .
s. g. johnson , p. bienstman , m. a. skorobogatiy , m. ibanescu , e. lidorikis , and j. d. joannopoulos .
`` adiabatic theorem and continuous coupled - mode theory for efficient taper transitions in photonic crystals , '' phys .
rev .
e * 66*(066608 ) ( 2002 ) . c. manolatou , s. g. johnson , s. fan , p. r. villeneuve , h. a. haus , and j. d. joannopoulos .
`` high - density integrated optics , '' j. lightwave technol . * 17*(9):16821692 ( 1999 ) .
|
the bilayer fluid membranes , which can form spontaneously when molecules with hydrophobic and hydrophilic parts are introduced to water , are a popular topic of research . in theoretical studies of such membranes , it is usual to work on length scales much larger than the molecular size , so that they can be considered as continuous surfaces ; two - dimensional curved spaces embedded in euclidean three - space .
features commonly incorporated into mathematical models of membranes include bending rigidity @xmath0 ( disaffinity for extrinsic curvature ) , gaussian curvature modulus @xmath1 ( disaffinity for intrinsic curvature ) , spontaneous curvature due to membrane asymmetries , and for closed surfaces , constraints of constant volume , constant area and constant difference in area between the inner and outer layers . in recent years
, interest has grown in membranes whose molecules have orientational order within the surface .
for instance , a smectic - a liquid crystal membrane , whose molecules have aliphatic tails which point in an average direction normal to the local tangent plane of the membrane , can undergo a continuous phase transition to smectic - c phase , in which the tails tilt .
the tilt is described by a two - component vector order parameter within the surface , given by the local thermal average of vectors parallel to the tails , projected onto to local tangent plane .
if such a membrane has intrinsic ( gaussian ) curvature , its orientational order is frustrated , since a parallel vector field can not exist in a curved space .
ordering is further frustrated on a closed surface of spherical topology ( genus zero ) , as an intrinsic vector field must have topological defects whose indices sum to the euler number of the surface ; in this case , two .
_ there must be at least two vortices in the order parameter field on a sphere .
the order parameter field vanishes at these defects to avoid infinite gradients .
the defects are energetically costly but topologically unavoidable .
the energetics of a smectic - c , genus zero membrane were studied by mackintosh and lubensky in @xcite .
there are other types of orientational order .
nematic order is similar to vector order except that its order parameter is locally invariant under rotations through @xmath2 .
hence nematics can form defects of index @xmath3 .
hexatic order is locally invariant under rotations through @xmath4 , as this is the molecular ` bond ' angle .
the notion may be generalized to fluid phases with @xmath5-atic orientational order being locally invariant under rotations through @xmath6 and forming defects of index @xmath7 .
if a vector ( ` monatic ' ) is represented by an arrow then we may picture an @xmath5-atic as an @xmath5-headed arrow or an @xmath5-spoked wheel . for a theoretical study of @xmath5-atic fluid membranes of genus zero ,
see @xcite . in this paper ,
theoretical findings on @xmath5-atic fluid membranes of toroidal topology are presented .
as the torus has genus one and euler number zero , it is possible to arrange an intrinsic vector field on a torus without topological defects .
_ furry tori can be brushed smooth . )
therefore it appears plausible that @xmath5-atic membranes with large order parameter coupling compared with bending rigidity may energetically favour a toroidal topology to a spherical one .
the object of embarking on this research was to confirm or refute this hypothesis .
as will be demonstrated , it turns out to be true , although the arrangements of ordering in the toroidal surface are far more diverse than expected . throughout this paper ,
a number of approximations are used .
one approximation is the use of mean field theory .
that is , no thermal fluctuations are considered .
so the configuration adopted by the system is assumed to be that which minimizes its free energy .
also , the lowest landau level approximation is employed in constructing the order parameter field , as discussed in section [ landau ] .
the validity of these approximations is investigated in section [ validity ] .
both of these approximations give rise to apparent phase transitions between separate , clearly defined states of the system , some of which may in fact be blurred by thermal fluctuations and higher landau levels , to the point where they disappear .
it should therefore be understood that , where reference is made to phase transitions , it is in the context of mean field theory and the lowest landau level approximation . before studying the thermal physics
, it will be necessary to introduce some tensor calculus .
this is needed to describe the behaviour of an @xmath5-atic field living within a two - dimensional curved space , as well as the intrinsic and extrinsic elastic properties of the space ( _ ie . _
the membrane ) itself . in the absence of intrinsic
ordering ( _ ie .
_ under the influence of the membrane s elasticity alone ) , one might expect the equilibrium shape of a toroidal vesicle to be axisymmetric .
however , such a vesicle is only neutrally stable with respect to one particular mode of non - axisymmetric deformation .
this goldstone mode is discussed in section [ goldstone ] , in which a further approximation is introduced , limiting the range of vesicle shapes to be explored . in the sections that follow ,
the ground - state configurations and energies of the order parameter field are calculated for each of these shapes , subject to the lowest landau level approximation .
although some of these shapes are never energetically favoured , the behaviour of the field living on them is of interest , if only academic .
finally , in section [ pd ] , the total free energy ( consisting of order - parameter field and membrane elasticity parts ) is minimized with respect to vesicle shape , to find the equilibrium configuration of the system as a function of temperature and the other thermodynamic parameters .
the result is a set of phase diagrams for various values of @xmath5 , which contain phases of vesicles with different shapes and ordering .
the shapes should be experimentally observable , although the intrinsic ordering is most likely not .
@xmath5 is the order of rotational symmetry .
as stated earlier , the fluid membrane is locally invariant under rotations through @xmath6 , and the @xmath5-atic field may be represented by a set of @xmath5-spoked wheels tessellating .
clearly @xmath5 is integer .
furthermore , it must be one of the set of numbers for which regular @xmath5-gons tessellate .
that is , @xmath8 .
these are the values of @xmath5 which park et .
al . studied in @xcite .
however , consider a triangular ( @xmath9 ) lattice with long - range _ orientational _ , but no _ positional _ order ; the @xmath9 analogue of a hexatic lattice .
bonds between neighbouring molecules lie at angles which are integer multiples of @xmath10 .
its symmetry properties in the absence of positional order are therefore indistinguishable from its dual lattice , the hexatic , and it will form defects of index @xmath11 , in preference to the more energetically expensive index @xmath12 defects . to convince oneself of this , it might be helpful to attempt to draw a lattice which has @xmath13 rotational symmetry , but does not have @xmath4 rotational symmetry .
this is only possible if one keeps strict account of the positions of the lattice sites and polygons ( _ ie .
_ in the absence of dislocations ) .
the following research is therefore conducted for @xmath14 .
the @xmath15 curved surface is embedded in euclidean @xmath16-space .
if @xmath17 is a two - dimensional coordinate of a point @xmath18 on the surface then @xmath19 is the position vector of @xmath18 in the embedding space relative to some origin .
coordinate - dependent unit basis vectors @xmath20 , where @xmath21 takes the values @xmath22 and @xmath23 to label the coordinates , are defined in the directions of @xmath24 .
a coordinate - invariant metric tensor is defined on the riemannian @xmath23-space , with components , from which the area of the closed surface is calculated : @xmath25 where @xmath26 .
indices are raised and lowered by the metric and its inverse , @xmath27 , in the standard manner .
the extrinsic curvature tensor has components @xmath28 , where @xmath29 is a unit vector normal to the surface .
dot products are evaluated in @xmath30 . )
its trace , @xmath31 , is the ` total curvature ' .
readers may be familiar with the _ mean _ curvature which is half of this quantity .
the intrinsic or gaussian curvature is given by @xmath32 .
as an example , the above quantities are calculated for a sphere of radius @xmath33 , using spherical polar coordinates @xmath34 .
the metric tensor may be represented by a matrix : @xmath35 with determinant @xmath36 .
hence the area element for integration is @xmath37 , which is just the jacobian for spherical polars .
any continuous , doubly differentiable surface has two principal radii of curvature , @xmath38 and @xmath39 at each point .
these are the radii of curvature of the geodesics passing through that point which have respectively the greatest and least curvature as measured in the euclidean embedding space .
it can be shown @xcite that these two geodesics are always perpendicular .
if the particular coordinate basis used corresponds to these two perpendicular directions , then the curvature tensor will be diagonal , and given by @xmath40 of course , on a sphere , curvature is constant .
therefore @xmath41 where the sign convention denotes the direction of the normal .
however , it should be noted that , without one index raised , the components are not constant : @xmath42 clearly , on a sphere , mean curvature is @xmath43 and gaussian curvature is @xmath44 .
recall that the intrinsic @xmath5-atic order in the system is orientational , not positional , and that it is locally invariant under rotations through @xmath6 . as
this orientational order is intrinsic to a 2-dimensional space , in can be represented by a complex order parameter in the following way .
let @xmath45 be the local angle between @xmath46 and the orientation of ordering . for vector order , this orientation is easily defined .
for instance , in a smectic - c membrane , it is the direction of a molecule s aliphatic tail projected onto the local tangent plane of the membrane .
however , in a nematic liquid crystal , @xmath45 may take one of two values for each molecule , depending on which end of the molecule one chooses to measure from . for our purposes
, it does not matter which of these two values is assigned to @xmath45 at each point . for bond - angle
orientational order , _ eg . _
hexatic , there is an @xmath5-fold ambiguity in the definition of @xmath45 but again , it does not matter which of the @xmath5 orientations is chosen .
it is conventional in this case to define @xmath45 at a given molecule as the angle between the basis vector @xmath46 and the line joining that molecule to its nearest neighbour .
having defined @xmath45 , the order parameter is given by @xmath47 where @xmath48 denotes a thermal average .
note that @xmath49 has the appropriate rotational invariance , since a rotation of the orientation of ordering through an angle @xmath50 within the surface is represented by a rotation of @xmath49 through @xmath51 in the argand plane . hence rotating the orientation of ordering through @xmath6 corresponds an identity transformation in the complex plane .
the model used is identical to that in @xcite .
there is zero spontaneous curvature as the bilayers considered are symmetric , and no constraint on bilayer area difference .
the vesicle s volume is also unconstrained since small solvent molecules may diffuse freely in and out , so long as the solvent is very pure ; containing no free ions , which would lead to osmotic pressure .
the membrane s area @xmath52 is fixed by the number of constituent molecules .
the free energy functional @xmath53/t = \int d^2 \mbox{\boldmath $ \sigma$}\sqrt{g}\ , \left ( \,r|\psi |^2 + \mbox{$\frac{1}{2}$}u|\psi |^4 + cd^{a}\psi d^{*}_{a}\psi ^ * + \mbox{$\frac{1}{2}$}\kappa ( k^{a}_{a})^2 + \kappa_{g } k \,\right)\ ] ] closely resembles the ginzburg - landau free energy for classical superconductors . for a discussion of how the thermodynamic coefficients relate to real physical and chemical quantities ,
see for instance @xcite . as defined in eq.([psi ] ) , the phase of @xmath49 depends on the choice of basis vectors . hence the derivatives @xmath54 in eq.([f ] ) must be covariant in order for the quantity @xmath55 to be physical . let the basis vectors be rotated through an angle @xmath56 . consequently , @xmath57 where @xmath58 are the components of the rotation matrix .
given that @xmath59 , demanding that @xmath55 is invariant under the gauge transformation implies that the components of @xmath60 transform like @xmath61 so @xmath60 undergoes the usual transformation , @xmath62 , of a vector potential in a gauge - invariant derivative .
but , in addition there is a coordinate rotation , @xmath63 since , in this complex number representation , a rotation of the argand plane is always accompanied by a rotation of the basis vectors .
the ` spin connection ' , @xmath64 has the appropriate transformation . on the sphere , @xmath65 and @xmath66 .
above the continuous transition to the disordered phase , @xmath49 vanishes and only the last two terms in the free energy eq.([f ] ) survive .
these constitute the helfrich hamiltonian , describing the elastic energy of a constant area membrane .
the _ intrinsic _ curvature energy , being proportional to the integral of gaussian curvature , is a _ topological _ invariant , according to the gauss - bonnet formula @xmath67 where @xmath50 is the euler number of the closed surface @xcite . for an orientable surface , @xmath68 , where @xmath69 is the surface s genus ; _ ie . _
how many ` handles ' it has .
hence the intrinsic curvature energy is zero for a torus and @xmath70 for a sphere .
the _ total _ curvature energy , @xmath71 , is a _ conformal _
invariant @xcite .
_ ie . _ it is invariant under shape changes of the form @xmath72 where @xmath73 is a constant vector which parameterizes the conformal transformation @xcite .
a sphere has total curvature energy @xmath74 and its shape is not changed by such a transformation .
this is not true for tori .
an axisymmetric , circular cross section torus may be deformed by a conformal transformation . as a consequence of the circular cross - section , a component of @xmath73 parallel to the symmetry axis serves only to change the torus s size .
hence there remains only a one - parameter family of non - trivial conformal transformations . with @xmath73 lying in the symmetry plane ,
the surface is deformed into a non - axisymmetric torus such as the one in fig.([piccy ] ) , and as @xmath73 increases in magnitude , the asymmetry increases until , in the limit , the torus becomes a perfect sphere with an infinitesimal handle .
the handle , although infinitesimal , must contain an intrinsic curvature energy of @xmath75 and finite total curvature energy . a circular
, axisymmetric torus of aspect ratio ( ratio of radii of generating has total curvature energy @xmath76 which is a minimum at @xmath77 , the clifford torus , and infinite at @xmath78 , where the torus becomes self - intersecting .
willmore conjectured @xcite that clifford s torus and its conformal transformations have the _ absolute _ minimum curvature energy of any surface of toroidal topology ; a conjecture that is difficult to prove , as the most general toroidal surface can not easily be parameterized : tori can even be knotted .
accordingly , this paper will not explore the whole of phase space .
the discussion will be restricted to axisymmetric , circular cross - section tori and their conformal transformations , with the assumption that these are close to the ground state shapes . to be consistent ,
the ground - state genus - zero shape is approximated by a sphere : the deviations from sphericity calculated in @xcite are not included .
the field gradient energy term in eq.([f ] ) may be re - formulated . using integration by parts ,
it is simple to show that , on a closed surface , @xmath79 let @xmath80 be eigenfunctions satisfying the hermitian equation @xmath81 and the orthonormalization condition @xmath82 these are the landau levels .
@xmath49 may be expanded in this complete set of functions .
however , close to the mean - field transition , it is a good approximation to use only the landau levels with the lowest eigenvalue , @xmath83 . as is common in such problems , there is a degenerate set of lowest landau levels .
@xmath84 scales inversely with area .
so let us define a scale - invariant eigenvalue @xmath85 , by @xmath86 the free energy becomes @xmath87 where @xmath88 and @xmath89 are complex constants .
note that the mean field transition has been lowered from @xmath90 to @xmath91 where @xmath92 . on an intrinsically flat surface , @xmath93 but it is finite and positive - definite on a surface with finite gaussian curvature .
so ordering is frustrated by intrinsic curvature .
eq.([eigen ] ) may be re - written as the differential equation @xmath94 \partial_{a } \partial_{b } \phi_{p } + \left[(\frac{\partial_{b}-2ina_{b}}{\sqrt{g}})(\sqrt{g}\,g^{ab})\right ] \partial_{a } \phi_{p } + \left [ \frac{-ind_{a}}{\sqrt{g}}(\sqrt{g}\,g^{ab}a_{b } ) \right ] \psi_{p } = -\frac{4\pi}{\cal a } \lambda \phi_{p } .\ ] ] this is still general to any surface . on a sphere , in spherical polar
coordinates @xmath34 , eq.([eigen2 ] ) becomes @xmath95 which is solved by @xmath96 and @xmath80 of the form @xmath97 where @xmath98 takes integer values from @xmath99 to @xmath5 .
these functions form an alternative basis set to that used in @xcite and are linear combinations of that set .
they are given in @xcite in a different gauge .
the functions have @xmath100 zeros , which are topological defects of index @xmath7 .
the @xmath101-fold degeneracy reflects the freedom of the defects to lie anywhere on the surface at no energy cost to order @xmath102 .
on tori , @xmath80 and @xmath85 depend on @xmath103 and @xmath73 .
consider first the axisymmetric case ( @xmath104 ) .
let us define coordinates @xmath105 .
the azimuthal angle is @xmath106 , varying around the large generating circle and subtended at the centre .
the angle @xmath107 varies around the smaller generating circle and is measured from the symmetry plane , as shown in fig.([coords ] ) .
let the small generating circle have radius @xmath108 . in this coordinate system ,
the following quantities are found : @xmath109 the landau level equation eq.([eigen2 ] ) becomes @xmath110 \phi_{p } = \frac{4\pi\lambda}{\cal a}\phi_{p } , \ ] ] which is manifestly invariant under @xmath111 with complex conjugation , reflecting the up - down mirror symmetry of the torus .
note that changing variable from @xmath107 to @xmath112 and setting @xmath103 to zero re - creates the sphere equation ( eq.([eigensphere ] ) ) .
eq.([eigentorus ] ) has separable variables and is solved by functions of the form @xmath113 for integer @xmath98 , where @xmath114f_{p } = \frac{4\pi\lambda}{\cal a}f_{p}\ ] ] and @xmath115 is real .
it is immediately apparent from the form of eq.([eigentorus2 ] ) that @xmath116 and consequently @xmath80 and @xmath117 are degenerate pairs of states .
the ground - state solutions of eq.([eigentorus2 ] ) were found numerically for various values of @xmath98 and @xmath103 , using rayleigh - ritz variational method .
trial functions were synthesized from a truncated fourier series , @xmath118 which has the appropriate periodic boundary conditions .
the fourier coefficients were used as variational parameters to minimize the integral @xmath119 f_{p}^{\mbox{\tiny trial}}\:d\eta\ ] ] subject to eq.([normalization ] ) .
the minimal value of eq.([variationalfunctional ] ) is an upper bound on @xmath120 , and the corresponding function @xmath121 is approximately @xmath115 .
using a fourier series truncated after frequency @xmath122 creates a @xmath123-dimensional parameter space to explore in the numerical minimization .
so cpu time increases very rapidly with @xmath122 .
a second numerical method was also used , which involved numerically ` integrating ' eq.([eigentorus2 ] ) forward in @xmath107 from @xmath124 to @xmath125 and minimizing the resultant discontinuities in @xmath115 and its derivative , with respect to the initial conditions and @xmath85 .
this method is more accurate than the rayleigh - ritz method but more difficult to use since , depending on the starting conditions , it may find a non - ground - state solution .
however , it was useful in establishing the accuracy of the upper bounds .
using @xmath126 , @xmath120 was found to an accuracy of 0.05% in the worst case . in fig.([lambdagraphs ] ) , @xmath127 is plotted for @xmath128 , with data points at increments of 0.5 in @xmath103 and cubic spline interpolation .
the states are labelled @xmath129 .
one might have guessed that @xmath130 would be the ground state in each case . and indeed , for large aspect ratios , when the torus is almost cylinder - like , the @xmath131 states do have the lowest energy .
these states are even functions of @xmath107 , being larger at @xmath132 than at @xmath133 where there is more absolute intrinsic curvature , and do not vary with @xmath106 .
_ the field does not twist with respect to the coordinate lines as @xmath106 increases .
however , for smaller @xmath103 , states with non - zero @xmath98 are energetically favoured . before explaining the reason for this , it is necessary to describe these states .
fig.([states ] ) depicts a selection of states for @xmath134 on a clifford torus .
the nematic - order field is represented by sets of double - ended lines directed along the orientations of order in the surface of the torus .
the length of each line is proportional to @xmath135 at that point .
note however that the three - dimensional views of the toroidal surface cause the lines to be foreshortened at oblique angles to the line of sight .
two views are shown of each field on the toroidal surface , as well as a graph of @xmath136 .
note that positive @xmath98 states have a large field on the bottom of the torus and small on the top , and vice versa for negative @xmath98 .
this asymmetry becomes more pronounced with increasing @xmath98 , with increasing @xmath5 , and with decreasing @xmath103 . as stated earlier , ordering is frustrated by intrinsic curvature . and
the net amount of intrinsic curvature on the torus is constant .
but as @xmath103 decreases , the curvature is redistributed .
it becomes more negative around the hole and more positive around the outside of the torus , with the top and bottom remaining fairly flat .
so this is where the field prefers to reside and be parallel .
although in eq.([form ] ) non - zero @xmath98 states appear to twist with respect to the coordinate basis as @xmath106 increases , the basis itself twists on the top and bottom of the torus as the hole is circled .
the two twists cancel .
note that the diagram of the @xmath137 state depicts a field which is as parallel as possible on the top ` face ' of the torus .
of course , the field on the inner and outer ` rims ' still contributes to the free energy so , as @xmath103 increases , the ground - state value of @xmath98 decreases by a series of ` transitions ' .
this is clear in fig.([lambdagraphs ] ) .
these should not be regarded as true phase transitions , since the lowest landau level approximation breaks down completely at these points . as a crossover between eigenvalues
is approached , the true state is a mixture of the two states . it should be emphasized that the tori studied here are true , curved tori , embedded in 3-space .
they should not be confused with a flat torus , which is a plane with opposite edges identified , for which the above energetics would not apply .
the two have the same topology , but different local geometries .
when @xmath138 , the landau level equation ( eq.([eigen2 ] ) ) becomes much more complicated and its variables are no longer separable .
however , rayleigh - ritz variational method is again possible , as @xmath139 but the trial wavefunctions are now complex , periodic functions of two variables .
thus , synthesizing them with a truncated , two - dimensional , complex fourier series up to frequency @xmath122 produces an @xmath140-dimensional space of variational parameters to be numerically explored .
hence ground - state functions of the coordinates with high frequency components can not viably be found .
this is unfortunate since conformally transforming an axisymmetric torus with the coordinate system described above , leaves the coordinates very unevenly distributed over the surface , with most of the interval from @xmath141 to @xmath142 in @xmath107 and @xmath106 covering only the small ` handle ' part of the torus and only a small range covering most of the area .
this problem is partly solved by a coordinate transformation @xmath143 which preserves the metric s diagonalization , where @xmath144\tan\frac{\rho}{2 } & = & [ 1-\beta(\alpha+1)]\tan\frac{\eta}{2 } \\
\left[1+\beta(\alpha+1)\right]\tan\frac{\sigma}{2 } & = & [ 1-\beta(\alpha+1)]\tan\frac{\phi}{2}\end{aligned}\ ] ] although this still leaves some high frequency components in the distribution of the coordinates on the handle .
using these coordinates , eq.([thing ] ) is very complicated but two special cases are already known : * when @xmath104 , the torus is axisymmetric and @xmath120 was found by the method in section [ axisym ] .
* when @xmath145 , where @xmath108 is the radius of the small generating circle , the torus becomes a sphere with an infinitesimal handle .
hence @xmath96 and the ground - state function is that of a sphere with all @xmath100 zeros at the handle .
unfortunately , the problem is too computer - intensive to produce any useful results for intermediate values of @xmath73 .
using @xmath146 and performing the minimization in a 98-dimensional parameter space took a considerable amount of cpu time per data point . at this level , no states of intermediate @xmath73 were found to have a lower eigenvalue than the known axisymmetric or spherical states . some _ local _ minima were found but these are very likely to be artifacts of the varying reliability of the upper bound , which was high by around 10% for @xmath104 and 60% in the spherical limit .
more research on this problem is planned .
but for the moment , it is necessary to work with the ansatz that : * a spherical vesicle exists when its total free energy is lower than that of an axisymmetric torus . * an axisymmetric torus exists when its total free energy is lower than that of a sphere _ and _ its field energy alone is lower than that of a sphere . * a non - axisymmetric torus exists otherwise .
it was noted earlier that @xmath80 and @xmath117 are degenerate to quadratic order in @xmath49 .
so , from eq.([expansion ] ) , @xmath49 is a linear combination of @xmath80 and @xmath117 , where @xmath98 has the ground - state value for the given aspect ratio
. the relative proportions of the two states are determined by the @xmath147 term in eq.([f ] ) .
let @xmath148 be the function which minimizes @xmath149 on a surface of area @xmath150 subject to @xmath151 , and let j be the minimal value of that integral .
then eq.([f ] ) is minimized by @xmath152 in the regime of the lowest landau level approximation , where @xmath21 is real .
the gauge invariance and axial symmetry of the problem allow us to choose the phases of both @xmath153 and @xmath154 .
hence @xmath155 is determined by a minimization with respect to just one parameter . for some values of @xmath103 and @xmath98 , it is found that @xmath156 .
so the ` top ' and ` bottom ' states co - exist and , although each is fairly parallel on its own side of the torus , the two do not mesh smoothly where they meet .
consequently , there are @xmath157 defects each of index @xmath7 around the torus s outer ` rim ' and @xmath157 of index @xmath158 round the inside of the hole : a total of @xmath159 defects .
hence as @xmath103 varies and @xmath98 jumps between integer values as described above , vortex - antivortex quartets are formed or destroyed .
let these symmetrical states be labelled @xmath160 .
the @xmath161 state for @xmath77 is represented in fig.([n4p3 ] ) . for other values of @xmath103 and @xmath98 , where the overlap between @xmath80 and @xmath117 is large and
hence the defects in the field become expensive , the up - down symmetry is spontaneously broken , so that @xmath162 or @xmath163 .
let these broken - symmetry states be labelled @xmath164 .
as @xmath103 varies , there are transitions between s and b states , which , unlike the transitions betwen states of differing @xmath98 described in section [ axisym ] , may be regarded as true phase transitions within mean - field theory , although they may be blurred by the inclusion of fluctuations . as a rule of thumb ,
the @xmath147 term chooses s or b on the basis of which covers the surface more uniformly .
a table , giving the values of @xmath103 for all transitions between ground - state field configurations on the torus , is produced in the appendix .
the free energy is next minimized with respect to @xmath21 , giving the overall magnitude of the order parameter field . below the continuous transition ,
the free energy is a minimum when @xmath165 it was explained in section [ sphere ] that the spherical landau levels are degenerate with respect to the positions of the vortices .
the @xmath147 term lifts this degeneracy and makes the vortices repel .
hence , following park et al .
@xcite , the ground state configuration is taken to be that of highest symmetry .
so , when @xmath166 the two zeros are antipodal .
when @xmath134 , the zeros lie at the corners of a regular tetrahedron and , when @xmath167 , at the corners of an icosahedron .
the values of @xmath168 were found using these configurations . for @xmath169 ,
the defects lie at the corners of a cube whose top face has been rotated through @xmath170 relative to the bottom face .
the distance between these two faces is then varied to minimize @xmath171 .
the optimal separation is found to be very close to the length of a side of the original cube .
the minimized free energy of the sphere is now completely determined . expressing it as a fraction of the bending rigidity gives an equation relating just four dimensionless combinations of thermodynamic parameters : @xmath172 & = & 4\pi\left(2+\left[\frac{\kappa_{g}}{\kappa}\right]\right ) -2\pi\left[\frac{c^2}{{\cal a}u\kappa}\right ] \frac{\left(\frac{1}{4\pi}\left[\frac{{\cal a}r}{c}\right]+n\right)^2 } { j_{(\mbox{\tiny sph})}(n ) } \;\;\;\;\ ; \mbox{for } \;\;\ ; r < \frac{-4\pi cn}{\cal a } , \nonumber\\ & = & 4\pi \left(2+\left[\frac{\kappa_{g}}{\kappa}\right]\right ) \hspace{2.25 in } \mbox{otherwise.}\end{aligned}\ ] ] on the torus , the free energy remains a function of @xmath103 only : @xmath173 & = & \frac{2\pi^2\alpha^2}{\sqrt{\alpha^2 - 1 } } -2\pi\left[\frac{c^2}{{\cal a}u\kappa}\right ] \frac{\left(\frac{1}{4\pi}\left[\frac{{\cal a}r}{c}\right ] + \lambda_{0}(\alpha)\right)^2 } { j_{(\mbox{\tiny tor})}(\alpha , n ) } \;\;\;\;\;\ , \mbox{for } \;\;\ ; r < \frac{-4\pi c\lambda_{0}(\alpha ) } { \cal a } , \nonumber\\ & = & \frac{2\pi^2\alpha^2}{\sqrt{\alpha^2 - 1 } } \hspace{2.575 in } \mbox{otherwise}\end{aligned}\ ] ] which relates just three dimensionless combinations of thermodynamic parameters .
when eq.([ftor ] ) is minimized with respect to @xmath103 , some of the physical richness of the problem is lost , as many of the interesting phases and all the broken symmetry states occur on tori of aspect ratios which are not the energetic optimum . however , other interesting features arise , as will be seen in the phase diagrams in figs.([phasediags1])-([phasediags1end ] ) , which result from a phase space restricted to only axisymmetric tori , and show contours of constant @xmath103 as well as lines of first order and continuous phase transitions .
the phases are labelled with the notation introduced earlier .
the diagrams are plotted in the plane of the two independent combinations of thermodynamic parameters in eq.([ftor ] ) . in each case , to the far right of the diagram is the high - temperature phase , for which @xmath174 . in this phase ,
only the curvature energy is non - vanishing so the aspect ratio is @xmath175 ; the clifford torus .
note that the continuous transition , at which ordering arises , occurs not at @xmath90 , but is suppressed by the manifold s curvature to @xmath91 , as defined in section [ landau ] for the clifford torus .
note that @xmath176 depends on aspect ratio , aspect ratio depends on @xmath49 and @xmath49 depends on @xmath176 .
this non - linearity is the cause of the first - order phase transitions .
consider for instance the @xmath166 diagram ( fig.([phasediags1 ] ) ) .
a vesicle with small @xmath0 is easily deformed by the @xmath177-state ordering to large aspect ratios at low temperatures . at large aspect ratios , @xmath176 is close to zero . as the temperature rises , @xmath49 diminishes , and can no longer hold the aspect ratio so far from the clifford torus . as the aspect ratio falls
, ordering becomes more difficult .
there comes a temperature at which the torus jumps down a first order phase transition to the clifford torus , where @xmath178 . at higher values of @xmath0 ,
the transition is continuous .
the @xmath166 phase diagram has a tri - critical point at @xmath179 and @xmath180 . for higher values of @xmath5 ,
the first order transition persists for all values of @xmath0 , and meets the continuous transition at a multi - critical point .
the continuous transition here is between the high - temperature phase and the @xmath181 phase , for which @xmath182 since the gradients in fig.([lambdagraphs ] ) are positive close to the clifford torus .
the free energy ( eq.([f ] ) ) has now been minimized with respect to the order parameter field s configuration and magnitude and with respect to the toroidal aspect ratio .
it remains only to minimize with respect to topology , according to the prescription given in section [ non - axisym ] .
figs.([phasediags2])-([phasediags2end ] ) show the resulting phase diagrams , which include spherical , and axisymmetric and non - axisymmetric toroidal phases of vesicles .
no @xmath103 contours are shown in these figures .
the positions of the transitions vary with @xmath183 , and transition lines are shown for various values of this quantity .
figs.([phasediags2])-([phasediags2end ] ) should therefore each be regarded as several phase diagrams superimposed .
the @xmath181 phases do not persist when this larger phase space is explored , except for @xmath184 , for which @xmath120 is less than the sphere value of @xmath185 , when @xmath103 is close to @xmath175 .
the diagrams are each divided into shaded and unshaded regions by a heavy line . in the shaded regions ,
vesicles are either spheres or non - axisymmetric tori while , in unshaded regions , either spheres or axisymmetric tori occur .
consider first the @xmath166 , @xmath134 and @xmath169 phase diagrams .
a transition line , whose position depends on @xmath186 separates tori from spheres .
note that for @xmath187 , the transition line is closed , with spheres on the inside and tori of one sort or another on the outside , while for @xmath188 , the line is open , with spheres on the high - temperature side and axisymmetric tori on the low - temperature side .
this magical value of @xmath189 is the critical value of @xmath186 at which spheres would become tori if the model described curvature energy alone and did not include terms in @xmath49 .
note that in these three phase diagrams the separatrix ( the @xmath190 transition ) has a vertical limb at the value of @xmath191 at which ordering first forms on the sphere , _ ie . _
at @xmath192 . the @xmath167 diagram in fig.([phasediags2end ] )
is a little more complicated since there are two distinct phases of ordering on axisymmetric tori .
again , the diagram is divided into shaded and unshaded regions .
the shaded region is now split into two parts .
the part on the right contains spheres if @xmath188 , and non - axisymmetric tori otherwise .
the part on the left contains spheres on the high @xmath193 side of the transition line ( whose position depends , of course , on @xmath186 ) and non - axisymmetric tori on the other side .
the un - shaded region , which contains axisymmetric tori and spheres , is also divided into two parts .
the part in the top left of the diagram behaves as described above for the other phase diagrams .
the other part may ( depending on @xmath186 ) contain a closed transition line , which , unlike the other closed transition lines we have met , has spheres on the outside and axisymmetric tori , of small aspect ratio and @xmath194 field configuration , on the inside .
the four phase diagram of figs.([phasediags2])-([phasediags2end ] ) may seem complicated but they are easily understood if one remembers that : _ a ) _ they are several diagrams superimposed , and _
b ) _ the regions in which spherical vesicles exist become smaller as @xmath1 increases , since spheres have more intrinsic curvature than tori . having established from these diagrams that an axisymmetric torus exists in a certain region , its aspect ratio may be found from figs.([phasediags1])-([phasediags1end ] ) .
it should be mentioned that the curves defining the edges of the shaded regions in figs.([phasediags2])-([phasediags2end ] ) have been slightly smoothed by hand to remove pixcellation which arose from numerical rounding errors .
some order - of - magnitude calculations are now presented to give an indication of the regimes of validity of the various approximations used .
the landau levels form a complete set of orthogonal functions in which to expand the order parameter ; @xmath195 .
each eigenfunction @xmath80 has an eigenvalue @xmath196 defined by eqs.([eigen ] ) and ( [ lambda ] ) .
the lowest landau levels alone may be used if @xmath197 where @xmath198 . as an indicator of the region
in which this is valid , let us calculate @xmath199 for gaussian fluctuations above the continuous transition , using the quadratic hamiltonian @xmath200 where @xmath201 and @xmath202 .
the partition function is @xmath203 hence @xmath204 so for the lowest landau levels , @xmath205 and for the next lowest level , @xmath206 .
hence the ratio of correlators is @xmath207 where @xmath208 is the x - coordinate in the phase diagrams , measured with respect to the transition .
we see that @xmath209 is small , and hence the lowest landau level approximation is good , when @xmath210 , the right hand side of which may be estimated from fig.([lambdagraphs ] ) . for the clifford torus , @xmath211 so the approximation is quantitatively accurate only within @xmath2123 of the transition , along the x - axis .
however , figs.([phasediags1])-([phasediags2end ] ) cover a much larger region than this , in order to clarify the qualitative features of the diagrams , for which the calculation at this level of approximation is a useful guide .
it should be noted that , at ` transitions ' from one lowest landau level configuration to another , @xmath213 , so the approximation must break down here , as noted in section [ axisym ] . also , in the large-@xmath103 phases below the first - order transitions , @xmath176 is higher , so the system is far from the lowest landau level regime .
these phases are included in the diagrams in order to show the first - order transitions . for this order - of - magnitude analysis ,
let us treat two parts of the free energy separately : first we will demand that fluctuations in the helfrich hamiltonian may be neglected in the absence of orientational order , then , within this fixed , non - fluctuating environment , it will be required that the terms in @xmath49 satisfy the ginzburg criterion for the validity of mean - field theory . in a simplistic calculation ,
good over only a short range , the monge gauge is used to parameterize a locally horizontal surface , defining its z - coordinate as a function of x and y. keeping terms of second order in z , and expanding z in plane - wave modes : @xmath214 gives @xmath215 note that the gaussian curvature energy does not influence fluctuations as it is a topological invariant . hence @xmath216 so , for fluctuations to be negligible on the scale of the system size requires @xmath217 .
this is intuitively obvious ; that a large bending rigidity is required to combat fluctuations .
in fact , this simple calculation on a flatish surface is over - stringent , since intrinsic curvature , as exists in closed surfaces , reduces the size of fluctuations @xcite .
the ginzburg criterion is now computed for the ginzburg - landau part of the free energy .
this is a comparison between the size of the discontinuity in specific heat at the phase transition , calculated with mean - field theory , and the size of the divergence in specific heat due to gaussian fluctuations .
close to the transition , @xmath218 goes approximately linearly with temperature @xmath219 .
hence @xmath220 , where @xmath221 is the shifted mean - field transition temperature , and @xmath222 is some constant . from eq.([fsph ] ) it follows that the mean - field discontinuity in specific heat is @xmath223 for simplicity , the specific heat of fluctuations is calculated for the sphere , above the phase transition .
using the quadratic form eq.([quadratic ] ) , and only the @xmath101 lowest landau levels , gives a free energy density @xmath224 .
hence the singular part or the specific heat is @xmath225 giving a critical exponent @xmath226 since we are in a zero dimensional @xmath227-space .
the ratio of these two specific heats is @xmath228 where @xmath229 and mean field theory is good for @xmath230 .
this condition is equivalent to demanding that tree graphs dominate over one - loop graphs . )
so the threshold , in the phase diagrams , between the mean - field regime and the fluctuation - dominated regime is given by the curve @xmath231 , where @xmath232 and @xmath233 , having set @xmath234 and @xmath235 .
at a height @xmath236 , this region has width @xmath237
. this should be much smaller than the region of interest _ ie . _
that in which the lowest landau level approximation is valid .
so @xmath238 and , by inspection , in all the phase diagrams , we are interested in @xmath239 . consequently the condition for a mean - field treatment of the order parameter to be valid in this investigation is just @xmath240 , which is always consistent with the validity of a mean - field treatment of the membrane shape , calculated above .
the ginzburg - landau free energy functional is and expansion in @xmath49 , truncated at order @xmath147 .
this is valid when @xmath241 . in the mean - field regime
, this is always the case above the continuous transition . below the transition , @xmath242 , from eq.([size ] ) .
so the inequality is marginal when @xmath243 where @xmath244 and @xmath245 are given above .
this defines a region in the phase diagrams of width @xmath246 at a height @xmath247 , which is in the region of interest .
again , let s demand that @xmath248 to be consistend with the region of validity of the lowest landau level approximation . consequently
, it is required that @xmath249 .
dropping constants of order unity , we find that mean - field theory , the ginzburg - landau model and the lowest landau level approximation are all valid in a region of half - width @xmath250 in the phase diagrams , given that @xmath251 close to the ordering transition , @xmath49 is very small , so the order can not deform the torus far from clifford s aspect ratio @xmath175 , or the genus - zero shape far from a sphere , so the vesicle shapes used in this theoretical investigation are undoubtedly good . as park et al .
discovered in @xcite , @xmath49 wants to live in a flat space , and can achieve a flattening of the membrane via the gauge - like coupling in eq.([f ] ) .
this it will do where it is large , at the expense of an increased curvature where @xmath49 is small _ eg .
_ at the defects
. this pay - off is unavoidable , since the integral of intrinsic curvature is topologically fixed .
hence spherical vesicles are slightly deformed by the in - plane order , into rounded polyhedra , whose vertices are at the order parameter s vortices .
this is the effect which has been ignored in this paper .
the same process will also occur in tori , especially in those parts of the phase diagrams where ordering is predicted to deform the tori to large aspect ratios .
in general , one might expect that the larger the aspect ratio , the worse the approximations of circular cross - section and axisymmetry .
but nature is on our side here . since , in these large - aspect - ratio tori , @xmath49 is always in the @xmath252 configuration , which has no vortices , and becomes very uniform for large aspect ratios .
so , although this approximation results in some quantitative inaccuracies , they are likely to be small .
also , they must change the free energy of the torus and the sphere , if not by the same amount , then at least in the same direction .
so the phase diagrams may not have exactly the right numbers , but they do have more - or - less the right shape .
it is clear that orientational order in a toroidal geometry is a subject rich in physical diversity , and deserving of further study in condensed matter theory .
also , introducing orientational order into the surface is an interesting way of stabilizing a torus against the massless goldstone mode of conformal transformations , without resorting to spontaneous curvature or a constraint of constant volume or bilayer area difference . as far as experimental verification
is concerned , the principal results of this paper are embodied in the final four phase diagrams .
the approximations used in their derivation may result in quantitative inaccuracies , but the main qualitative features of these diagrams are predicted to be seen in experiment in the future . at present an experimental realization is somewhat difficult , due to the fragile nature of laboratory - produced vesicles .
i would like to thank prof .
mike moore for his guidance in this work .
i am grateful for the helpful comments of tom lubensky , udo seifert , renko de vries and the numerous others with whom i have conversed on the subject .
i wish to acknowledge the epsrc for my funding .
the axisymmetric toroidal aspect ratios at which the order parameter field , which minimizes the free energy in the regime of the lowest landau level approximation , changes its configuration between states of different @xmath98 or different symmetry are given in the table below .
see sections [ axisym ] and [ axisym2 ] for explanations of these transitions .
note that an @xmath166 ( vector ) field has no such transitions , as its ground state is always @xmath253 .
999 f.c.mackintosh and t.c.lubensky , phys.rev.lett . * 67 * 1169 ( 1991 ) .
jeongman park , t.c.lubensky and f.c.mackintosh , europhys.lett .
* 20,3 * 279 ( 1992 ) .
d.e.rutherford , _ vector methods _ ( oliver & boyd ltd . , 1959 ) .
jeffrey r.weeks , _ the shape of space _ ( marcel dekker , 1985 ) .
b.fourcade , j.phys.ii france * 2 * 1705 ( 1992 ) .
f.jlicher , u.seifert and r.lipowsky , j.phys.ii france * 3 * 1681 ( 1993 ) .
t.j.willmore , _ total curvature in riemannian geometry _ , ( ellis horwood ltd . ,
chicester , 1982 ) .
j.a.oneill and m.a.moore , phys.rev.b * 48 * 374 ( 1993 ) .
f.david , geometry and field theory of random surfaces and membranes , in _ statistical mechanics of membranes and surfaces _ ,
d.nelson , t.piran and s.weinberg ( world scientific publishing co. pte . ltd . , 1989 ) r.de vries , private communication .
|
a theoretical study of toroidal membranes with various degrees of intrinsic orientational order is presented at mean - field level .
the study uses a simple ginzburg - landau style free energy functional , which gives rise to a rich variety of physics and reveals some unusual ordered states .
the system is found to exhibit many different phases with continuous and first order phase transitions , and phenomena including spontaneous symmetry breaking , ground states with nodes and the formation of vortex - antivortex quartets .
transitions between toroidal phases with different configurations of the order parameter and different aspect ratios are plotted as functions of the thermodynamic parameters .
regions of the phase diagrams in which spherical vesicles form are also shown .
= msbm10 epsf
|
the multifractional brownian motion ( mbm ) has been independently introduced by @xcite and @xcite as a natural extension of the well - known fractional brownian motion ( fbm ) .
the main idea behind these two works was to drop the stationary assumption on the process , and allow the hurst exponent to change as time passes . in this way
, the mbm is parametrized by a function @xmath0 , usually continuous , and happens to be an interesting stochastic model for non - stationary phenomena ( e.g. signal processing , traffic on internet , terrain modelling , ) .
since its introduction , several authors have investigated sample path properties of the mbm .
for instance , @xcite and @xcite have respectively studied the law of iterated logarithm and the hlder regularity of its trajectories . in the latter , the box and hausdorff dimensions of the graph
have also been determined . moreover ,
the fine covariance structure have been analysed and refined by @xcite , @xcite and @xcite . finally , the local time and the hausdorff dimension of the level sets have recently been considered by @xcite .
several multiparameter extensions , which generalise the classic lvy fractional brownian motion and fractional brownian sheet , have also been introduced and investigated by @xcite @xcite and @xcite .
fine geometric properties obtained in the aforementioned works usually rely on a key assumption @xmath1 on the hurst function : @xmath2 owing to this hypothesis , the study of the @xmath1-mbm is usually easier and leads to results closely related to their counterparts on fractional brownian motion .
this structure of the mbm is due to the form of its covariance which , under @xmath1 , is locally equivalent to fbm s one with hurst exponent @xmath3 : @xmath4 } } \asymp { \lvertu - v\rvert}^{2h(t)},\ ] ] for any @xmath5 .
the study of the more general multifractional brownian motion , i.e. when the assumption @xmath1 does not hold , has only been recently considered by @xcite and @xcite . both have investigated the hlder regularity of the sample paths , and more precisely the pointwise and local hlder exponents , which are defined by @xmath6 and @xmath7 these two exponents aim to characterise the local hlder behaviour of the trajectories at @xmath8 , by respectively comparing its oscillations to a power of @xmath9 or @xmath10 .
for instance , it is well - known that a fractional brownian motion @xmath11 satisfies with probability one and for all @xmath12 , @xmath13 .
the general form of the hlder regularity of the multifractional brownian motion has first been obtained by @xcite , proving that for all @xmath14 @xmath15 the previous result only holds for a fixed @xmath16 , and thus not uniformly on the sample paths .
we also note that the geometric properties of a generic mbm are not a simple extension of those on fbm , but display a more complicated structure where the fine geometry of the hurst function intervene .
@xcite has recently extended this study , obtaining a uniform characterisation of the local hlder exponent and proving that its pointwise exponent can behave randomly as time passes .
the resulting process , with unusual sample path properties , is sometimes called the _ irregular multifractional brownian motion _ in the literature . the main goal of this work is therefore to push further the study of this irregular mbm . in section [ sec : mbm_representation ] , we first discuss the existence of an alternative deterministic representation of the fractional brownian field , which will then be used to study more precisely the geometric properties of the mbm .
section [ sec : mbm_regularity ] is devoted to the 2-microlocal and hlder regularity of the trajectories ( theorem [ th:2ml_mbm ] and proposition [ prop : pointwise_mbm ] ) , and therefore extends the results obtained by @xcite and @xcite .
hausdorff and box dimensions of the graph , and images of fractal sets , are investigated in section [ sec : mbm_dimension ] ( theorems [ th : mbm_dim_box ] , [ th : mbm_dim_haus ] and [ th : mbm_dim_images ] ) , where further connections with the geometry of the hurst function are presented .
appendix [ sec : mbm_appendix ] gathers a few deterministic results related to the 2-microlocal frontier and the fractal dimensions which are used along this work .
finally , several examples are also given in these different sections to illustrate the main results and some particular aspects of the geometry of the irregular multifractional brownian motion .
the question of the equivalence of the different mbm representations is known to be non - trivial .
on one side , @xcite have introduced the mbm as the following stochastic integral @xmath17 } { \ensuremath{\mathrm d}\xspace}w_u.\ ] ] whereas on the other hand , @xcite have considered a real - harmonizable definition : @xmath18 @xcite have proved that the covariance structure is slightly different between these two definitions .
considering this fact , they have suggested to name multifractional brownian motion any process @xmath19 which has the following form @xmath20 where @xmath21 and the fractional brownian fields @xmath22 are defined by @xmath23 } { \ensuremath{\mathrm d}\xspace}w_u,\ ] ] for all @xmath12 and @xmath24 . in the next proposition , we present an alternative deterministic representation for these fractional brownian fields which will be very useful for the study of sample path properties . [
prop : rep_fgf ] suppose @xmath25 is a brownian motion such that @xmath26}})$ ] , where @xmath27 is a wiener measure . then , for all @xmath12 and any @xmath24 , the fractional brownian field @xmath28 is almost surely equal to : * if @xmath29 , @xmath30 } { \ensuremath{\mathrm d}\xspace}u .
\end{aligned}\ ] ] * if @xmath31}}$ ] and @xmath32 is a fixed number such that @xmath33 , @xmath34 or equivalently , @xmath35 } } } b_u ( s - u)_\pm^{h-\frac{1}{2 } } { \ensuremath{\mathrm d}\xspace}u \biggr)}{\bigl\{(t ) - ( 0)\bigr\ } } \nonumber \\ & \pm \frac{{h-\tfrac{1}{2 } } } { \gamma{\bigl(h+\tfrac{1}{2}\bigr ) } } \int_{{\ensuremath{(\pm\infty,\pm t ] } } } b_u { \bigl [ ( t - u)_\pm^{h-3/2 } - ( -u)_\pm^{h-3/2 } \bigr ] } { \ensuremath{\mathrm d}\xspace}u .
\end{aligned}\ ] ] without any loss of generality , we may only prove the equality for the term @xmath36 .
let first consider the case @xmath12 and @xmath37 . using ito s lemma on the product @xmath38
, we obtain for any @xmath39 @xmath40 hence , subtracting the case @xmath41 , @xmath42 } & = -\int_x^t { \bigl [ ( t - u)_+^{h-1/2 } - ( -u)_+^{h-1/2 } \bigr ] } { \ensuremath{\mathrm d}\xspace}b_u \\ & + ( h-\tfrac{1}{2 } ) \int_x^t b_u { \bigl [ ( t - u)_+^{h-3/2 } - ( -u)_+^{h-3/2 } \bigr ] } { \ensuremath{\mathrm d}\xspace}u .
\end{aligned}\ ] ] the brownian motion is known to satisfy @xmath43 .
moreover , we observe that @xmath44 } \sim_{-\infty } ( h-\tfrac{1}{2 } ) t^{h-1/2 } ( -x)^{h-3/2}$ ] . therefore , as @xmath45 , @xmath46 } \xrightarrow[x\rightarrow -\infty]{a.s . } 0.\ ] ] similarly , using the dominated convergence theorem , @xmath47 } { \ensuremath{\mathrm d}\xspace}u \xrightarrow[x\rightarrow -\infty]{a.s . }
\int_{-\infty}^t b_u { \bigl [ ( t - u)_+^{h-3/2 } - ( -u)_+^{h-3/2 } \bigr ] } { \ensuremath{\mathrm d}\xspace}u,\ ] ] finally , owing to the @xmath48-continuity of the stochastic integral , @xmath49 } { \ensuremath{\mathrm d}\xspace}b_u \xrightarrow[x\rightarrow -\infty]{l^2(\omega ) } \int_{-\infty}^t { \bigl [ ( t - u)_+^{h-1/2 } - ( -u)_+^{h-1/2 } \bigr ] } { \ensuremath{\mathrm d}\xspace}b_u,\ ] ] which proves the expected equality .
let now consider the second case @xmath50 .
the calculus is quite similar , with only a slight modification in the application of ito s lemma between @xmath32 and @xmath51 .
@xmath52 for any @xmath53 .
owing to the hlder continuity of the sample paths of @xmath54 , the previous expression converges when @xmath55 .
in addition , the calculus presented in the case @xmath56 holds as well on the interval @xmath57}}$ ] , therefore proving equation .
finally , the second form of the fractional brownian field is obtained using an alternative representation of the riemann
liouville fractional derivative of a function on the interval @xmath58}}$ ] ( see the book of @xcite for more information on the subject ) .
this type of representation has already been exhibited and studied by @xcite for the fractional brownian motion , even though his proof is based on a different approach ( approximation of the brownian motion with lipschitz functions ) .
the case @xmath59 has also been considered by @xcite .
as presented in the next simple lemma and owing to the hlder continuity of the sample paths of brownian motion , we obtain a deterministic representation of the fractional brownian field @xmath60 which is continuous in both variables .
it will be very useful in the two next sections to study in the sample path properties of the multifractional brownian motion .
[ lemma : cont_fgf ] let @xmath25 be a continuous brownian motion and @xmath61 be the fractional brownian field defined in proposition [ prop : rep_fgf ] .
then , with probability one for all @xmath62 , the field @xmath63 is continuous at @xmath64 . as previously
, we only need to consider the component @xmath36 .
the continuity on the domains @xmath65 and @xmath66 is a simple consequence of the dominated convergence theorem : the deterministic integral is split in two parts on which the theorem can be applied with different bounds .
to threat the case @xmath67 , we observe that the representation in fact holds for all @xmath24 .
then , similarly to the previous case , the dominated convergence theorem implies that the first term converges to zero when @xmath68 whereas the second one simply converges to @xmath69 . as observed by @xcite ,
the introduction of the mbm is more delicate in the well - balanced case @xmath70 . indeed ,
when @xmath71 , the aforementioned definition of the fractional brownian field @xmath61 implies that @xmath72 almost surely .
hence , as presented in @xcite , to obtain a well - defined mbm , one needs to change the normalisation term , and more specifically , add a diverging factor : @xmath73 . considering this slightly modified definition of the well - balanced mbm
, we observe that it corresponds to a particular deterministic expression .
[ prop : rep_fgf_balanced ] suppose @xmath74 is a continuous brownian motion and @xmath75 .
then , for all @xmath76 , @xmath77 } { \ensuremath{\mathrm d}\xspace}u\quad\text{a.s.}\ ] ] where for any @xmath78 , @xmath79 .
let @xmath76 , @xmath80 and @xmath81 . owing to the representation obtained in proposition [ prop : rep_fgf ] and using the well - balanced re - normalisation , @xmath82 } { \ensuremath{\mathrm d}\xspace}u.\ ] ] without any restriction
, we may assume that @xmath83 .
then , let us observe the component of the integral corresponding to the interval @xmath84}}$ ] .
it is equal to @xmath85 for any @xmath53 , we note that @xmath86 .
furthermore , the other part converges as well to @xmath87 } } } ( b_u - b_{2t - u } ) ( t - u)^{h-3/2 } { \ensuremath{\mathrm d}\xspace}u
\\ \longrightarrow_{{\varepsilon}\rightarrow0 } & \int_{{\ensuremath{[t-\delta , t ] } } } ( b_u - b_{2t - u } ) ( t - u)^{h-3/2 } { \ensuremath{\mathrm d}\xspace}u , \end{aligned}\ ] ] since @xmath54 is hlder continuous at @xmath8 .
hence , the integral converges when @xmath55 and we obtain the representation presented in equation . on the contrary to the formulas presented in proposition [ prop : rep_fgf ] , we observe that the representation is an improper deterministic integral when @xmath88 .
also note that we may check that when @xmath71 , the corresponding gaussian process is , up to a constant , a brownian motion .
more precisely , @xcite have shown that it has the following stochastic integral representation @xmath89 } { \ensuremath{\mathrm d}\xspace}w_u.\ ] ] as proved in the next proposition , the well - balanced case is in fact a classic mbm defined with respect to the latter brownian motion . [ prop : rep_fgf_balanced2 ] suppose @xmath54 is a continuous brownian motion and @xmath61 denotes the well - balanced fractional brownian field .
define the brownian motion @xmath90 by @xmath91 } { \ensuremath{\mathrm d}\xspace}b_u.\ ] ] then , for all @xmath12 and @xmath24 , @xmath92 where @xmath93 designates the fractional brownian field with respect to @xmath90 .
note that the normalising function is continuous at @xmath71 .
there exist several ways to prove this result .
one might directly obtain the expression of the fractional brownian field @xmath93 using propositions [ prop : rep_fgf ] and [ prop : rep_fgf_balanced ] , but the calculus is unfortunately a bit tricky .
the proof happens to be simpler if we consider the fourier representation of @xmath61 obtained by @xcite . for all @xmath12 and @xmath24 , @xmath94 where @xmath95 is a complex gaussian measure such that @xmath95 is a complex wiener measure on @xmath96 and @xmath97 for any @xmath98 .
the well - balanced fractional brownian field then has the following representation @xmath99 we introduce a slightly modified complex gaussian measure @xmath100 , given by @xmath101 then , let @xmath93 denote the fractional brownian field defined with respect to @xmath100 . since @xmath102 , @xmath103 to conclude the proof , we need to check that the field @xmath93 corresponds to the same process defined in the proposition , i.e. the integral with respect to the brownian motion @xmath90 .
for this purpose , we simply need to consider the previous equality when @xmath71 .
we first note that @xmath104 .
hence , @xmath105 according to equation , this last equality proves that @xmath90 is the brownian motion with respect to which are defined the fractional brownian fields , and thus we clearly obtain the expected representation of @xmath61 .
the equality in law between the two processes can be obtained more directly using theorem 4.1 from @xcite .
nevertheless , proposition [ prop : rep_fgf_balanced2 ] indicates precisely how these two fractional brownian fields are related , and proves that the well - balanced one is simply a classic field with a change of brownian motion .
besides , this statement allows us to simply the study of the sample path properties of the mbm since we do not need to consider separately the specific well - balanced case @xmath70 .
following the convention initiated by @xcite , we will call _ multifractional brownian motion _ a process @xmath19 which has the following form : @xmath106 where @xmath107 .
as previously said , we may also suppose without any loss of generality that @xmath108 .
the hurst function @xmath109 will be assume to be at least a deterministic cdlgfunction .
we may nevertheless note that several of the following results remain valid if @xmath110 is replaced by a stochastic process with cdlgsample paths .
this last model has already been considered by @xcite and @xcite , where the latter has obtained some general sample path properties . for any cdlgfunction @xmath110 ,
the set @xmath111 is countable , and therefore we can easily see from the definition of the mbm that with probability one , @xmath112 furthermore , @xmath113 is a centered gaussian variable whose variance is proportional to @xmath114 . in this section
, we will use specific functional spaces , called 2-microlocal spaces , to characterize the fine sample path properties of the mbm .
the use of the unusual framework is particularly motivated by the deterministic fractional integrals and derivatives of brownian motion exhibited in proposition [ prop : rep_fgf ] .
indeed , the latter spaces behave particularly well under the action of fractional operators , which will allow us to deduce many properties on the mbm from the geometry of the brownian motion .
historically , the 2-microlocal formalism has been introduced by @xcite to study the singularities of the solutions of pdes , and has only been recently introduced in a stochastic frame by @xcite to characterise the fine regularity of processes .
several characterizations of the 2-microlocal spaces have been investigated in the literature , including the following time - domain one due to @xcite and @xcite .
[ def:2ml_spaces_01 ] let @xmath12 , @xmath115 and @xmath116 such that @xmath117 .
a function @xmath118 belongs to the _ 2-microlocal space _
@xmath119 if there exist @xmath120 , @xmath121 and a polynomial @xmath122 such that @xmath123 for all @xmath124 . in this work
, it will only be necessary to consider the case @xmath116 .
nevertheless , definition [ eq : def_2ml_spaces_01 ] can be extended @xmath125 .
we refer to @xcite for a more complete presentation on the subject .
as previously outlined , a very useful properties of the 2-microlocal spaces is their stability under the application of fractional integrations and derivations : @xmath126 where @xmath127 denotes the fractional integral of @xmath128 of order @xmath129 : @xmath130 . following the presentation of these functional spaces
, we can introduce a related regularity tool , named the _
2-microlocal frontier _ and defined by @xmath131 we observe that when using the 2-microlocal formalism , the local regularity at @xmath8 is characterised by a function , not a single coefficient . owing to inclusion properties of the 2-microlocal spaces , the map @xmath132 is well - defined and satisfies several interesting properties : * @xmath133 is a concave non - decreasing function ; * @xmath133 has left and right derivatives between @xmath134 and @xmath135 .
furthermore , if the modulus of continuity of @xmath128 is such that @xmath136 , the pointwise and local hlder exponents can be retrieved from the frontier : @xmath137 finally , owing to the stability of the 2-microlocal spaces , the application of a fractional integration simply corresponds to a translation of the 2-microlocal frontier along the vertical axis : @xmath138 the 2-microlocal frontier of some gaussian processes has been determined by @xcite . for instance , a fractional brownian motion @xmath11 , @xmath24 satisfies with probability one and for all @xmath12 : @xmath139 to investigate the 2-microlocal and hlder regularity of the multifractional brownian motion , we first need to properly study the behaviour of the fractional brownian field and its successive partial derivatives .
hence , in the next proposition , we consider uniformly on the variables @xmath8 and @xmath140 the 2-microlocal frontier of @xmath141 along the time axis .
[ prop:2ml_fgf ] suppose @xmath61 is a fractional brownian field .
for any @xmath24 and any @xmath142 , let @xmath143 denotes the process @xmath144 . then , with probability one , for all @xmath24 , @xmath142 and @xmath12 @xmath145 since we use the deterministic representation obtained in proposition [ prop : rep_fgf ] , we can set a fix @xmath146 in the proof .
for the sake of readability , let us first consider @xmath56 and remove the normalisation constant @xmath147 .
then , for any @xmath142 , the partial derivative with respect to @xmath140 is equal to @xmath148 } { \ensuremath{\mathrm d}\xspace}u.\ ] ] suppose @xmath149 is a non - empty open interval .
since the 2-microlocal frontier is a purely local property of the sample paths , we may restrict our study to @xmath150 with any loss of generality .
then , we note that the component @xmath151 } } } b_u { \bigl [ ( t - u)_\pm^{h-3/2 } \log^k(t - u)_\pm - ( -u)_\pm^{h-3/2 } \log^k ( -u)_\pm \bigr ] } { \ensuremath{\mathrm d}\xspace}u\ ] ] is a smooth function , and therefore does not have any influence on the 2-microlocal regularity .
hence , we focus on the study of the remaining integral @xmath152 } } } b_u ( t - u)_\pm^{h-3/2 } \log^k(t - u)_\pm { \ensuremath{\mathrm d}\xspace}u,\ ] ] when @xmath59 .
similarly , the case @xmath153 is reduced to the analysis of @xmath154 } } } b_u ( s - u)_\pm^{h-\frac{1}{2 } } \log^k(t - u)_\pm { \ensuremath{\mathrm d}\xspace}u \biggr)}(t).\ ] ] if we define the process @xmath155}}}$ ] , we observe that the two previous formulas correspond respectively to fractional - like integral and derivative of @xmath156 on the real axis .
compared to the usual definition , there is only a slight modification in the kernel with a logarithmic term .
furthermore , when @xmath153 , the fractional - like derivative is defined as the classic derivative of a integral of order @xmath157 .
since we know the behaviour of the 2-microlocal regularity under derivation , we may restrict our study to the fine analysis of the following process @xmath158 for any @xmath159 and @xmath142 .
we expect the 2-microlocal frontier of this integral to behave similarly to the one of a classic fractional integral of @xmath156 , since the only difference lies in a logarithmic term .
this easiest way to prove this statement is to compute the fourier transform of the corresponding pseudo - differential operator and show it has an effect equivalent to the fractional integration in the fourier space .
the computation of fourier transform of the tempered distribution @xmath160 is similar to the classic calculus on @xmath161 .
we first consider the laplace transform , with @xmath162 .
since this transform is analytic on the previous domain , we can restrict ourselves to the case that @xmath163 .
then , @xmath164 as the last term is also analytic , it implies the equality for any @xmath165 such that @xmath162 . to obtain the fourier transform of @xmath160 as a tempered distribution
, we simply have to consider @xmath166 and let @xmath55 .
the scalar product with a test function converges , and proves that the generalized fourier transform is equal to @xmath167 therefore , the fourier transform corresponds to a weighted sum of elements @xmath168 , with a phase correction term . to express the action of these pseudo - differential operators on the regularity of the brownian motion , we rely on the original definition of the 2-microlocal spaces in the fourier space and
make use of the arguments presented by @xcite to prove the stability .
more precisely , the aforementioned definition is based on the littlewood
paley decomposition of tempered distributions .
shortly , consider @xmath169 such that its fourier transform satisfies @xmath170 \ , 0 & \text{si}{\hspace{1em}}{\lvert\xi\rvert}\geq 1 , \end{cases } \end{aligned}\ ] ] and define for any tempered distribution @xmath171 and for all @xmath172 , @xmath173 , @xmath174 , @xmath175 and @xmath176 .
then , the littlewood
paley decomposition tells us that @xmath128 can be expressed as following @xmath177 where we observe that every term in the sum belongs to @xmath178 . roughly speaking
, this decomposition corresponds to the application of band - pass filters in the fourier space . as presented by @xcite , @xmath179
if and only there exists @xmath120 such that for all @xmath180 , @xmath181 observe that , up to a simple translation , we may assume that @xmath41 . then , define the rescaled function @xmath182 .
its fourier transform @xmath183 is carried by @xmath184 and @xmath185 is such that @xmath186 .
applying the operator @xmath187 in the fourier space on @xmath128 , we obtain a corresponding collection of rescaled functions @xmath188 which satisfy for every @xmath180 , @xmath189 } , \end{aligned}\ ] ] where @xmath190 is a constant only depending on @xmath191 and @xmath192 .
note that similarly to the classic fractional integral , the fourier multipliers @xmath193 coincide with schwartz functions on @xmath194 .
coming back into the time space , @xmath195 is given by @xmath196 where @xmath197 for every @xmath191 .
the convolution with a schwartz function @xmath198 preserves the polynomial decay , hence for every @xmath180 , @xmath199 . rescaling by a factor @xmath200 , the latter upper bound leads to the characterization of 2-microlocal spaces , proving that @xmath201 for any @xmath53 .
note that the property is slightly weaker than the classic stability under the action of fractional integration due to the logarithmic terms .
the converse case is obtained similarly , @xmath202 the same property holds for any weighted sum of @xmath203 and @xmath204 , and therefore , the 2-microlocal frontier of @xmath205 satisfies for any @xmath24 and @xmath142 @xmath206 it concludes the proof , since we have shown previously that the 2-microlocal frontier of @xmath143 is equal to the regularity @xmath205 .
@xcite has obtained a similar result on the field @xmath207 based on a multiresolution analysis of the latter . as outlined in the introduction , the hlder regularity of a general multifractional brownian motion has already been investigated by @xcite and @xcite . in particular
, the latter has proved that with probability one , the local hlder exponent is given by @xmath208 note that the behaviour at @xmath41 is slightly different as the form of the variance of increments differs at this point .
let us first obtain a estimate for the general 2-microlocal frontier .
for this purpose , we need to introduce the multiplicity of the component @xmath209 of the fractional field at @xmath76 , i.e. @xmath210 [ th:2ml_mbm ] suppose @xmath19 is a multifractional brownian motion with hurst function @xmath140 .
then , with probability one , the 2-microlocal frontier of @xmath19 at any @xmath12 satisfies @xmath211 for all @xmath212
. furthermore , when @xmath213 , the frontier is equal to @xmath214 without any loss of generality , we can assume that @xmath215 , since otherwise the lower bound is equal to @xmath134 . then , let @xmath53 , @xmath121 , @xmath12 and @xmath216 be in the neighbourhood @xmath217 .
we first estimate the size of the increments @xmath218 : @xmath219 using a simple integration by parts , we observe that the first term is equal to @xmath220 owing to lemma [ prop:2ml_fgf ] , for any @xmath53 , there exists @xmath221 such that @xmath222}};\quad { \bigl\lvert \partial_h b(v , h ) - \partial_h b(u , h ) \bigr\rvert } \leq c_1 { \lvertu - v\rvert}^{h-{\varepsilon}}.\ ] ] therefore , the integral term satisfies @xmath223 and is thus negligible in front of the first one . still using lemma [ prop:2ml_fgf ] , we know that for all @xmath124 @xmath224 where @xmath225 and @xmath226 .
this inequality corresponds to the first term in the lower bound . using taylor s formula
, we can obtain an estimate of the second component @xmath227 .
@xmath228 where @xmath229 denotes the coefficient @xmath230 previously introduced .
there exists a neighbourhood @xmath217 of @xmath8 such that for every @xmath231 and for all @xmath232 , @xmath233 .
hence the term @xmath234 is dominant in the sum and for all @xmath124 @xmath235 for any @xmath236 and @xmath237 .
this inequality leads to the second term in the lower bound . finally , let consider the last integral .
@xmath238 can be simply bound by a constant , implying that the integral is upper bounded by @xmath239 .
the third lower bound is a direct consequence of this estimate . to end the proof , we study the case @xmath213 .
the lower bound is clearly equal to @xmath240 .
let set @xmath212 .
suppose first that @xmath241 .
then , the term @xmath242 will dominate in the increment @xmath243 , therefore implying the inequality @xmath244 . in the other case
, we observe that @xmath245 there exists @xmath121 and @xmath246 such that for all @xmath232 , @xmath247 , implying that the regularity of this component is @xmath248 furthermore , as previously noted , the second term is upper - bounded by @xmath249 , and thus is negligible in front of the first one .
therefore , the increment @xmath243 behave similarly to @xmath250 , proving that @xmath251 . as observed in the previous proof ,
the study of the general form of the 2-microlocal frontier is quite technical , in particular because of the interaction which may appear between the oscillations of the hurst function @xmath140 and the derivatives of the fractional brownian field @xmath61 .
nevertheless , a complete estimate can be obtained for the pointwise hlder regularity .
[ prop : pointwise_mbm ] suppose @xmath19 is a multifractional brownian motion with hurst function @xmath140 .
then , with probability one , the pointwise exponent is equal to @xmath252 let us proceed similarly to the proof of theorem [ th:2ml_mbm ] and study the increment @xmath253 . owing to lemma [ prop:2ml_fgf ]
, the pointwise regularity of the second term is exactly @xmath3 .
then , we consider a slightly different taylor expansion on the first increment , @xmath254 for every @xmath231 , there exists a simple upper bound : @xmath255 .
hence , the pointwise regularity of this type of term is strictly greater than @xmath3 , they are therefore negligible . in the case
@xmath256 , we note that there exists @xmath121 and @xmath257 such that for all @xmath258 , @xmath259 .
hence , the pointwise exponent of this component is exactly @xmath260 . finally , similarly to the previous proof
, the last term can be simply upper bounded by @xmath261 .
this last estimate is sufficient to conclude the proof .
the previous result constitutes an extension of the estimate obtained by @xcite .
as pointed out by the latter , it shows that the multifractional brownian motion can have a non - deterministic pointwise regularity .
more precisely , @xcite has proved that for any @xmath53 , with positive probability , @xmath262 } } : m_{t , h(t ) } > 1 \ } } \geq 1 - \inf_{u\in{\ensuremath{[a , b ] } } } h(u ) - { \varepsilon}.\ ] ] in the view of the result obtained in proposition [ prop : pointwise_mbm ] , it would be interesting to study the hausdorff dimension of the following sets @xmath263 } } : m_{t , h(t ) } = k\bigr\}}.\ ] ] this question seems to be much more difficult than the previous estimate , since it involves the successive derivatives @xmath141 of the fractional field and thus the correlation existing between them .
finally , we also note that in terms of regularity , the behaviour at @xmath41 is specific , since the multiplicity @xmath264 is equal to infinity .
to end this section , we present a couple of examples of hurst functions which illustrate this particular geometry of the multifractional brownian motion .
[ ex:2ml_mbm1 ] suppose @xmath265 is an hurst function such that for all @xmath12 , the 2-microlocal local frontier of @xmath140 at @xmath8 is equal to @xmath266 this kind of function can be easily constructed from the sample paths of a fractional brownian motion with parameter @xmath267 .
then , the 2-microlocal frontier of the corresponding multifractional brownian motion satisfies with probability one : * if @xmath268 , @xmath269 * if @xmath270 , @xmath271 these two cases are illustrated by figure [ sfig:2ml_mbm_ex1 ] . to obtain the complete determination of 2-microlocal frontier , we use theorem [ th:2ml_mbm ] , observing that it implies the inequality @xmath272 since @xmath273 . then , the upper bound is a consequence of the pointwise exponent obtained in proposition [ prop : pointwise_mbm ] and the following property of the 2-microlocal frontier : @xmath274 proved in the appendix ( lemma [ lemma:2ml_holder_bound ] ) .
0.49 0.49 in the following second example , we illustrate the particular hlder regularity which may appear at @xmath41 .
[ ex:2ml_mbm2 ] consider the hurst function @xmath275 , which locally corresponds to the `` chirp '' function .
then , the 2-microlocal frontier of the multifractional brownian motion at @xmath134 is equal to @xmath276 figure [ sfig:2ml_mbm_ex2 ] gives an illustration of this particular and hybrid 2-microlocal frontier . to obtain this estimate , we need to refine the proof of theorem [ th:2ml_mbm ] .
let first recall that the 2-microlocal frontier of @xmath140 at @xmath134 is equal to @xmath277 if we consider the regularity of the term @xmath278 , we know there exists for any @xmath53 a sequence @xmath279 such that @xmath280 . then , owing to form of the hurst function , there exists @xmath281 such that @xmath282 where @xmath283 .
combining the two , we obtain that the 2-microlocal frontier of the component @xmath278 is @xmath284 furthermore , we know that @xmath285 , inducing that the other terms in the taylor expansion are negligible .
we note that the 2-microlocal frontier obtained in this example is consistent with the estimate of the hlder exponents proved in proposition [ prop : pointwise_mbm ] .
in this section , we aim to extend our work on the sample path geometry of mbm by studying the fractal dimension of its graph and the images of sets . as previously , we will be particularly interested in the study of the irregular case in order to generalize the classic results of @xcite .
we first need to recall a few notations on fractal dimensions . for any set @xmath286
, we respectively denote by @xmath287 and @xmath288 ( or @xmath289 ) the hausdorff and the lower ( or upper ) box dimensions of @xmath290 . in the case @xmath291 , we denote by @xmath292 the previous quantity .
the reader might refer to the book of @xcite for the precise definition and the basic properties of these fractal dimensions . as we investigate the fractal geometry of the multifractional brownian motion , which is a purely non - stationary process , it is more relevant to consider the localised fractal dimension . for all @xmath286 and @xmath293 ,
it is defined by @xmath294 where @xmath295 denotes any fractal dimension .
the previous limit always exists as the right term is decreasing when @xmath296 .
we note that the map @xmath297 is upper semi - continuous .
finally , we also observe that the fractal dimension can always be retrieve from the localised values : @xmath298 . in the first part of this section
, we investigate the local fractal dimension of the graph of the mbm . for any function @xmath128
, its graph on the set @xmath299 will be denoted @xmath300 , i.e. @xmath301 in the case @xmath302 , we will simply use the notation @xmath303 . to simplify , we will write @xmath304 for the local fractal dimension at @xmath305 .
the latter is therefore equivalently defined by @xmath306 several deterministic properties satisfied by the local graph dimension are presented in the appendix [ sec : mbm_appendix ] , including some connections with the 2-microlocal frontier .
the graph dimension of an @xmath1-multifractional brownian motion has already been determined by @xcite , proving that with probability one @xmath307 in the next result , we investigate the extension of this formula to the box dimension of the graph of a general multifractional brownian motion .
[ th : mbm_dim_box ] suppose @xmath19 is a multifractional brownian motion . then
, with probability one , it satisfies @xmath308 and @xmath309 hence , the graph has a local box dimension at @xmath8 when the previous two quantities are equal .
we first obtain the upper bound for both the lower and upper box dimensions .
let @xmath12 and @xmath121 .
owing to lemma [ prop:2ml_fgf ] , with probability one , there exist @xmath310 and @xmath53 such that for all @xmath124 , @xmath311 let @xmath312 denotes the smallest number of balls of diameter @xmath313 necessary to cover the graph of @xmath140 on the interval @xmath314}}$ ] . up to a constant
, we can assume that this cover is organised on columns of size @xmath313 , as usually presented in the case of a graph covering .
owing to the upper bound of the increment @xmath243 , we know that by adding @xmath315 balls to every column covering , we obtain a covering of the graph of the process @xmath19 . hence , there exists a constant @xmath316 such that @xmath317 and therefore @xmath318 considering the inferior and superior limits of this inequality , we respectively obtain the upper bound of the lower and upper box dimensions .
to prove the lower bound @xmath319 , we use classic potential methods that have already been presented by @xcite and @xcite for the multifractional brownian motion .
briefly , for any interval @xmath320}}\subset{\ensuremath{\mathbf{r}}\xspace}_+$ ] , there exists @xmath120 such that @xmath321}}^2 \geq c { \lvertu - v\rvert}^{2\underline{h}}$ ] , where @xmath322 } } } h(u)$ ] .
this estimate is sufficient , using a standard potential argument to show that almost surely , @xmath323 } } } h(u ) \leq { \ensuremath{{\dim}_{\text{\normalfont\scriptsize h}}}\xspace}\operatorname{gr}(x,{\ensuremath{[a , b ] } } ) \leq { \ensuremath{\underline{\dim}_{\text{\normalfont\scriptsize b}}}\xspace}\operatorname{gr}(x,{\ensuremath{[a , b ] } } ) \leq { \ensuremath{\overline{\dim}_{\text{\normalfont\scriptsize b}}}\xspace}\operatorname{gr}(f,{\ensuremath{[a , b]}}).\ ] ] considering @xmath324 , we obtain the inequality with probability one for any rational interval . then , we note that equation holds as well if we use in the limit rational intervals containing @xmath8 .
hence , owing to the continuity of @xmath140 , with probability one and for all @xmath12 , we get @xmath325 and @xmath326 . to prove the remaining lower bound , let us first consider the case @xmath213 .
the constant @xmath9 can be chosen small enough such that for all @xmath327 in neighbourhood of @xmath328 , @xmath329 , where @xmath330 is a positive constant .
furthermore , for all @xmath124 , @xmath331 hence , in the neighbourhood of @xmath8 , @xmath332 using the covering argument described at the beginning of the proof , we obtain that the inequality @xmath333 and @xmath334 . since we already know that the lower and upper box dimensions are greater than @xmath319 , it induces the expected lower bound .
finally , let us extend the inequality to any @xmath5 and any value of @xmath335 .
we have already noted that the function @xmath336 is upper semi - continuous for every dimension considered . as a consequence , there exists a deterministic countable set @xmath337 such that for all @xmath8 , there is a sequence @xmath338 satisfying @xmath339 since for every @xmath340 , @xmath341 is a centered gaussian variable with a positive variance , with probability one and for all @xmath340 , @xmath342 .
hence , as the local dimension is upper semi - continuous , @xmath343 the argument holds as well for the upper box dimension , therefore uniformly proving the result .
interestingly , we have obtained an expression of the box dimension of the graph which is similar to the behaviour observed on a fractional brownian motion with variable drift ( see the recent works @xcite ) .
this common property happens to extend as well to the hausdorff dimension of the graph , recently obtained by @xcite on the fbm with variable drift .
the hausdorff dimension of the graph is a bit more complicate to analyse . for this purpose
, we have to introduce a more general dimension called the _
parabolic hausdorff dimension_. we first need to define a _
parabolic metric _
@xmath344 on @xmath345 , with @xmath346 : @xmath347 for any set @xmath348 , we denote by @xmath349 the _ parabolic hausdorff dimension _ of @xmath290 .
it is defined similarly to the classic hausdorff dimension using covering balls relatively to the metric @xmath344 , i.e. it corresponds to the infimum of @xmath350 for which @xmath351 note that given the form of the parabolic metric , a @xmath344-ball @xmath352 has the following form : @xmath353}}\times{\ensuremath{[x-\delta , x+\delta]}}$ ] .
the parabolic hausdorff dimension has first been considered by @xcite for the study of polar sets of the heat equation .
recently , it has been proved to be useful by @xcite to analyse the geometry of the images of brownian motion . finally , as previously outlined , @xcite have shown it is also a natural idea for the study of fractional brownian motion with variable drift . note
that the convention used in the latter work is slightly different from ours .
[ th : mbm_dim_haus ] suppose @xmath19 is a multifractional brownian motion . then
, with probability one , it satisfies @xmath354 to begin with , we prove that with probability one , @xmath355 the sketch of the proof of this equality is similar to theorem [ th : mbm_dim_box ] , as it uses as well an ad - hoc covering argument .
let us set @xmath356 , @xmath121 and @xmath357 .
suppose @xmath358 is a @xmath313-cover of @xmath359}})$ ] with of @xmath360-balls and such that @xmath361 we observe that @xmath362 for all @xmath124 .
for any @xmath363 , we need at most @xmath364 @xmath360-balls of diameter @xmath365 to cover the part of the graph related to the term @xmath366 .
hence , from @xmath358 , we construct a @xmath313-cover @xmath367 of @xmath368}})$ ] which satisfies @xmath369 the last series converges when @xmath370 .
hence , considering the limits @xmath55 , @xmath296 and @xmath371 , we obtain @xmath372 .
to prove the other side inequality , we proceed similarly to the proof of theorem [ th : mbm_dim_box ] .
we first assume that @xmath213 . in this case , the estimate of the increments presented in theorem [ th : mbm_dim_box ] and the previous reasoning similarly imply that @xmath373 .
then , we use the upper semi - continuity of the function @xmath374 , obtained in lemma [ lemma : dim_para_usc ] , to extend the inequality to any @xmath5 .
the upper bound of @xmath375 is then a consequence of the deterministic lemma [ lemma : dim_para_bounds ] , with @xmath376 and @xmath377 . with probability one , for all @xmath12
, we have @xmath378 } } ) \leq 1 + h(t){\bigl ( { \ensuremath{{\dim}_{\text{\normalfont\scriptsize h}}}\xspace}{\bigl ( \operatorname{gr}(x,{\ensuremath{[t-\rho , t+\rho]}})\,;\varrho_{h(t)}\bigr ) } - 1 \bigr)}$ ] . therefore ,
when @xmath296 , we obtain the first upper bound with probability one : for all @xmath5 , @xmath379 . to prove the lower bound , we make use of potential arguments on the parabolic and euclidean metrics .
we first set @xmath5 .
then , let @xmath121 and @xmath380 .
using frostman s lemma applied on the parabolic metric @xmath381 , there exists a probability measure @xmath382 on @xmath217 such that @xmath383 suppose @xmath384 denotes the image of the measure @xmath382 by the map @xmath385 . to prove the lower on the dimension of the graph of @xmath19
, we need to study @xmath386-energy of @xmath384 , i.e. @xmath387 where @xmath388 .
let us set @xmath53 .
to study the previous energy , we need to split the integral into two components , respectively denoted @xmath389 and @xmath390 : @xmath391 let us prove @xmath389 is finite . for this purpose
, we estimate its expectation @xmath392 } } = \iint_{b(t,\rho)^2 } { \mathbb{e}{\bigl [ { \bigl ( { \lvertu - v\rvert}^{2 } + { \lvert x_u - x_v \rvert}^2 \bigr)}^{-s/2 } \bigr]}}{\ensuremath{\mathbf{1}}\xspace}_{{\bigl\{{\lverth(u)-h(v)\rvert}\leq{\lvertu - v\rvert}^{h(t)(1-{\varepsilon})}\bigr\ } } } \mu({\ensuremath{\mathrm d}\xspace}u)\,\mu({\ensuremath{\mathrm d}\xspace}v ) .
\end{aligned}\ ] ] using the notation @xmath393}}^2 $ ] , the inner term is equal to @xmath394 } } & = \frac{2}{\sigma_{u , v}\sqrt{2\pi } } \int_{{\ensuremath{\mathbf{r}}\xspace}_+ } { \bigl ( x^2 + { \lvertu - v\rvert}^2 \bigr)}^{-s/2 } \exp{\biggl ( -\frac{x^2}{2\sigma^2_{u , v } } \biggr ) } \,{\ensuremath{\mathrm d}\xspace}x \\ & = c_1 \int_{{\ensuremath{\mathbf{r}}\xspace}_+ } { \bigl ( r \sigma_{u , v}^2 + { \lvertu - v\rvert}^2 \bigr)}^{-s/2 } r^{-1/2 } { \ensuremath{\mathrm{e}}\xspace}^{-r/2 } \,{\ensuremath{\mathrm d}\xspace}r .
\end{aligned}\ ] ] splitting the integral into two terms , and using classic approximations ( see e.g. @xcite ) , the previous term is upper bounded by @xmath395 we recall that @xcite has proved that @xmath396}}^2 \asymp { \lvertu - v\rvert}^{2h(t ) } + { \lverth(u)-h(v)\rvert}^2\ ] ] in the neighbourhood of @xmath8 .
hence , on the domain @xmath397 , we observe that @xmath398 the expectation of @xmath389 is therefore upper bounded by @xmath392 } } & \leq c_4 \iint_{b(t,\rho)^2 } \sigma_{u , v}^{-1+(1-s)/h(t)(1-{\varepsilon } ) } \,\mu({\ensuremath{\mathrm d}\xspace}u)\,\mu({\ensuremath{\mathrm d}\xspace}v ) \\ & \leq c_5 \iint_{b(t,\rho)^2 } { \bigl ( { \lvertu - v\rvert}^{h(t ) } + { \lvert h(u)-h(v ) \rvert } \bigr)}^{-1+(1-s)/h(t)(1-{\varepsilon})}\,\mu({\ensuremath{\mathrm d}\xspace}u)\,\mu({\ensuremath{\mathrm d}\xspace}v ) .
\end{aligned}\ ] ] the term @xmath399}}$ ] is thus finite when @xmath400 , i.e. when @xmath401 hence , considering the limits on @xmath402 and @xmath403 , the upper bound obtained on @xmath386 corresponds to the expected estimate .
let now consider the second term @xmath390 .
we know that @xmath404 almost surely .
recall that the increment @xmath218 has the following form @xmath405 if @xmath9 is sufficiently small , there exists a positive constant @xmath221 such that for all @xmath406 and any @xmath407}}$ ] , we have @xmath408 .
furthermore , there also exists @xmath409 such that @xmath410 using the previous estimates , @xmath411 hence , on the domain @xmath412 , we obtain @xmath413 when @xmath9 is sufficiently small .
therefore , the energy term @xmath390 satisfies @xmath414 the last integral is finite when @xmath415 .
but , since @xmath416 , we remark that @xmath417 and thus @xmath418 , showing that @xmath390 is always finite when @xmath419 .
therefore , we have proved that the @xmath386-energy @xmath420 is almost surely finite for all @xmath421 .
considering the limits @xmath55 and @xmath422 , we obtain @xmath423 the lower bound has been obtained for a fixed @xmath5 . to extend the latter uniformly on @xmath424
, we proceed similarly to theorem [ th : mbm_dim_box ] . shortly ,
as proved in lemma [ lemma : dim_para_usc ] , the map @xmath374 is upper semi - continuous .
hence , there exists a countable set @xmath425 such that for any @xmath8 , there is sequence @xmath338 satisfying @xmath426 the equality on the dimension of the graph holds almost surely for all @xmath340 .
therefore , with probability one , for all @xmath427 , @xmath428 this last inequality concludes the proof , since the upper bound is already uniform on @xmath429 .
note that theorem [ th : mbm_dim_haus ] does not contradict the result obtained by @xcite on @xmath1-mbms . indeed , under this assumption @xmath1 and due to the form of the parabolic metric @xmath360 , covering the graph of @xmath140 in the neighbourhood of @xmath8 is equivalent to covering a straight line , implying that the local dimension satisfies : @xmath430 .
hence , the application of theorem [ th : mbm_dim_haus ] proves that with probability one @xmath431 where @xmath19 is an @xmath1-multifractional brownian motion .
it is interesting to observe that the hausdorff and box dimensions of the graph of mbm and fbm with variable drift have a similar form despite the rather different gaussian structures of these two processes .
indeed , the first one is a centered process whose incremental variance satisfies @xmath432}}^2 \asymp { \lvertu - v\rvert}^{2h(t ) } + { \lverth(u)-h(v)\rvert}^2,\ ] ] whereas the latter still has the covariance of fbm , but a mean equal to the drift @xmath433 . as a consequence of these similarities , the examples constructed by @xcite can also be applied to the mbm , proving that the graph of the latter may have different hausdorff and box dimensions .
[ ex : mbm_gr_dim1 ] @xcite have computed the parabolic hausdorff dimension of the celebrated deterministic construction of @xcite .
hence , there exists a continuous function @xmath434 such that for all @xmath12 , @xmath435 , @xmath436 then , considering the mbm @xmath19 with hurst function @xmath140 , we first observe that with probability one , for all @xmath16 , @xmath437 .
furthermore , as a consequence of the properties of @xmath140 , at every @xmath16 such that @xmath438 , @xmath439 hence , at these particular times , the local hausdorff dimension is strictly smaller that the box one .
in addition , as shown by @xcite , the local hausdorff dimension is such that @xmath440 , meaning the simple and straightforward lower bound on the hausdorff dimension is not optimal .
this example illustrates the kind of unusual fractal geometry the mbm can present .
it clearly shows that the parabolic hausdorff dimension is the relevant concept , as in this case , @xmath441 happens to be insufficient to characterise entirely the hausdorff dimension of the graph .
we also note that in contrary to fbm , the fractal dimension of the graph can be completely unrelated to the local hlder exponent . as presented in the following second example
, more classic sample path properties also occur .
[ ex : mbm_gr_dim2 ] suppose @xmath140 is a sample path of a fractional brownian motion with hurst exponent @xmath129 .
up to a rescaling , it is assumed to be valued in the interval @xmath442 .
then , using some classic estimates ( see e.g. @xcite ) , we easily obtain that @xmath443 then , owing to theorem [ th : mbm_dim_haus ] , a mbm @xmath19 with hurst function @xmath140 satisfies with probability one : @xmath444 this example displays a more common fractal geometry , where the hausdorff and box dimensions are both equal to @xmath445 .
nevertheless , the local hlder exponent may still have an irregular behaviour , depending on the value of @xmath3 .
in the second part of this section , we are interested in studying the fractal dimension of an image @xmath446 , where @xmath447 can be any fractal set .
a large literature has already investigated the question of the geometry of images of gaussian processes .
it is a well - known result ( see e.g. @xcite ) that for any fixed borel set @xmath448 , @xmath449 @xcite have generalized the previous result to @xmath450-(fractional ) brownian sheet , and in particular , obtained uniform results on @xmath447 under some assumption on the dimensions @xmath451 and @xmath452 . adapting classic arguments from @xcite
, we can easily extend equation to an @xmath1-mbm , i.e. for any fixed borel set @xmath448 and all @xmath453 @xmath454 in the irregular case , the statement is not as simple and straightforward . furthermore , the _ parabolic hausdorff dimension
_ again happens to be the relevant tool to describe the geometry of @xmath446 .
[ th : mbm_dim_images ] suppose @xmath19 is a multifractional brownian motion .
then , with probability one , @xmath455 the proof of this result is similar to the work of @xcite on the drifted fbm .
let us first remark that @xmath456 this inequality is a simple consequence of the property of the hausdorff under the application of lipschitz function ( here , the projection @xmath457 ) . then , using lemma 2.1 from @xcite , we obtain @xmath458 , thus proving the upper bound . to obtain the lower bound , we use a classic potential argument . owing to frostman s lemma , for any @xmath121 and any @xmath459
, there exists a probability measure @xmath382 on @xmath217 such that @xmath460 recall , as proved by @xcite the covariance of the mbm has the following form @xmath461}}^2 \asymp { \lvertu - v\rvert}^{2h(t ) } + { \lverth(u)-h(v)\rvert}^2\ ] ] suppose @xmath384 denotes the image of the measure @xmath382 under the map @xmath462 . then , for any @xmath463 , the expectation of its @xmath386-energy is equal to @xmath464 } } & = \iint_{{\ensuremath{\mathbf{r}}\xspace}^2 } { \mathbb{e}{\bigl[\hspace{1pt } { \lvert x_u - x_v \rvert}^{-s } \hspace{1pt}\bigr ] } } \mu({\ensuremath{\mathrm d}\xspace}u)\,\mu({\ensuremath{\mathrm d}\xspace}v ) \\ & = c_2 \iint_{{\ensuremath{\mathbf{r}}\xspace}^2 } \sigma_{u , v}^{-2s } \,\mu({\ensuremath{\mathrm d}\xspace}u)\,\mu({\ensuremath{\mathrm d}\xspace}v ) \\
\end{aligned}\ ] ] therefore , owing to the form of the covariance , the integral is finite when @xmath465 .
hence , considering the limit @xmath296 , we obtain @xmath466 , which concludes the proof .
the local parabolic hausdorff dimension @xmath467 might not seem to be a very intuitive notion at first . to have a better intuition of the result described in theorem [ th : mbm_dim_images ] ,
we may first note , similarly to theorem [ th : mbm_dim_haus ] , it does not contradict equation in the case of an @xmath1-multifractional brownian motion . indeed , when @xmath468 , due to the property of the parabolic metric @xmath360 , it is completely equivalent to cover @xmath447 and @xmath469 in the neighbourhood of @xmath8 .
hence , in this case @xmath470 the next example shows that different behaviours may occur when the mbm is irregular .
[ ex : mbm_dim_images ] let @xmath140 be a trajectory of fractional brownian motion defined on the interval @xmath471}}$ ] , with hurst index @xmath159 .
up to a rescaling and a translation , the sample path satisfies @xmath472}})\subset{\ensuremath{(0,1)}}$ ] .
we first assume that @xmath473 .
owing to equation , there exists a borel set @xmath447 such that @xmath474 .
furthermore , since the proof of this statement is based on frostman s lemma , there exists a probability measure on @xmath475 such that @xmath476 where @xmath477 .
therefore , still using frostman s lemma , we obtain that @xmath478 .
the other side inequality is simply obtained by observing that @xmath479 , inducing the upper bound using classic covering arguments .
hence , we obtain @xmath480 for the second example , we consider the set @xmath481 , which might be assumed to be non - empty . @xcite have proved that @xmath482 . in this case , as the hurst function @xmath140 is constant on the set @xmath447 , we easily prove that the dimension of @xmath469 is characterised by the dimension on @xmath447 .
therefore , @xmath483 the two previous examples show that the estimate obtained in theorem [ th : mbm_dim_images ] can lead to very different results , depending on the properties of the set @xmath447 chosen . in particular , we observe that the hausdorff geometry of images displays a richer structure than the local hlder regularity . indeed , in the irregular case , we know that the later only depends on @xmath110 sample path properties , whereas equations and show a more complex geometry of the image @xmath446 . in the first case , we observe that the formula depends only on the local exponent of @xmath140 and the hausdorff dimension of @xmath447 , whereas in the second one , equation presents a case where the hausdorff dimension follows the classic formula , even though the mbm is irregular at @xmath8 . to conclude this section , we discuss the question of obtaining the hausdorff dimension of the level sets @xmath484 .
this problem has already been investigated by @xcite for the @xmath1-multifractional brownian motion , proving in particular that for all @xmath5 , @xmath485 it is quite natural to wonder if the previous result can be extended to the irregular mbm .
the upper bound can easily extended , since it is a direct consequence of the hlder regularity of the process at @xmath8 .
in addition , we note that the proof from @xcite of the lower bound stands as well in the irregular case . as a consequence ,
we easily obtain bounds for all @xmath5 , @xmath486 lower and upper bounds are not equal in the irregular case , and the remaining question is whether the lower one can be improved .
the sketch of the proof of this inequality is rather classic and relies on the study of the hlder continuity of the local time ( see e.g. the seminal works of @xcite and @xcite on the subject ) .
the key point to prove the hlder regularity is a property called the _ local nondeterminism ( lnd)_. in the case of the mbm , it states that for every @xmath487 , there exists @xmath488 such that for all @xmath489 @xmath490 to obtain a better lower bound for the local dimension of the level set , we would need to improve the exponent in the right term .
unfortunately , the next lemma proves that it is optimal under a mild assumption on the hurst function .
let @xmath12 and suppose there exists a sequence @xmath491 such that @xmath492 .
then , for any @xmath487 , any @xmath53 and any @xmath488 , there exists @xmath493 such that @xmath494 for all @xmath493 and any @xmath495 , we note that @xmath496}}$ ] . using elements @xmath497 and the estimate of the variance of the increments , we obtain that @xmath498 let set @xmath121 .
we may assume that @xmath499 for every @xmath500 , implying that @xmath501 using @xmath9 sufficiently small , it proves that the property of local nondeterminism does not stand with the exponent @xmath502 , for any @xmath53 .
hence , the classic exponent @xmath503 in the lnd property of the mbm is most of the time optimal .
intuitively , it can be understood as following : in the neighbourhood of @xmath8 , the hurst function can influence the hlder regularity , but is still a deterministic component and thus does not bring any randomness which could increase the conditional variance .
as the consequence , the latter is only a result of the local fbm form of the multifractional brownian motion .
therefore , the estimate of the hausdorff dimension of the level sets can not be improved using the classic methods based on the lnd property , and it remains an open question to determine the precise form of the fractal dimension in terms of the geometric properties of the hurst function @xmath140 .
we gather in the appendix a few deterministic properties related to 2-microlocal analysis and local graph dimensions and which have been used through the paper .
we begin with a simple upper bound on the 2-microlocal which extends the classic property of lower semi - continuity on the local hlder exponent .
let us set @xmath121 and @xmath53 .
there exists @xmath406 such that @xmath506 .
then , we can find @xmath507 such that @xmath508 therefore , for any @xmath509 , we have @xmath510 proving that @xmath511 , for any @xmath53 .
we need to construct a local cover of the graph which is sufficiently efficient .
let first set @xmath12 and @xmath514 .
then , there exists @xmath121 and @xmath120 such that for all @xmath124 , @xmath515 the cover is constructed similarly to the original proof .
hence , if @xmath516 denotes the number of @xmath313-boxes necessary to cover @xmath517}})$ ] , we have @xmath518 } } } { \lvertf(u)-f(v)\rvert } \cdot \delta^{-1 } \\ & \leq c \delta^{-1 } + \sum_{k=-{\lceil\rho/\delta\rceil}}^{{\lceil\rho/\delta\rceil } } c \delta^\sigma ( k\delta)^{-1}\cdot \delta^{-1 } \\ & = c \delta^{-1 } + 2c \delta^{\sigma-2 } \sum_{k=1}^{{\lceil\rho/\delta\rceil } } k^{-1 } \leq 2c \,\delta^{\sigma-2 } { \bigl ( 1 + \log{\lceil\rho/\delta\rceil } \bigr)}. \end{aligned}\ ] ] since @xmath519 .
then , @xmath520 } } ) = \limsup_{\delta\rightarrow 0 } \frac{n_\delta}{-\log\delta}\leq 2-\sigma,\ ] ] for all @xmath521 , which proves to the expected inequality . as the estimate we want to obtain differs from the previous lemma
, we need to adopt a constructing procedure which is slightly different .
let @xmath12 and @xmath523 .
then , there exist @xmath524 , @xmath121 and @xmath120 such that for all @xmath124 @xmath525 let us set @xmath356 and @xmath526 .
the @xmath313-cover of @xmath517}})$ ] is constructed as following .
there exists @xmath487 such that for all @xmath527 , @xmath528 . to cover the set @xmath314}}$ ] ,
we first choose successive intervals @xmath529 of size @xmath530 in the neighbourhood of @xmath8 .
the rest is simply covered by intervals of size @xmath313 .
the first type of intervals cover a set whose length is equal to : @xmath531 hence , the number @xmath532 of intervals of size @xmath313 which are necessary is upper bounded by @xmath533 let now consider the cover @xmath358 of @xmath517}})$ ] which can be defined from the previous intervals ( still using the same idea of covering by columns ) .
it satisfies for any @xmath534 @xmath535 let us estimate the first term , @xmath536 where we have used that @xmath537 .
the previous series converges if @xmath386 satisfies @xmath538 , i.e. if @xmath539 .
let now consider the second term , @xmath540 the sum can upper bounded by @xmath541 combining the previous estimate , we obtain @xmath542 hence , the latter converges as @xmath543 when @xmath544 .
therefore , combining the two conditions we have obtained , the hausdorff dimension satisfies @xmath545 } } ) \leq \min{\bigl\ { 1 -\sigma + s'(1-\gamma^{-1 } ) + \gamma^{-1 } , 2-\sigma+(\gamma-1)(s'-1)/\gamma \bigr\}},\ ] ] for all @xmath416 and @xmath509 .
considering the limit @xmath546 , we obtain @xmath547 } } ) \leq 2 - \sigma$ ] , which implies the expected inequality .
we note that the two new upper bounds presented in the previous lemmas are consistent with the original one , since @xmath548 .
furthermore , it proves to be the optimal ones for a classic deterministic function called the _ chirp function _ : @xmath549 , with @xmath550 . as proved by @xcite ,
its 2-microlocal frontier at @xmath134 is equal to @xmath551 it is an easy calculus to check that the local hausdorff and box dimensions at zero are given by @xmath552 hence , they are equal to the upper bounds presented in lemmas [ lemma:2ml_dimb ] and [ lemma:2ml_dimh ] , on the contrary to the usual hlder estimate @xmath553 which is not optimal .
suppose @xmath348 and @xmath558 .
let us first prove the upper bound on @xmath559 .
for any @xmath560 , there exists a @xmath561-cover @xmath358 of @xmath290 such that @xmath562 we denote by @xmath563 the quantity @xmath564 . without any loss of generality , we may assume that for every @xmath363 , @xmath565 is a rectangle of size @xmath566 ( as a simple consequence of the definition of @xmath561 ) .
we want to construct a @xmath567-cover of @xmath290 from @xmath358 .
since @xmath568 , we need to split @xmath565 along the space axis to obtain @xmath567-balls .
more precisely , there will be at most @xmath569 resulting @xmath567-balls of diameter @xmath563 .
hence , we obtain a @xmath567-cover @xmath367 of @xmath290 such that @xmath570 the last sum is finite when @xmath571 , i.e. when @xmath572 . considering @xmath573
, we obtain the expected upper bound .
the lower bound is obtained using a similar reasoning .
for any @xmath574 , there exists a @xmath567-cover @xmath367 of @xmath290 such that @xmath575 similarly , we may assume that @xmath576 is a rectangle of size @xmath577 for every @xmath142 . as @xmath558 , we note that @xmath578 .
hence , to obtain a @xmath561-cover of @xmath290 , we split the previous balls along the time axis , in at most @xmath579 sub - balls .
therefore , we obtain a @xmath561-cover @xmath358 of @xmath290 such that @xmath580 this sum is finite when @xmath581 , inducing the lower bound on @xmath559 . for any @xmath12 ,
the local parabolic hausdorff dimension satisfies @xmath585 this inequality is straightforward extension of the usual one on the hausdorff dimension .
let @xmath80 .
owing to the uniform bounds obtained in lemma [ lemma : dim_para_bounds ] , there exists @xmath586 such that for all @xmath587 and all @xmath121 , @xmath588 hence , considering the limit @xmath296 , we obtain @xmath589 , proving , jointly with the first inequality , the upper semi - continuity of the map @xmath583 .
a. ayache , s. cohen , and j. lvy vhel .
the covariance structure of multifractional brownian motion , with application to long range dependence . in _ proceedings of the acoustics , speech , and signal processing , 2000 _ , volume 6 , pages 38103813 , washington , dc , usa , 2000 .
ieee computer society .
second microlocalization and propagation of singularities for semilinear hyperbolic equations . in _ hyperbolic equations and related topics ( katata / kyoto , 1984 ) _ , pages 1149 . academic press , boston , ma , 1986 .
d. monrad and l. d. pitt . local nondeterminism and hausdorff dimension . in _
seminar on stochastic processes , 1986 ( charlottesville , va . , 1986 )
_ , volume 13 of _ progr .
_ , pages 163189 .
birkhuser boston , boston , ma , 1987 .
|
the geometry of the multifractional brownian motion ( mbm ) is known to present a complex and surprising form when the hurst function is greatly irregular .
nevertheless , most of the literature devoted to the subject considers sufficiently smooth cases which lead to sample paths locally similar to a fractional brownian motion ( fbm ) .
the main goal of this paper is therefore to extend these results to a more general frame and consider any type of continuous hurst function .
more specifically , we mainly focus on obtaining a complete characterization of the pointwise hlder regularity of the sample paths , and the box and hausdorff dimensions of the graph
. these results , which are somehow unusual for a gaussian process , are illustrated by several examples , presenting in this way different aspects of the geometry of the mbm with irregular hurst functions .
|
our starting point is that the key question in the physics of unstable nuclei is the knowledge of the @xmath0 for asymmetric nuclear matter away from normal conditions .
we remark the effect of the symmetry term at low densities on the neutron skin structure , while the knowledge in high densities region is crucial for supernovae dynamics and neutron star cooling @xcite .
effective interactions are obviously tuned to symmetry properties around normal conditions and any extrapolation can be quite dangerous .
microscopic approaches based on realistic @xmath3 interactions , brueckner or variational schemes , or on effective field theories show a quite large variety of predictions , see fig.1 . in the reaction dynamics with intermediate energy radioactive beams we can probe highly asymmetric nuclear matter in compressed as well as dilute phases : the aim of this paper is to show that fragmentation events have new features due to isospin effects and that some observables are particularly sensitive to the symmetry term of the @xmath0 . _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ * fig.1 - * @xmath0 for various effective forces , @xmath4 @xcite(dashed ) , @xmath5 @xcite(squares ) and @xmath6 @xcite(crosses ) .
top : neutron matter ( up ) , symmetric matter ( down ) ; bottom : potential symmetry term .
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ for fragmentation mechanisms a new qualitative feature
is expected : the onset of chemical effects on spinodal instabilities @xcite , due to the intrinsic coupling between isoscalar and isovector modes .
the formation of a very neutron rich gas vs. an almost symmetric liquid phase is expected for neutron excess systems , in a dynamical non - equilibrium mechanism on short time scales @xcite .
this effect , which appears very much reduced in a statistical multifragmentation calculation @xcite , will show up in a production of more stable primary intermediate mass fragments and in enhanced yields for neutron rich light isobars .
all these predictions were actually based on linear response approaches @xcite , nuclear matter dynamics in a box @xcite or on the evolution of a suitably prepared excited source @xcite . in this paper
we will present the first results of a fully `` ab initio '' calculation of fragmentation reactions for systems with different charge asymmetry .
a new code for the solution of microscopic transport equations , the _ stochastic iso - bnv _ , has been written where asymmetry effects are suitably accounted for and the dynamics of fluctuations is included @xcite .
a density dependent symmetry term is used also in the ground state construction of the initial conditions .
isospin effects on nucleon cross section @xcite and pauli blocking are consistently evaluated . in order to simplify the analysis of the most sensitive observables to isospin effects we have chosen a skyrme force with the same @xmath7 for symmetric nuclear matter ( @xmath8 ) ( @xmath9 ) and with two different choices for the density dependence of the symmetry term ,
see fig.1(bottom ) , _ asy - stiff _ ( like in @xmath6 , see also @xcite ) and _ asy - soft _ ( like in @xmath4 ) . in this way we force the symmetric part of the @xmath0 to be exactly the same in order to disentangle dynamical symmetry term effects .
we will show that the reaction mechanism is sensitive to the different behaviours , although with stable nuclei we will limit the possible asymmetries and moreover we will certainly not reach high compression regions .
we have studied the @xmath2 collisions of the systems @xmath10 and @xmath11 , where new data are under analysis at @xmath12 @xcite .
we will comment @xmath13 events generated in semi - central ( @xmath14 ) and semi - peripheral ( @xmath15 ) reactions . in fig.2 we show a typical impact parameter evolution of the density plot ( projected on the reaction plane ) for one event ( neutron rich case , asy - stiff @xmath0 ) .
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ * fig.2 - * @xmath10 @xmath2 : time evolution of the nucleon density projected on the reaction plane .
first two columns : @xmath14 collision , approaching , compression and separation phases .
third column : @xmath16 , fourth column : @xmath15 , separation phase up to the _ freeze - out_. _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ we remark : i ) in the cluster formation we see a quite clear transition from _ bulk _ @xcite to _ neck _
@xcite instabilities .
ii ) the _ `` freeze - out times '' _ , when the nuclear interaction among clusters disappears , are decreasing with impact parameter .
these two dynamical effects will influence the isospin content of the produced primary fragments , as shown later .
a detailed analysis of the results from @xmath13 events for the same system ( _ asy - stiff _ calculation ) is shown in fig.3 ( central , @xmath14 ) and fig.4 ( peripheral , @xmath15 ) .
each figure is organized in this way : top row , time evolution of : ( a ) _ mass _ in the liquid ( up ) and gas ( down ) phase ; ( b ) _ asymmetry _ @xmath17 in the gas `` central '' ( solid line and squares ) , gas total ( dashed+squares ) , liquid `` central '' ( solid+circles ) and clusters ( stars ) .
`` central '' means in a cubic box of side @xmath18 around the c.m .. the horizontal line shows the initial average asymmetry ; ( c ) _ mean fragment multiplicity _ @xmath19 .
the saturation of this curve defines the freeze - out configuration , as we can also check from the density plots like in fig.2 .
bottom row , properties of the `` primary '' fragments in the _ freeze - out configuration _ : ( d ) _ charge distribution _ , ( e ) _ asymmetry distribution _ and ( f ) _ fragment multiplicity distribution _ ( normalized to @xmath20 ) .
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ * fig.3 - * @xmath10 @xmath2 @xmath14 collisions : time evolution and freeze - out properties .
_ asy - stiff _ @xmath0 . _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ * fig.4 - * like fig.3 for @xmath15 .
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ we see a neutron dominated prompt particle emission and a second _ neutron burst _ at the time of fragment formation in the `` central region '' .
the latter is consistent with the dynamical spinodal mechanism in dilute asymmetric nuclear matter , as discussed before .
the effect is quite reduced for semi - peripheral collisions ( compare fig.s 3b and 4b ) and the @xmath21 s ( @xmath22 ) produced in the neck are more neutron rich ( fig.s 3e and 4e ) .
this seems to indicate a different nature of the fragmentation mechanism in central and neck regions , i.e. a transition from volume to shape instabilities with different isospin dynamics . in more peripheral collisions
the interaction time scale is also very reduced ( fig.4c ) and this will quench the isospin migration . _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ * fig.5 - * @xmath10 @xmath2 @xmath14 collisions : time evolution and freeze - out properties .
_ asy - soft _ @xmath0 . _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ * fig.6 - * like fig.5 for @xmath15 . _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ in fig.s 5,6 we have the corresponding results for the _ asy - soft _ @xmath0 . the main qualitative difference is a larger prompt neutron emission ( compare the gas asymmetry in fig.s 3b,4b with the corresponding fig.s 5b,6b ) joined to a `` slower '' dynamics ( see the ( c)-plots ) , as expected from the more attractive nature of the asymmetric @xmath0 @xcite .
the interesting point is that the effect on fragment production is different for central and peripheral events . with respect to the _ asy - stiff _ case
we have a smaller mean @xmath21 multiplicity for central collisions ( fig.s 3f vs. 5f ) and more fragments produced in the neck region ( fig.s 4f vs. 6f ) .
neck instabilities have also the feature of forming clusters on the `` spectator '' side @xcite leading to a `` fission - like '' splitting of the projectile-/target - like fragment .
this mechanism has been clearly observed in accurate kinematical selections @xcite . in our study of n - rich systems
the rate of such dynamical fission processes is systematically larger for the asy - soft @xmath0 , as expected from the previous discussion .
this seems to be an quite sensitive observable to look at .
the most important qualitative difference of the neutron poor case , @xmath11 , is a larger prompt proton emission , in particular for central collisions . the number of protons available to produce clusters is then reduced with a related smaller mean @xmath21 multiplicity .
moreover the reduction of the overall asymmetry is acting in the same direction .
in fig.7 we show the average @xmath21 multiplicity vs. the average charge of the heaviest produced fragment , which measures the centrality of the collision , for the two symmetry terms and for the two systems .
the effect , in agreement with recent data @xcite , is more evident with the _ asy - stiff _ choice , in particular for central collisions ( low @xmath23 ) . _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ * fig.7 - * correlation between mean @xmath21 multiplicity and charge of the heaviest fragment : _ squares _ n - rich `` asy - stiff '' ; _ diamonds _ n - poor `` asy - stiff '' ; _ circles _ n - rich `` asy - soft '' ; _ stars _ n - poor `` asy - soft '' . _
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
starting from simulations of reaction dynamics performed with a new transport code where isospin and fluctuation effects are suitably accounted for , we have shown that dissipative heavy ion collisions , and in particular fragment production reactions , at medium energies are rather sensitive to the density dependence of the symmetry contribution to the nuclear equation of state .
interesting differences appear between isospin effects on central and neck fragmentation .
an indication for a `` stiff '' symmetry term is emerging , consistent with transverse flow calculations in the same energy range @xcite .
the effects are not large but quite encouraging for similar studies with radioactive beams .
a possibility is emerging of obtaining in terrestrial accelerator laboratories important information on the symmetry term of large astrophysical interest .
it appears essential to have good charge asymmetric beams available at intermediate energies and to perform more exclusive experiments .
i. bombaci , t.t.s . kuo and u. lombardo , phys .
reports . 242 ( 1994 ) 165 .
+ i. bombaci , phys .
c55 ( 1997 ) 1 + m.prakash et al . , phys.rep.280 ( 1997 ) 1 i.m.irvine , `` neutron stars '' oxford univ . press 1978 d.pines ,
r.tamagaki and s.tsuruta ed.s , `` neutron stars '' , addison - wesley n.y.1992 `` it is important to put the properties of _ neutron _ matter at _ subnuclear _ densities on as firm a footing as possible , not only for astrophysical applications , but also for interpreting terrestrial experiments with coming radioactive beam facilities '' c.j .
pethick and d.g .
ravenhall , in `` the lives of neutron stars '' , nato asi series c , vol.450 , 1993 .
v.a.urpin , same proceedings , pp.163 - 166 h.krivine , j.treiner and o.bohigas , nucl.phys . *
a336 * ( 1980 ) 155 e.chabanat , p.bonche , p.haensel , j.meyer , r.schaeffer , nucl.phys .
* a627 * ( 1997 ) 710 h.mueller and b.d.serot , phys.rev . *
c52 * ( 1995 ) 2072 bao - an li and c.m.ko , nucl.phys .
* a618 * ( 1997 ) 498 m.colonna , m.di toro and a.larionov , phys.lett . * b428 * ( 1998 ) 1 .
v.baran , a.larionov , m.colonna and m.di toro , nucl.phys . * a632 * ( 1998 ) 287 m.di toro et al .
, bormio winter meeting on nuclear physics , ed.i.iori , univ.milano 1998 p.319 , + progr.part.nucl.phys .
* 42 * ( 1999 ) 125 - 136 a.b.larionov , a.s.botvina , m.colonna and m.di toro , nucl.phys . * a658 * ( 1999 ) 375 m.colonna , m.ditoro , g.fabbri and s.maccarone , phys.rev .
* c57 * ( 1998 ) 1410 m.colonna , m.di toro , a.guarnera , s.maccarone , m.zielinska- pfabe , h.h.wolter , nucl.phys . *
a642 * ( 1998 ) 449 g.q.li and r.machleidt , phys.rev .
c48 ( 1993 ) 1702 and phys.rev .
c49 ( 1994 ) 566 we remark here that also all relativistic mean field approaches in the presently used approximation scheme are predicting a _ stiff - like _ simmetry term , proportional to the barion density , coming from the @xmath24-meson field contribution , see ref.@xcite .
s.yoshida , h.sagawa and n.takigawa , phys.rev .
* c58 * ( 1998 ) 2796 m.b.tsang , contribution to this conference , + h.s.xu et al .
, phys.rev.lett .
* 85 * ( 2000 ) 716 m.colonna and ph.chomaz , phys.rev .
* c49 * ( 1994 ) 1908 m. colonna , m. di toro and a. guarnera , nucl .
phys . a * 580 * ( 1994 ) 312 .
a. guarnera , m. colonna and ph .
chomaz , phys . lett .
* b373 * ( 1996 ) 267 ; m. colonna , nucl .
phys . a * 630 * ( 1998 ) 136c .
m.colonna , m.di toro and a.guarnera , nucl.phys . *
a589 * ( 1995 ) 160 m.colonna et al . , _ perspectives in heavy ion physics : 3rd italy - japan joint meeting _ , ed.s c.signorini et al . , world.sci,1999 pp.172 - 185 d. durand , nucl .
phys . a * 630 * ( 1998 ) 52c .
j.f.lecolley et al . , phys.lett .
* b354 * ( 1995 ) 202 , + @xmath25 indra data : j.colin , private communication ; f.bocage , caen thesis 1998 .
miller et al . , phys .
* 82 * ( 1999 ) 1399 , and ref.s therein .
l.scalone , m.colonna and m.di toro , phys.lett . *
b461 * ( 1999 ) 9
|
we show the possibility of extracting important information on the symmetry term of the equation of state ( @xmath0 ) directly from multifragmentation reactions using stable isotopes with different charge asymmetries .
we study n - rich and n - poor @xmath1 collisions at @xmath2 using a new stochastic transport approach with all isospin effects suitably accounted for . for central collisions a chemical component in the spinodal instabilities
is clearly seen .
this effect is reduced in the neck fragmentation observed for semiperipheral collisions , pointing to a different nature of the instability . in spite of the low asymmetry tested with stable isotopes
the results are showing an interesting and promising dependence on the stiffness of the symmetry term , with an indication towards an increase of the repulsion above normal density .
|
a large fraction of edge - on spiral galaxies have been observed to have boxy or peanut - shaped ( hereafter b / ps ) bulges . as the observations have improved and the samples have grown , the percentage of galaxies with these bulge shapes has increased from less than 20% @xcite to up to 45% @xcite . because the observed frequency of b / ps bulges is so high , significant effort has been invested in determining the formation mechanism responsible for these bulge shapes .
@xcite and @xcite demonstrated that the accretion of satellite galaxies could result in a galaxy bulge looking boxy or peanut - shaped .
the currently favored formation mechanism , however , is the buckling and subsequent vertical thickening of a bar ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
this is a plausible explanation , as bars are known to be common features in disk galaxies .
@xcite found that 79% of disk galaxies are barred in the near - infrared and 74% are barred in the optical ( see also * ? ? ?
* ; * ? ? ?
barred galaxies are also known to occur in a variety of environments . using a sample of 930 galaxies
, @xcite showed that the presence of a bar does not depend on whether the galaxy is in the field or is a member of a group or a cluster .
similarly , @xcite studied the bar frequency in isolated galaxies and found roughly equal numbers of barred and unbarred systems . to confirm the bar buckling hypothesis ,
bars must be detected in edge - on galaxies with b / ps bulges . while bars can be easily seen in face - on galaxies , it is much more difficult , however , to unambiguously photometrically confirm the presence of a bar in an edge - on galaxy .
this problem was simplified when @xcite and @xcite demonstrated that bars in edge - on galaxies could be detected kinematically from features like parallelograms and figures - of - eight in the position - velocity diagrams .
subsequent observations and simulations ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) have confirmed the relationship between b / ps bulges , complex kinematics , and bars .
we have obtained the optical long - slit spectrum and sparsepak integral field unit ( ifu ) velocity field , as well as optical and near - ir images , of ngc 2683 , an isolated , nearly edge - on sb galaxy with a b / ps bulge .
ngc 2683 is highly inclined ; @xcite report an inclination of 78@xmath1 , while others suggest a _ minimum _
inclination of 80@xmath1 ( e.g. * ? ? ?
* ; * ? ? ?
the complex kinematic and photometric features of the data are suggestive of a bar in the galaxy . although previous authors have observed ngc 2683 and linked its complex velocity structure and bulge shape with a bar , there has been little discussion about the bar orientation .
in this paper we present our spectroscopic and photometric observations as evidence that ngc 2683 contains a bar that is 6@xmath1 on the sky away from the galaxy major axis . in @xmath2 2
we review the data for ngc 2683 in the literature .
our new observations and data reduction are discussed in @xmath2 3 .
we present the spectroscopic data in @xmath2 4 and the photometric data in @xmath2 5 .
we describe in each section how these data support the hypothesis that a bar is present in the galaxy , and use the features of the kinematic and photometric signatures to constrain the orientation of the bar .
the results for ngc 2683 are compared in @xmath2 6 with the results for m31 by @xcite .
a comparison to m31 is insightful , as it has photometric and kinematic characteristics , as well as an inclination , that are similar to ngc 2683 .
finally , a summary is presented in @xmath2 7 .
because of its prominent boxy / peanut - shaped bulge and interesting kinematics , ngc 2683 has been a well - studied object . in 1974 ,
de vaucouleurs classified the bulge as peanut - shaped .
@xcite obtained a long - slit spectrum of the galaxy and noted that outside of the nucleus , the [ ] @xmath36584 emission was fainter than the emission , whereas inside the nucleus , it was brighter than the emission .
@xcite presented _ _ b-__band photometry and optical emission line spectroscopy . based on the galaxy image
, they found the bulge to be peanut - shaped and the dust - obscured nw side of the galaxy to be the near side .
the spectroscopic data showed trailing spiral arms , a [ ] / ratio in the outer parts of the galaxy consistent with what is seen in hii regions ( [ ] emission that is fainter than emission ) , and strong evidence for deviations from circular motions .
@xcite classified the bulge+disk shapes of 72 galaxies and categorized ngc 2683 as a bs - ii galaxy : a system in which the disk and bulge components are not sharply distinct and the bulge is an elongated ellipsoid that is sometimes rectangular .
an hirotation curve of the galaxy was presented by @xcite , and noncircular motions were noted .
_ j- _ and _ _ k-__band photometry , as well as _ ( j - k ) _ color , eccentricity , position angle and @xmath4 4@xmath5 profiles as functions of radius were presented by @xcite , as were models of the observed luminosity distribution .
@xcite found the and [ ] emission to show multi - valued , figure - of - eight velocity structure in the inner 60@xmath0 of the galaxy .
light from bulge stars or foreground gas warping out of the galaxy plane and into the line - of - sight were suggested as possible explanations for the observed complex velocity structure .
an hi position - velocity map of ngc 2683 was presented by @xcite .
the link between boxy bulges and bars was investigated by @xcite using a sample of 10 galaxies , including ngc 2683 .
they found that the emission line kinematics of galaxies with boxy bulges show complex velocity structure ; the position - velocity diagrams can be described as being x - shaped or containing features similar to figures - of - eight or parallelograms .
the complex kinematics of ngc 2683 were also observed by @xcite who noted the presence of a counter - rotating stellar system and two gas components , one of which was rotating in the same sense as the stellar component , and the other rotating in the opposite direction . a nuclear bar close to the minor axis or a merger event were two scenarios given to explain the counter - rotation .
additionally , @xcite observed the bulge isophotes to be boxy inside 50@xmath0 .
optical and nir photometry also led @xcite to classify the bulge of ngc 2683 as box - shaped .
stellar and ionized gas velocity curves as well as velocity dispersions were presented by @xcite .
they , too , detected kinematically distinct stellar and gas components , but remarked that they were unable to resolve the fast and slow rotating components in the gas .
@xcite presented a position - velocity diagram for ngc 2683 derived from , [ ] @xmath36584 , and [ ] @xmath35007 emission . due to the low _
s / n _ of their data , the figure - of - eight shape in the position - velocity diagram was faint , but they did observe two spatially distinct gas components . * *
we observed ngc 2683 with the rc spectrograph on the kitt peak national observatory ( kpno ) 4-meter telescope during the nights of 2007 october 16 and 18 .
we used the t2 kb ccd with the 860 line mm@xmath6 grating in second order , centered near .
the slit width was 1.5@xmath0 , giving @xmath7 1.0 spectral resolution and a spatial scale of 0.69@xmath0 pixel@xmath6 .
we centered the slit on a nearby bright star and then offset to the optical center of the galaxy .
two 600 s exposures were taken with the slit aligned along the major axis of the galaxy ( position angle @xmath8 41.5@xmath1 ) .
a henear lamp was observed before and after each science exposure to provide wavelength calibration .
we also observed ngc 2683 with the sparsepak ifu on the 3.5-meter wiyn @xmath9 telescope at kpno on the night of 18 february 2009 .
sparsepak is a 70@xmath0 @xmath10 70@xmath0 fixed array of 82 5@xmath0 diameter fibers .
we used the sta1 ccd with the [email protected] grating in eighth order , centered near h@xmath11 , giving a 40 km s@xmath6 velocity resolution .
the sparsepak array was aligned with the major axis of the galaxy and five pointings were used to cover the entire length of the galaxy . each exposure was 1200 s , and two exposures were taken at each pointing .
a thar lamp was observed to provide wavelength calibration .
the spectral data were reduced using standard reduction routines in iraf@xmath12 .
the data were bias - subtracted and flattened .
the iraf task ` dohydra ` was used to extract the ifu spectra .
the galaxy spectra were wavelength - calibrated using a wavelength solution created from the observations of the comparison henear and thar lamps . to increase the signal - to - noise and remove cosmic rays ,
the two galaxy frames per pointing were combined . in general ,
the galactic emission lines in the spectra are much stronger than the night - sky emission lines , and the night - sky lines bracket , rather than overlap , the galaxy lines .
sky subtraction was not performed for either data set and we subsequently used the night - sky emission lines as the reference wavelengths @xcite by which the velocities of the galactic emission lines were measured .
the velocities were measured by fitting gaussians to both the sky lines and the five galactic emission lines of interest : , [ ] @xmath36548 , [ ] @xmath36584 , [ ] @xmath36717 , [ ] @xmath36731 .
there was less scatter between the measured galactic emission line velocities when using the night - sky calibration than either the henear or thar calibrations .
the galaxy was also imaged at the kpno 2.1-meter telescope during the night of 2007 march 17 .
the galaxy was imaged in @xmath13 , @xmath14 , @xmath15 , and @xmath16 , with total exposure times of 600 s , 600 s , 300 s , and 300 s , respectively .
two exposures were taken in each band to correct for cosmic rays .
the t2 kb ccd was used , and the spatial resolution was 0.305@xmath0 pixel@xmath6 .
these data were taken under non - photometric conditions .
the optical broad - band images were bias - subtracted , flattened , and combined using standard iraf routines .
2mass _ j- _ and _
_ k-__band images of ngc 2683 are also available , and were downloaded from the on - line 2mass catalog @xcite .
in figure 1 we show the observed and [ ] @xmath3@xmath36548 , 6584 region of the long - slit position - velocity diagram ( pvd ) of ngc 2683 .
the pvd shows complex structure out to radii of @xmath7 45@xmath0 , almost half of its entire length . the and [ ] emission ( as well as the [ ] @xmath36717 and [ ] @xmath36731 emission that are not shown ) trace a parallelogram shape extending to @xmath7 45@xmath0 on each side .
the edges of this region are relatively bright compared to the faint interior emission . at radii @xmath17 45@xmath0 ,
the velocities are double , sometimes triple , valued .
the maximum velocity reached is comparable to the velocities reached in the outer parts of the galaxy .
the parallelogram velocities are seen to populate the `` forbidden '' quadrants of the pvd ; there are red - shifted velocities on the blue - shifted side of the galaxy and vice versa .
there is also a second parallelogram feature seen in the pvd . in approximately the inner @xmath18 5@xmath0
there is a steeper and more compact parallelogram that is brighter in the [ ] @xmath36584 emission than .
the velocities reached in this region do not exceed the velocities in the more extended parallelogram or those of the outer disk , and there is a smooth transition between the compact , inner and extended , outer parallelogram - shaped features .
this complex `` figure - of - eight '' pvd can be explained by the presence of a bar in the galaxy . in a series of papers , (
* hereafter ba99 ) , ( * ? ? ?
* hereafter ab99 ) , and ( * ? ?
* hereafter ba05 ) used simulations of edge - on ( @xmath19 = 90@xmath1 ) galaxies to demonstrate that features in pvds like those shown in figure 1 can be explained by gas ( and stars ) moving in the _ x@xmath20 _ and _ x@xmath21 _ families of periodic orbits of barred potentials .
each of these orbit families produces unique signatures in the pvd and , by virtue of the orientations of the orbits with respect to the bar , the appearance of the pvd features can be used to put constraints on the position angle of the bar . , , and showed that in edge - on galaxies _ x@xmath20 _ orbits produce extended features in the pvd that range from bow tie - shaped to parallelogram - shaped depending on the orientation of the bar with respect to the observer s line - of - sight ( see figure 2 ) .
x@xmath20 _ orbits are elongated parallel to the major axis of the bar , so that when the bar is viewed end - on , the line - of - sight is parallel to the major axis of both the bar and _ x@xmath20 _ orbits . viewed at this angle , the `` fast '' parts of the elongated _ x@xmath20 _ orbits are pointing directly toward or away from the observer and very high radial velocities at small projected distances are observed .
this produces a bow tie - shaped feature in the pvd .
when the bar is viewed side - on , the elongated axis of the _ x@xmath20 _ orbits is perpendicular to the line - of - sight .
the `` slow '' parts of the _ x@xmath20 _ orbits are thus pointing directly toward or away from the observer , and as a result , much lower radial velocities are observed .
in addition , because the long axis of the orbits is across the line - of - sight , these observed velocities will extend to larger projected distances than when the bar ( and _ x@xmath20 _ orbits ) is viewed end - on .
the resulting feature in the pvd is therefore a relatively thin , diagonal band of velocities extending to large projected distances .
when the bar in an edge - on galaxy is not viewed directly end - on or side - on , the pvd will contain a parallelogram - shaped feature with velocities appearing in the `` forbidden '' quadrants .
this velocity structure arises because the orbits are elongated and not circular .
the line - of - sight does not fall along either the major axis or minor axis of the orbits .
thus , if they are viewed at an angle other than end - on or side - on , the positions along the orbits where the tangential velocity is greatest ( and radial velocity is zero ) will be offset from the center .
this means that the velocities transition from being blue - shifted to red - shifted , and vice versa , at radii greater than zero .
this leads to red(blue)-shifted velocities being observed on the blue(red)-shifted side of the galaxy , thus falling in the `` forbidden '' quadrants .
the closer the bar is to end - on , the sharper and more peaked the parallelogram is at small projected radii ( and the more it looks like a bow tie ) .
the interior of these pvds also tend to be brighter than the edges .
as the bar angles closer to side - on , the corners of the parallelogram round out and the parallelogram becomes thinner as it extends to larger projected distances and rises to lower radial velocities . in the observed pvd of ngc 2683
, the broad ( @xmath1845@xmath0 ) parallelogram shape corresponds to the _
x@xmath20 _ feature .
the simulated pvds of , , and can serve only as a _ guide _ to the bar orientation in our galaxy because ( a ) ngc 2683 is not exactly edge - on ( @xmath19 @xmath22 78@xmath1 ) , and ( b ) the models are not tailored to specifically match ngc 2683 ( bar strength , pattern speed , mass distribution , etc ) . with that caveat in mind , both
the bright emission along the edges of the _ x@xmath20 _ parallelogram and a lack of clear central peaks suggest that the bar is oriented at an intermediate angle ( perhaps between 10@xmath1 and 45@xmath1 ) on the sky away from the galaxy major axis and therefore viewed closer to side - on than end - on . at the center of the observed pvd of ngc 2683
is the signature of the _ x@xmath21 _ orbits : a compact feature that is brighter in [ ] emission than emission ( figure 1 ; see also * ? ? ?
this relatively large [ ] emission may be caused by the shocks that develop at the transition between the _ x@xmath20 _ and _ x@xmath21 _ orbits ( baldwin , phillips , & terlevich 1981 ; dopita & sutherland 1996 ; ab99 ) , as well as the agn that is known to be at the center of ngc 2683 @xcite . without additional diagnostic emission lines such as [ ] or [ ] , we can not say for certain that it is the agn rather than shocks that is causing the bright [ ] emission .
it is not uncommon for barred galaxies to host agn .
@xcite observed a number of relatively face - on sb galaxies with large [ ] / ratios and confirmed the presence of active nuclei .
@xcite have also found that galaxies with active centers are more often barred ( 79% ) than galaxies without active nuclei ( 59% ) ( see also * ? ? ?
x@xmath21 _ orbits are elongated perpendicular to the bar ( making them also perpendicular to the _ x@xmath20 _ orbits ) and are less radially extended than the _ x@xmath20 _ orbits
. because the major axes of the _ x@xmath21 _ and _ x@xmath20 _ orbits are perpendicular , when one orbit family reaches its maximum radial velocity , the other is at its minimum .
thus , the radial velocities of the _ x@xmath21 _ orbits peak when the bar is viewed side - on and are lowest when the bar is viewed end - on . rather than using the shape of the _
x@xmath21 _ feature in the pvd , advocate the ratio of the maximum observed radial velocity of the _ x@xmath21 _ orbits to the maximum velocity observed in the outer parts of the galaxy ( @xmath23/@xmath24 ) as the best measure of bar viewing angle . when the bar is viewed close to end - on , the maximum velocity of the _ x@xmath21 _ orbits will be low compared to the outer disk ( the line - of - sight is perpendicular to the elongated axis of the _ x@xmath21 _ orbits and the `` slow '' parts of the orbit are observed ) ; @xmath23/@xmath24 @xmath17 1 in this case .
the `` fast '' parts of the _ x@xmath21 _ orbits are observed when the bar is viewed side - on ; it therefore follows that @xmath23/@xmath24 @xmath22 1 in this orientation . in ngc 2683 , we find that @xmath23/@xmath24 @xmath25 0.6 .
this value of @xmath23/@xmath24 suggests that the bar is viewed nearer to end - on than side - on .
this result contrasts with the other indicators we have , both photometric and kinematic , that suggest a more side - on orientation .
the @xmath26 indicator advocated by appears to be one of the more promising indicators of bar orientation .
however , it must depend on the details of the disk mass distribution .
some galaxies have a distinct peak in their rotation curves before declining to a flat level while others rise continuously .
consequently , @xmath24 of this particular galaxy may not correspond well to that of the ab99 model .
more generally , it may be an indication that we still do not have a complete understanding of the complex kinematics in ngc 2683 . in figure 3
we show the observed sparsepak velocity field of ngc 2683 .
the major and minor kinematic axes are not perpendicular and there is an -shaped twist to the minor axis .
these are characteristic of velocity fields with oval distortions or bars ( e.g. * ? ? ?
similar -shaped velocity fields can be seen in observations of more face - on barred galaxies such as ngc 6300 ( _ _ i__=52@xmath1 , * ? ? ? * ) and ngc 5383 ( _ _ i__=50@xmath1 , * ? ? ? * ) . as in the long - slit spectrum ,
double and triple - valued velocities ( as well as [ ] and [ ] velocities ) are detected in some of the ifu fibers .
the multi - valued velocities are more striking in the long - slit spectrum than in the velocity field .
analysis of the velocity field with the tilted - ring fitting program ` rotcur ` @xcite shows that the position angle ( pa ) varies by @xmath7 25@xmath1 across the velocity field , with the largest deviation from the pa of the outer galaxy occuring at @xmath7 45@xmath0 , the length of the _ x@xmath20 _ parallelogram in the long - slit pvd ( see figure 4 ) .
if we assume that there should be a steep velocity gradient ( i.e. the isovelocity contours bunch up ) across the bar and that the central minor axis velocity contours run along the length of the bar , then the center of the velocity field qualitatively indicates ( visually ) that the bar is angled not far ( perhaps @xmath7 15@xmath1 ) from the galaxy major axis on the sky , running from northeast to southwest .
this is supported by the change in pa determined by ` rotcur ` ; the bar is a maximum of 25@xmath1 from the major axis .
this angle implies that the bar is viewed more side - on than end - on , in agreement with the constraints from the parallelogram in the long - slit pvd produced by _
x@xmath20 _ orbits . in figure 5
we show the image of ngc 2683 constructed from the ifu data with isovelocity contours from the velocity field overplotted .
we find that while there is diffuse emission throughout the galaxy , the brightest concentrations of emission coincide with the bends in the -shape ( the ends of the bar ) .
the ne ( upper ) end of the bar is slightly brighter than the sw ( lower ) end and there is no strong emission along the length of the bar .
strong emission at the bar ends and relatively weak emission along its length has also been seen in the relatively face - on galaxies ngc 1530 @xcite , ngc 5383 @xcite , and ngc 6300 and ngc 1433 @xcite .
similar to what is seen in ngc 2683 , @xcite found that the brightest emission in ngc 1530 coincides with the regions of lowest velocity gradient perpendicular to the bar ( regions were the isovelocity contours are not bunched up ) .
boxy / peanut - shaped bulges in edge - on galaxies have been shown to correlate with the presence of a bar ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
it is thought that these bulge shapes are a result of a bar that has buckled and thickened due to vertical instabilities ( e.g. * ? ? ?
depending on the bar strength and the orientation of the bar with respect to the line - of - sight , the bulge may look round ( the bar is viewed end - on ) , boxy ( the bar is viewed at an intermediate angle ) , or peanut - shaped ( the bar is viewed side - on ) ( e.g. * ? ? ? * ; * ? ? ?
* ) . in figure 6
, we show contour plots of the bulge region of the _ b , v , r , i , j , _ and _ k _ images of ngc 2683 .
@xcite define peanut - shaped bulges as having isophotes that pinch inwards along the minor axis on both sides of the galaxy major axis , and boxy bulges as having isophotes that remain parallel to the major axis . applying these definitions to figure 6
, it is clear that the bulge of ngc 2683 is boxy .
this is most easily seen in the _ _ k-__band image of the galaxy where the obscuration from dust is at a minimum . on both sides of the major axis ,
the isophotes are parallel and remain flat as they cross the minor axis .
even as the galaxy is viewed through progressively bluer filters and the effects of dust on the northwest side of the galaxy become more severe , the isophotes on the less - obscured side of the galaxy do not show signs of pinching along the minor axis that are characteristic of a peanut - shaped bulge .
the classification of the bulge shape is independent of wavelength ( see also * ? ? ?
two of us ( rkd and ssm ) made independent ellipse fits with different fitting routines ( ` ellipse ` in iraf and archangel @xcite ) . while these ellipse fits are broadly consistent , they give slightly different results for the boxiness / pointiness parameter @xmath27 .
thorough examination of the images and residuals of the fits shows that this can be traced to isophotes that are both boxy and pointy .
this is most apparent in the @xmath28 and @xmath29-band images in figure 6 . along the minor axis of the bulge ,
the isophotes are flatter than a pure ellipse , indicating a boxy bulge . along the major axis ,
the isophotes become pointy , so that the entire isophote shape is both pointy and boxy .
we interpret the pointiness to be due to the bar itself extending out into the disk ; see figures 7 and 8 .
the length of the boxy region , corresponding to the thick inner part of the bar , is about @xmath1823@xmath0 in the @xmath29-band .
the pointy isophotes extend to roughly @xmath1854@xmath0 and correspond to the thin outer part of the bar .
it is not uncommon for barred galaxies to have features that connect to the ends of the bar such as ansae ( `` handles '' ) or rings ( e.g. * ? ? ?
* ; * ? ? ?
because ngc 2683 is so highly inclined , it is difficult to determine if these structures are also present .
the boxy bulge shape of ngc 2683 indicates that the bar is not so strong as to induce the pinched isophotes of a peanut and that the bar is positioned at an intermediate viewing angle . to determine the general orientation of the bar , we plot in figure 7 the observed surface brightness measured along 5@xmath30 slits positioned parallel to the galaxy major axis .
we find that the surface brightness profiles are asymmetric around the minor axis .
specifically , in the slices above the major axis ( the northwest slices ) , there is a `` hump '' in the profile indicating excess light to the left ( north ) of the galaxy minor axis and in the slices below the major axis ( the southeast slices ) , the ehancement in the light is to the right ( south ) of the galaxy minor axis .
we interpret this asymmetry in the light profiles as an indication that the bar is aligned in a n - s direction , consistent with the orientation derived from the sparsepak velocity field .
consistent with the approximate lengths of the boxy and pointy isophote regions , the excess light begins around @xmath18 20@xmath0@xmath3125@xmath0 , peaks around 45@xmath0 , and drops off between 50@xmath0 and 60@xmath0 .
based on the photometry , we estimate the length of the bar to be @xmath7 55@xmath0 . in figure 8 , we confirm this n - s orientation by plotting surface brightness profiles along slits oriented at different position angles .
we define 0@xmath1 as the minor axis and 90@xmath1 as the major axis .
there is an enhancement in the light along the slits rotated counterclockwise ( positive angles ) from the galaxy minor axis .
the humps in the profile reach a maximum above the level of the light along the major axis at a position angle of + 84@xmath1 from the minor axis ( or , equivalently , 6@xmath1 from the major axis ) .
this angle coincides with the asymmetric , elongated isophotes seen most clearly in the _ k_-band image . in agreement with the kinematic data , the humps in the surface brightness profiles extend to @xmath7 45@xmath0@xmath3150@xmath0 .
based on the photometric signatures , we conclude that the bar is oriented @xmath7 6@xmath1 from the galaxy major axis on the sky , running from north to south .
it is instructive to compare the observations of ngc 2683 with observations of other galaxies having similar inclinations ( @xmath19 @xmath22 78@xmath1 ) and bar orientations ( @xmath32 @xmath7 6@xmath1 , @xmath33 @xmath7 10@xmath1 @xmath31 25@xmath1 ) .
one such example is m31 .
@xcite recently used kinematic and photometric data to argue for the presence of a bar in the galaxy ( see also * ? ? ?
* ; * ? ? ?
they find that m31 ( @xmath19 = 77@xmath1 ) contains a medium - strength bar that is on the order of @xmath34 @xmath7 7@xmath1 on the sky away from the galaxy major axis ( pa@xmath35 = 38@xmath1 , pa@xmath36 @xmath7 45@xmath1 ) .
the observed pvds of ngc 2683 and m31 show similar complex kinematics .
the observed hi pvd of m31 ( their figure 9 ) is more bow tie - shaped than the long - slit pvd of ngc 2683 .
as discussed in @xmath24.1 , in edge - on galaxies , the pvd appears more bow tie - like as the bar angles farther away from the major axis .
taken at face value , the observed pvds would therefore suggest that the bar in m31 is seen more end - on than the bar in ngc 2683 .
the simulations described in @xmath24.1 , however , are _ ideal _ and in reality , the observed kinematics are highly influenced by physical conditions in the galaxy , such as the presence of dust or the density of the gas . without further modeling , we can not say whether the differences in the pvds originate from these physical differences or from different bar orientations or some combination of both .
the isophotes seen in the images of m31 ( their figure 2 ) are similar to the isophotes of the @xmath29-band image of ngc 2683 ; the isophotes are boxy rather than pinched in a peanut shape . based on isophote shapes in @xmath37-body simulations tailored to the orientation of m31
, they argue that the absence of peanut - shaped isophotes means the bar is not a strong bar ( see also * ? ? ?
* ; * ? ? ?
* ) . by this standard
, the bar in ngc 2683 is not strong .
similar to ngc 2683 , m31 has elongated isophotes that are angled away from the major axis and indicate the pa of the bar .
additionally , they find similar asymmetric enhancements in the light profile in slits parallel to the major axis ( their figures 5 and 6 ) .
it is encouraging that there are so many similarities between the observations of these two galaxies .
bars in nearly edge - on galaxies can not be unambiguously visually confirmed as such , but strong evidence for their presence comes from the comparison of observed kinematic and photometric signatures to the results of numerical simulations
. that simulations are consistent with numerous observed galaxies increases our confidence in using bars to explain boxy / ps bulges and figure - of - eight pvds .
we have presented new spectroscopic and photometric observations of the near edge - on spiral galaxy ngc 2683 .
the long - slit pvd displays a complex , figure - of - eight distribution of velocities , and the sparsepak ifu velocity field shows characteristics of an oval distortion .
isophotes in optical and near - ir images of the galaxy are boxy .
we find an asymmetric enhancement in the light profile in slits placed parallel to , and offset from , the major axis .
we argue that the kinematic and photometric signatures are evidence that the galaxy hosts a bar .
we find that the data support a bar that is viewed closer to side - on than end - on ( i.e. closer to the galaxy major axis than the minor axis ) .
based on kinematic and photometric constraints , we determine that the bar is 6@xmath1 away from the major axis on the sky .
our results are consistent with previous observations of ngc 2683 in the literature , as well as recent results for m31 , and add to the growing body of evidence linking b / ps bulges and complex pvds with bars in edge - on galaxies .
we would like to thank the referee for helpful comments .
is supported by an nsf astronomy & astrophysics postdoctoral fellowship under award ast 07 - 02496 .
s.s.m . and m.j.z .
are supported by nsf grant ast 05 - 05956 .
we would like to thank l. athanassoula for helpful conversations .
r.k.d . would also like to thank brian marsteller and misty bentz for useful 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 infrared processing and analysis center / california institute of technology , funded by the national aeronautics and space administration and the national science foundation .
athanassoula , e. & beaton , r. l. 2006 , , 370 , 1499 athanassoula , e. & bureau , m. 1999 , , 522 , 699 ( ab99 ) athanassoula , e. & misiriotis , a. 2002 , , 330 , 35 baldwin , j. a. , phillips , m. m. , & terlevich , r. 1981 , , 93 , 5 barbon , r. & capaccioli , m. 1975 , , 42 , 221 begeman , k. 1989 , , 223 , 47 binney , j. , & petrou , m. 1985 , , 214 , 449 bosma , a. 1981 , , 86 , 1825 broeils , a. h. , & van woerden , h. 1994 , , 107 , 129 bureau , m. & athanassoula , e. 1999 , , 522 , 686 ( ba99 ) bureau , m. & athanassoula , e. 1999 , , 626 , 159 ( ba05 ) bureau , m. & freeman , k. c. 1999 , , 118 , 126 buta , r. 1987 , , 64 , 383 buta , r. , ryder , s. d. , madsen , g. j. , wesson , k. , crocker , d. a. , & combes , f. 2001 , , 121 , 225 casertano , s. & van gorkom , j. h. 1991 , , 101 , 1231 chung , a. , & bureau , m. 2004 , , 127 , 3192 combes , f. & sanders , r. h. 1981 , , 96 , 164 combes , f. , debbasch , f. , friedli , d. , & pfenniger , d. 1990 , , 233 , 82 de souza , r. e. , & dos anjos , s. 1987 , , 70 , 465 de vaucouleurs , g. 1974 , in iau symp .
58 , formation and dynamics of galaxies , ed .
j. r. shakeshaft ( reidel : dordrecht ) , 335 dopita , m. a. , & sutherland , r. s. 1996 , , 102 , 161 eskridge , p. b. , frogel , j. a. , pogge , r. w. , quillen , a. c. , davie , r. l. , depoy , d. l. , houdashelt , m. l. , kuchinski , l. e. , ramirez , s. v. , sellgren , k. , terndrup , d. m. , & tiede , g. p. 2000 , , 119 ,
536 ford , w. k. jr . , rubin , v. c. , & roberts , m. s. 1971 , , 76 , 22 funes , j. g. , corsini , e. m. , cappellari , m. , pizzella , a. , vega beltrn , j. c. , scarlata , c. , & bertola , f. 2002 , , 388 , 50 irwin , j. a. , saika , d. j. , & english , j. 2000 , , 119 , 1592 jarvis , b. j. 1986 , , 91 , 65 knapen , j. h. , shlosman , i. , & peletier , r. f. 2000 , , 529 , 93 kuijken , k. , & merrifield , m. r. 1995 , , 443 , l13 laurikainen , e. , salo , h. , & buta , r. 2004 , , 607 , 103 lindblad , b. 1956 , stockholms observatorium annaler , band 19 , no . 2 ltticke , r. , dettmar , r .- j . , & pohlen m. 2000a , , 145 , 405 ltticke , r. , dettmar , r .- j . , & pohlen m. 2000b , , 362 , 435 marinova , i. , & jogee , s. 2007 , , 659 , 1176 martinez - valpuesta , i. , knapen , j. h. , & buta , r. 2007 , , 134 , 1863 merrifield , m. & kuijken , k. 1999 , 345l , 47 osterbrock , d. e. , fulbright , j. p. , martel , a. r. , keane , m. j. , trager , s. c. , & basri , g. 1996 , , 108 , 277 peterson , c. j. , rubin , v. c. , ford , w. k. , jr . , & thonnard , n. 1978 , , 219 , 31 philips , m. m. , charles , p. a. , & baldwin , j. a. 1983 , , 266 , 485 pompei , e. , & terndrup , d. m. 1999 , in asp conf .
182 , galaxy dynamics , ed .
d. r. merritt , m. valluri , & j. a. sellwood ( san francisco : asp ) , 221 raha , n. , sellwood , j. a. , james , r. a. , kahn , f. d. 1991 , nature , 352 , 411 rowley , g. 1988 , , 331 , 124 rubin , v. c. 1993 , in proc .
, 90 , 4814 schombert , j. 2007 , arxiv : astro - ph/0703646 shaw , m. a. 1987 , , 229 , 691 shaw , m. a. 1993 , , 261 , 718 sheth , k. , regan , m. w. , vogel , s. n. , & teuben , p. j. 2000 , , 532 , 221 skrutskie , m. f. , et al .
2006 , , 131 , 1163 stark , a. a. , & binney , j. 1994 , , 426 , l31 teuben , p. j. the stellar dynamics toolbox nemo , in : astronomical data analysis software and systems iv , ed .
r. shaw , h. e. payne and j. j. e. hayes .
( 1995 ) , pasp conf series 77 , p398 van den bergh , s. 2002 , , 124 , 782 vega beltrn , j. c. , pizzella , a. , corsini , e. m. , funes , j. g. , zeilinger , w. w. , beckman , j. e. , & bertola , f. 2001 , , 374 , 394 veilleux , s. , & osterbrock , d. e. 1987 , , 63 , 295 verley , s. , combes , f. , verdes - montenegro , l. , bergond , g. , & leon , s. 2007 , , 474 , 43 whyte , l. f. , abraham , r. g. , merrifield , m. r. , eskridge , p. b. , frogel , j. a. , & pogge , r. w. 2002 , , 336 , 1281 zurita , a. , relao , m. , beckman , j. e. , & knapen , j. h. 2004 , , 413 , 73
|
we present optical long - slit and sparsepak integral field unit emission line spectroscopy along with optical broadband and near ir images of the edge - on spiral galaxy ngc 2683 .
we find a multi - valued , figure - of - eight velocity structure in the inner 45@xmath0 of the long - slit spectrum and twisted isovelocity contours in the velocity field .
we also find , regardless of wavelength , that the galaxy isophotes are boxy .
we argue that taken together , these kinematic and photometric features are evidence for the presence of a bar in ngc 2683 .
we use our data to constrain the orientation and strength of the bar .
|
pseudo - cubic manganites are remarkable for their rich phase diagram although properties of various phases have not been studied equally well .
for example , the mechanisms of the `` charge ordering '' phenomenon ( co - phase ) or of the metallic anti - ferromagnetic phase ( a - phase ) , which usually appear near @xmath12 , to a large extent remain to be poorly understood .
it is also surprising that the potential of metallic manganites as the basic elements for various hetero - structures in development of alternate devices is practically unexplored . meanwhile
the diversity of their phases depending on temperature and doping concentration poses interesting questions regarding phenomena that may take place at the interface of the contacts between them .
further understanding of magneto - transport properties of these materials also remains of prime importance because of several reasons .
one of them is the colossal magneto - resistance effect , observed in these compounds at room temperatures . at the same time
there is an increased general interest in the artificial engineering of nano - structure materials with novel physical properties , such as the giant magneto - resistance ( gmr ) effect ( see @xcite for a review ) .
the discovery of the gmr effect has generated a lot of investigations focused on understanding the physical mechanism of this phenomena and that is why use of artificial hetero - structures composed of manganites suggests an interesting direction of research .
indeed , the a - phase with its weakly coupled ferromagnetic planes is a natural spin - valve system itself @xcite .
recent experimental results for large magneto - resistance on the domain walls in strained ferromagnetic manganite films provide another interesting example of such properties that may become important for various applications @xcite .
last but not least is a richness of magnetic and transport properties of manganites . concentration dependence and even overall topology pattern of the phase diagram varies from material to material in these compounds with general formula a@xmath0b@xmath1mno@xmath2 ( a = la , pr , nd ; b = sr , ca ) @xcite . in what follows we concentrate our attention mainly on la@xmath0sr@xmath1mno@xmath2 ( lsmo ) compounds because there are no complications due to `` charge ordering '' . instead their phase diagram is remarkable for the metallic ferromagnetic and metallic anti - ferromagnetic phases inside the concentration ranges @xmath13 and @xmath14 correspondingly @xcite .
in addition , the lsmo compounds allow the use of band description for broad concentration range . usually among the main difficulties in trying to describe the properties of magnetic hetero - structures on the theory part lies the fact that a contact area between the two materials with different magnetic ordering corresponds to an abrupt change in the underlying ground states : microscopic effects at the boundaries , values of transmission coefficients of carriers , change in the exchange mechanisms between the core spins _ etc .
_ , all of them taking place at the contact on the atomic scale .
we are motivated by the idea that the boundary between two manganite phases with rather close doping levels may present a better ground for the understanding of interface phenomena .
it is worth mentioning , that attempts to fabricate and investigate the characteristics of manganite hetero - structures have already been made @xcite . in this paper
we pursue a study of an electronic spectrum , mechanisms of conductivity and magneto - conductivity of both the ferromagnetic and anti - ferromagnetic ( a - phase ) phases .
in particular , we address the phenomena that may appear at the interface of their contact .
we use our results as an attempt to rationalize some experimental observations @xcite performed on films and hetero - structures made of manganites having these two phases as ground states .
manganites are thought to belong to the group of the materials with strongly correlated electrons . as it was mentioned above
, their phase diagram consists of numerous phases , with a crossover from one phase to another upon change in temperature @xmath15 , doping concentration @xmath16 or composition , a@xmath0b@xmath1mno@xmath2 , ( see , for example @xcite and references therein ) .
the parent lamno@xmath2 compound has the anti - ferromagnetic insulating state ( a@xmath17 ) at low temperatures . upon sr substitution with la above @xmath18 , it undergoes a transition to a ferromagnetic ( fm ) metallic state . with further increase in doping ( @xmath19 ) , as it was mentioned above , there is a transition from a ferromagnetic metallic to another anti - ferromagnetic metallic state ( a - phase ) .
the latter is defined as the phase , in which the core ( @xmath20 ) spins are aligned ferromagnetically in the planes ( for example , @xmath21-plane ) and anti - ferromagnetically along the axis perpendicular to the planes ( @xmath22-axis ) . in the ferromagnetic phase electrons
are fully polarized ( `` half - metallic '' state ) .
the `` half - metallic '' state is also realized inside the @xmath21-planes of the a - phase .
below we restrict our consideration to low temperatures . despite of the mutual consent on the prevailing role of the double exchange ( de ) mechanism @xcite as the basic mechanism responsible for ferromagnetism in manganites ,
there are several theories describing their low - temperature properties from different perspectives ( for a recent review , see @xcite ) .
for our purposes it is worth mentioning right here that for the concentration range @xmath23 , the physical properties of the lsmo compounds at least at low temperatures can be accounted for within the double exchange ( de ) model generalized for the degenerate orbitals @xcite .
the scheme suggested in @xcite is basically the band model utilizing however such important features of manganites as large values of hund s coupling constant , @xmath24 , and double degeneracy of the @xmath25 orbitals .
it was shown to be capable of some predictions regarding the evolution of the physical characteristics with change in electron concentration upon doping . at larger concentrations the super - exchange interaction between the @xmath20
core spins also plays an important role @xcite .
the fact that the band picture captures the main physics of manganites is not obvious .
often a generalized hubbard model is used to account for strong electron - electron interactions @xcite .
however , instead of a direct hubbard on - site @xmath26 which hinders the double occupation on each mn site , one may consider the local jahn - teller effect as another way to describe the same physics . indeed , a single electron positioned on the degenerate @xmath27-orbital of the mn site will cause a local lattice distortion , reducing the energy of a system . on the other hand two electrons on the same site do not lead to jahn - teller instability and the jahn - teller energy gain does not realize itself . therefore in the local picture
it is energetically favorable to have a single electron on a given site .
we should also mention the large values of the hund s coupling ( @xmath28ev ) , which is responsible for the polarization of all electrons on the mn site .
a tendency to the jahn - teller effect causes strong electron - phonon interaction which is thought to be the reason for various structural transitions in manganites , affecting their electronic properties .
the jahn - teller instability is inherent to the _ local _ high - spin @xmath29 configuration of a manganese ion due to double degeneracy of the @xmath27 orbitals .
when such an instability realizes itself as a _ cooperative _ jahn - teller transition , it also accounts for the coulomb interactions in a system , as it was discussed above .
in addition the coherent long - range order correlations of the local distortions along the crystal lattice favor the band description for electrons .
stability of the anti - ferromagnetic ( a-phase ) for the parent lamno@xmath2 compound was indeed explained in ref .
[ ] in terms of the jahn - teller cooperative effect incorporated into the two - band tight - binding approximation .
a primary manifestation of the jahn - teller effect comes about with an appearance of a proper deformation of the mno@xmath4 octahedron .
however substitutional doping also leads to the lattice deformations .
mismatch between the sizes of different ions is expressed by the value of the tolerance factor , @xmath3 : @xmath30 where @xmath31 are the ionic radii of each element in a@xmath0b@xmath1mno@xmath2 . a process of doping results in an average change of the lattice parameters , which may be described by the average @xmath32 .
when the value of @xmath32 is close to 1 , the `` cubic '' perovskite structure is realized as a whole .
one must distinguish @xmath32 from its `` local '' value : two materials may have close @xmath32 , i.e. close lattice parameters , but the local distortions or disorder may differ strongly from one site to another .
experiments show ( see @xcite and references there in ) that two compounds with close overall @xmath33 often have very different values of residual resistivity , @xmath34 .
analysis in terms of the band electrons is applicable if local disorder remains reasonably small , i.e. the mean free path of electrons is large enough . from this point of view , the considerable differences in the values of conductivity of various ferromagnetic manganites at the same doping level should be ascribed to strong variations in the values of the local tolerance factor . with the increase in disorder the mean free path
may become so short that the anderson mobility edge would hinder metallic conductivity even in the ferromagnetic phase .
such an unifying view turns out to be rather helpful for classification of the physical properties of the whole series of doped manganites @xcite . before proceeding further another simple observation regarding the role of the jahn - teller instability
can be made . at @xmath35 , according to @xcite , the a-phase in lamno@xmath2 is in the cooperative state with all bands filled up by electrons .
the sr doping lifts the local degeneracy of the @xmath27 orbitals and introduces random strains in the lattice .
de - localization in the electronic bands and hence gain in the kinetic energy also competes with _ cooperative _ jahn - teller effects .
therefore , it is expected that at intermediate concentrations @xmath16 the a- phase is destroyed resulting in an appearance of the ferromagnetic phase in the framework of the de mechanism .
there is no electron - hole symmetry in the two orbital model . at @xmath36
there are no electrons in the system and , therefore , departing from the `` other end '' of the phase diagram , @xmath37 , small number of carriers , @xmath38 , experience the jahn - teller instability and should be trapped locally at the mn sites . with an increase in the number of electrons ,
local jahn - teller traps will merge into a coherent state which is responsible for another lattice structure with the directions of spins remaining to be arranged .
although the cooperative jahn - teller effect seems to be a major mechanism responsible for the diversity of the phase diagram of manganites at @xmath12 and higher , spin interactions must be included to finally specify the resulting ground states @xcite . due to the small differences in ionic radii between the la and sr atoms ( @xmath33 ) , it seems , that the band description works best for la@xmath0sr@xmath1mno@xmath2 compounds @xcite . for these materials , being in the ferromagnetic metallic state
, the residual resistivity can be as small as @xmath39 @xcite , which is a good metallic conductivity range .
unfortunately , so far there are not so many experimental papers that deal with lsmo .
nevertheless , there are few @xcite and in what follows we keep lsmo materials in mind in our approach to different phenomena . in this paper
we first study the magneto - transport properties of the manganites for a more general case of a _ canted _ a - phase @xcite .
this term means that the core @xmath20 spins of the mn ions maintain both the fm and afm components ( e.g. spins in the a - phase can be canted by applying an external magnetic field ) .
we first derive an expression for the energy spectrum of electrons for the canted a - phase manganites ( section ii ) and then proceed with the derivation of the formulas for the magneto - conductivity by generalizing the diagrammatic `` cross '' technique @xcite for green functions in the two - band model ( section iii ) .
specific case of manganites makes a modification in the form of the `` impurity potentials '' necessary , to include a dependence on the orbital indexes .
more specifically , the distortions of the mn - o - mn conduction network are primarily caused by random octahedral distortions . with sr doping ,
the number of distorted octahedron increases so that one may expect a correlation between the number of carriers and the value of conductivity even for the high - quality samples . throughout this paper
we use the tight binding approximation in the frame of the two - orbital de model @xcite .
this model , as we already mentioned , has proved to capture the main physical properties of manganites at least in the metallic concentration range . finally in section
iv we briefly discuss phenomena in the vicinity of the interface between the a - phase and fm - phase manganites .
the details of our numerical calculations with some exact results are provided in appendix .
in the concluding section together with a general discussion an attempt also made to apply the results to recent experiments on la@xmath8sr@xmath7mno@xmath2/la@xmath6sr@xmath5mno@xmath2 superlattices @xcite .
we start directly with a general case of the canted a - phase magnetic structure .
the band hamiltonian has the form : @xmath40 here @xmath41 is an electron hopping matrix for the two - band model , @xmath24 is the hund s coupling constant on the mn sites , and @xmath42 is the fourier component of the af ordering along the _ c _ direction , @xmath43 .
the magnetic structural vector @xmath44 reduces brillouin zone ( @xmath45 is the cubic lattice constant ) ; @xmath46 is a canted magnetic moment , so that at each site @xmath47 : @xmath48 ( when @xmath49 we obtain the ferromagnetic phase ) .
the orientations of @xmath46 and @xmath50 are fixed by magnetic anisotropy ( easy plane ) and/or by an external field .
the matrix elements @xmath41 in the eq.([eq1 ] ) are calculated with the basis functions of the following form @xcite : @xmath51 as the result the hopping matrix elements are equal to : @xmath52 where @xmath53 and @xmath54ev being a hopping amplitude @xcite . from ( [ eq1 ] )
we obtain the following equation of motion : @xmath55 and similar equation for @xmath56 .
thus , the secular equation , is now an @xmath57 determinant from which one must calculate the eigenvectors and eigenvalues .
let us remind that the double exchange ( de ) mechanism for the manganites @xcite exploits the large value of the hund interaction , @xmath58 ev , so that @xmath59 . using this approximation , we solve eq .
( [ eq4 ] ) up to the terms of the order of @xmath60 .
the electrons can occupy only four lowest bands : @xmath61 , \\
e_{3,4}({\mbox{\boldmath $ p$}};m / s ) = -j_hs - |a|{\cdot}[c_x + c_y - ( m / s)c_z \pm r_{34}({\mbox{\boldmath $ p$}};m / s ) ] , \end{split } \label{eq5}\ ] ] where for brevity we introduced the notations @xmath62 and @xmath63 performing the canonical transformation in accordance with the eigenvalues of eqs .
( [ eq5 ] ) , it is straightforward to express operators @xmath64 and @xmath65 in eq .
( [ eq1 ] ) in terms of the new eigenfunctions . quite generally , the transformation has the form : @xmath66 where @xmath67 are creation and annihilation operators for the true energy branches ( [ eq5 ] ) , @xmath68 .
below we write down explicitly the expressions for @xmath69 for the four lowest bands : @xmath70 where the following notation has been used : @xmath71
it is straightforward to extend the standard `` cross '' technique for static defects @xcite for the two - band model . as for the nature of the defects ,
when an ion b is substituted for an ion a in the unit formula a@xmath0b@xmath1mno@xmath2 it immediately lifts the cubic symmetry at the mn - site .
the @xmath27 doublet gets split and the oxygen octahedron becomes distorted .
since , in accordance with our introductory remarks , this effect is of prime importance for the mn - o - mn conduction network , disorder in manganites to a large extent comes about through a change in doping . in application of diagrammatic cross - technique , however , we assume that positions of `` impurities '' ( i.e. of b ions ) remain random and there is no correlations between the scattering processes from any two of them . secondly , for the `` impurity '' potential @xmath72 in @xmath73 ( where summation is held over the random realizations of the `` impurities '' ) one can assume the jahn - teller form of the defect potential . using the basis given by eq .
( [ eq2 ] ) , the expression for @xmath74 is : @xmath75 where @xmath76 is an amplitude of the jahn - teller distortion at site @xmath47 , @xmath77 is an electron - lattice coupling constant and the angle @xmath78 specifies the shape of the distorted octahedron at a given mn site . going to momentum representation ,
the expression for @xmath79 is : @xmath80 with @xmath81 keeping in mind the energy spectrum obtained in the preceding section it is helpful to re - write expression ( [ eq9a ] ) in terms of new variables defined by ( [ eq6 ] ) : @xmath82 the so - called `` cross - technique '' @xcite can now be straightforwardly applied to the calculation of the average of new band green function given by : @xmath83 in the absence of the defects , the green function ( [ eq10 ] ) is : @xmath84 where @xmath85 ( again we can leave only four essential bands , since @xmath86 )
. for the green function averaged over defect s positions @xmath87 we obtain the well known form of equation schematically shown in fig.[fig74 ] .
the self - energy part @xmath88 on fig.[fig74 ] is again expressed in terms of the corresponding relaxation times as : @xmath89 where @xmath90 denotes an average over the fermi surface .
let us emphasize that attenuation @xmath91 in ( [ eq14 ] ) contains contributions from scattering between different bands ( [ eq5 ] ) . in the representation ( [ eq9b ] ) , the expressions for relaxation times @xmath92 are : @xmath93 where @xmath94 is concentration of `` impurities '' ( if our model with the potential ( [ eq8]-[eq9 ] ) for sr atom , one would have for @xmath95 , where @xmath16 is a dopant concentration ; we will come back to this in section iv ) ; @xmath96 is a density of states at the fermi level of the @xmath97th band . in the process of derivation of eq .
( [ eq15 ] ) we took into account that the main contribution to the integrals in eq .
( [ eq15 ] ) comes from the region close to the fermi surface .
we also used the following averages over disorder : @xmath98 , @xmath99 ( the latter one means averaging over the local shapes of distorted octahedra ) . with the use of eqs .
( [ eq14],[eq15 ] ) , the expression for the averaged green function ( [ eq13 ] ) can be written in the form : @xmath100 the dc - conductivity can be calculated from the kubo formula ( see , for example , in [ ] ) : @xmath101 where @xmath102 can be obtained with the help of the corresponding product of retarded green functions , averaged over impurities : @xmath103 via analytic continuation @xmath104 ( in eq .
( [ eq18 ] ) @xmath105 is the velocity operator defined as a derivative of energy with respect to the momentum for each band given by eqs .
( [ eq5 ] ) ) . with the impurity potential given by eqs .
( [ eq9 ] , [ eq9a ] ) , the average of product in expression ( [ eq18 ] ) can be re - written as @xmath106 now , taking into account equations ( [ eq13]-[eq19 ] ) and performing the integration in ( [ eq18 ] ) with respect to @xmath107 , we finally obtain the following expression for the in - plane and out - of - plane dc - conductivities : @xmath108 where @xmath109 where @xmath110 are defined by eq .
( [ eq15 ] ) , @xmath111 and integration runs over each fermi surface .
expressions ( [ eq20 ] ) for both in - plane and out - of - plane conductivities are rather complicated and the procedure of evaluating the conductivity using them deserves further explanation .
parameters of the electronic spectrum are chosen beginning with the initial two - band `` cubic '' phase which , as we believe , realizes itself in the ferromagnetic state .
let us also recall that the disorder in the calculations presented above expresses itself through the local octahedra distortion which , in turn , is produced by the substitutional disorder , i.e. by non - stoichiometric a@xmath0b@xmath1mno@xmath2 .
therefore , in ( [ eq20 ] ) the `` concentration '' of `` impurities '' is the concentration of b atoms , while the amplitude of the disorder potential is given by the value of @xmath112 in eq .
( [ eq15 ] ) . change in composition by increasing @xmath16 decreases the number of carriers , @xmath38 , while increasing the number of defects _
simultaneously_. another significant simplification above was that the distortions of the oxygen octahedra were treated independently .
indeed this is a good approximation , because two octahedra surrounding two neighboring mn atoms share one oxygen atom only .
if it were not so , the values of resistivity would depend on the b atom concentration only .
in reality a sample s quality also depends on how these octahedra adjust themselves .
factor @xmath94 in ( [ eq15 ] ) is not the only reason for dependence of conductivity on @xmath16 in ( [ eq20 ] ) .
change in carrier concentration results in a shift of the chemical potential relative to the bottom of the bands which reflects itself in an immediate change in the occupation number in each of the four active energy bands ( [ eq5 ] ) .
such a non - trivial intimate dependence between the number of carriers and the number of defects presents itself as a new feature for conductivity behavior in manganites .
it would be of great interest to investigate such a trend experimentally .
currently a shortage of experimental data for lsmo compounds for large enough sr concentrations deprives us the possibility to trace that dependence in some more details .
some estimates have been done for the fm - phase in our previous paper @xcite for @xmath113 . in this presentation
we perform the calculations for canted a - phase for @xmath114 . in fig .
[ cond](a , b ) we plotted our results .
energy spectrum of the a - phase itself in the de approximation ( @xmath86 ) would not allow current to flow in the out - of - plane direction : the dispersion @xmath115 drops out from ( [ eq5 ] ) for @xmath116 . therefore fig .
[ cond ] describes , as expected , a dramatic magneto - resistance effect inherent to the canted a - phase for the perpendicular - to - the - plane current ( @xmath117 ) .
surprisingly , it turned out that even the in - plane ( @xmath118 ) components of conductivity display considerable change in its value at the transition from the 3d conductivity regime in the ferromagnetic state ( i.e. at @xmath119 ) to the 2d one for the a - phase ( @xmath116 ) .
the origin of such a rapid change comes about from the re - distribution of carriers between the energy bands with change in the value of @xmath120 .
the effect of carrier re - distribution in the bands is seen in fig .
[ msvsef ] which shows the calculated position of the fermi level for doping concentration @xmath114 . with variation of @xmath120
the system undergoes dimensional transition between the 2d and 3d conductivity regimes . in fig .
[ cond ] ( a , b ) it seemed more convenient to present our results for the chosen concentration as a function of @xmath120 in accordance with eqs .
( [ eq19 ] ) .
we remind that @xmath46 is a ferromagnetic component of the _ core spins _ only . in order to find the values of conductivity as a function of the total magnetization , @xmath121 , which also includes the electronic component and
is induced by an external magnetic field , one may use the following simple relation : @xmath122 equation ( [ eq21 ] ) expresses the value of the full magnetic moment @xmath123 , in units of bohr magneton per mn ion , and @xmath124 is a magnetic susceptibility .
we now turn to some problems which involve artificial contacts of manganites with different magnetic ground states or just two manganites having the same ground states but different doping concentrations .
we will discuss both charge and magnetic structure in the vicinity of the boundary .
below we consider the plane geometry , so that all quantities depend on one coordinate only .
in addition we simplify our discussion by choosing the single band de model @xmath125 , where @xmath3 is a hopping amplitude .
we first consider two ferromagnetic manganites with different doping concentrations brought into contact with each other with the parallel orientations of the local moments .
some pronounced effects come about due to the schottky layers formed at the contact .
difference in doping concentrations produces a difference in the chemical potentials on both sides away from the contact , @xmath126 and @xmath127 , i.e. difference in the `` work functions '' of two components .
that leads to a redistribution of carriers near the contact plane .
this effect is general and well known for the contacts between metals or semiconductors .
a simplifying feature for the contacts of two manganites is in similarity of the underlying band structures on the both sides of a contact . at the same time
all major changes still take place at the atomic scale so that one needs to apply the kohn - sham scheme to solve for potential and charge distributions self - consistently . to elucidate some qualitative features we provide in appendix the solution of this problem in its continuous formulation .
we proceed as follows .
let @xmath128 be the concentration ( i.e. the number per @xmath129 ) of positive charges in ( lasr ) plane on the left(right ) side far away from the contact .
in the process of preparation of the hetero - structure ( film deposition)@xcite , the sr concentration @xmath130 changes sharply from @xmath131 to @xmath132 at the contact . the system of kohn - sham equations ( the poisson equation for the potential distribution and the schrdinger equation ) is : @xmath133,\\ -\psi_{\lambda}{(i+1 ) } + 2\psi_{\lambda}{(i ) } - \psi_{\lambda}{(i-1 ) } + \phi(i)\psi_{\lambda}(i ) = e_{\lambda}\psi_{\lambda}(i ) , \end{split } \label{h-1}\ ] ] where @xmath47 is an index , which runs through the mn planes , @xmath134 is an eigenvalue index , @xmath135 is an energy in units of @xmath3 , @xmath136 is a dimensionless potential defined by electrostatic potential @xmath137 as @xmath138 @xmath139 is a concentration of electrons on a plane @xmath47 and @xmath140 is a sr concentration , which depends on which side of the contact an electron is located and @xmath141 with @xmath142 being equal to the local fermi level in the units of @xmath3 .
we have obtained a numerical solution of ( [ h-1 ] ) with the boundary conditions providing the equality of the electrochemical potentials across the contact . in the calculations below
the total number of layers was equal to twenty ( ten on each side of the contact ) with : @xmath143 the solution for the potential and electron distribution is shown on fig .
[ potel](a , b ) .
the consistency of our numerical results have been verified by comparing them with the analytical ones obtained from the solution of the same problem in its continuous form .
the latter are described in appendix in more details . the significance of these calculations for the further discussion is as follows .
first of all , one sees that the electron screening ( the `` thomas - fermi '' length ) extends over four - five atomic distances implying that the sharpness of the contact is smoothed out considerably .
secondly , there is a re - distribution of charge between phases : repletion and depletion regions form close to the interface .
we now turn to the discussion of a contact between the manganites in the ferromagnetic and anti - ferromagnetic a - phase .
the results of the previous section have demonstrated the pronounced magneto - resistance effects in the canted a - phase for both in - plane and out - of - plane directions of a current .
this justifies an interest and need for better understanding of the f / a contact properties .
stacks of manganite films of different thickness with abrupt change in sr concentration in the ( lasr ) planes in order to stabilize fm- or a - phase , can be made using the state of the art deposition techniques @xcite .
however contacts between the fm state and the a - phase studied in @xcite pose more questions than the preceding example of the contact between the two unequally doped ferromagnetic metallic manganites .
schematically such a contact is shown on fig .
[ spins ] in the plane geometry .
a transitional layer lies in the shaded area and its structure may become complicated by a number of reasons .
first of all one should expect that the charge effects discussed above for a contact between the two fm phases exists at present case as well .
re - distribution of the carriers may shift the boundary between the magnetic phases .
less clear is the magnetic structure of such a transitional layer : whether there is a sharp boundary for the spins orientations on both sides of a contact or the direction of spins may rotate going through the interface from left to right , needs further discussion .
in particular , since the energy of magnetic anisotropy in manganites is rather weak ( magnetization of ferromagnetic manganites is saturated @xcite at external fields of the order of 100 oe ) orientation of moments in fig .
[ spins ] may even change gradually forming a structure similar to the one in the bloch or neel domain wall .
the transitional region between the two phases may include a canted a - phase , magneto - conductivity of which has been studied above .
it is highly desirable to further address these issues experimentally .
we return to the applicability of our band model for different doped manganites .
preparation of high - quality samples needs significant efforts and therefore experiments with doped manganites having good metallic conductivity at low temperatures are rare in the literature .
ferromagnetic metallic lsmo samples @xcite with @xmath144 have been reported .
estimates of ref .
[ ] give mean free path values as large as @xmath145 .
more typical values of residual resistivity are of the order of @xmath146 @xcite , which means that the mean free path lies on the scale of several inter - atomic distances .
higher values of resistivity often correlate with larger mismatch in the ionic radii of the constituent elements ( tolerance factor @xmath147 ) . from various experimental data on a - phase in manganites
pr@xmath148sr@xmath148mno@xmath2 , nd@xmath8sr@xmath7mno@xmath2 and la@xmath8sr@xmath7mno@xmath2 @xcite the values of the in - plane resistivity of these compounds lie in the range @xmath149 . in these samples one obviously has the conductivity regime in the a - phase lying close to the mobility edge .
as we have seen , magnetoconductivity and other properties of the a - phase are interesting enough to justify the efforts to improve the values of resistivity in the a - phase manganites . as for the lsmo films and hetero - structures prepared in @xcite ,
the value of residual resistivity in the ferromagnetic state of la@xmath6sr@xmath5mno@xmath2 compound and the value of the in - plane residual resistivity of the a - phase la@xmath8sr@xmath7mno@xmath2 compound are @xmath150 and @xmath151 respectively .
the films were prepared by a pulsed deposition method and obviously are not clean enough .
nevertheless , qualitatively there is an agreement : our calculations of the values of resistivity for ferromagnetic phase and a - phase ( when @xmath116 ) for the sr concentrations @xmath152 and @xmath153 correspondingly gave a factor of ten difference while the experimentally observed difference is only a factor of two .
as far as an interface is concerned , the complications that lie on the theory part are as follows .
the de mechanism that played such an important role in exploring the properties of the half - metallic magnetic ground state , bears a non - local character .
therefore it is not straightforward to account for it in an inhomogeneous problem with a spatial dependence near the interface ( recently , there were attempts to present the de mechanism in a local form to describe the various domain structures in fm metallic manganites @xcite ) .
secondly , diminishing the number of band electrons , the anti - ferromagnetic super - exchange interaction between the @xmath20 spins becomes comparable with the de interaction .
the task of combining these two mechanisms to study an inhomogeneous problem is already a serious problem .
the a - phase ground state has been found in a number of other compounds , such as pr@xmath148sr@xmath148mno@xmath2 and nd@xmath8sr@xmath7mno@xmath2 @xcite .
the transition into the a - phase along the temperature axis is of the first order as it is expected from symmetry considerations .
it is accompanied by a change in the @xmath154 ratio @xcite .
such a transition is often described in the literature in terms of the `` orbital ordering '' @xcite .
we suggest to interpret this transition as a cooperative jahn - teller effect involving the proper lattice distortion .
indeed , so far we have discussed changes in the electronic band structure caused by spin re - arrangements only . meanwhile , as it was discussed in introduction , lattice effects may also play an important role .
judging from various experimental results ( for a review , see @xcite ) , the importance of the lattice effects may vary and depend on a specific compound . for example , the lattice deformations strongly prevail in nd@xmath8sr@xmath7mno@xmath2 compound @xcite .
its ground state shows huge anisotropy in resistivity ( @xmath155 ) and much lower in - plane conductivity compared to other members of the a - phase family @xcite . as it was shown in ref .
[ ] the strong enough shear deformation @xmath156 of the oxygen octahedra ( @xmath11 ) alone may lead to a in practically two - dimensional electronic spectrum .
we suppose that nd@xmath8sr@xmath7mno@xmath2 is such an extreme case ( @xmath157 ) @xcite .
after carriers are added to the 2d bands both their and core spins adjust themselves via de and super - exchange mechanisms . if in other a - phase compounds the tetragonal deformations of the lattice , @xmath11 , is less pronounced , one may neglect it as we have done above for la@xmath8sr@xmath7mno@xmath2 .
the phase diagram for the la@xmath0sr@xmath1mno@xmath2 as a function of sr doping @xmath16 and fixed tetragonal distortion @xmath154 has been theoretically studied in @xcite .
we note , that the jahn - teller interaction has not been considered in @xcite and the electronic band structure was obtained for a fixed values of @xmath154 . much attention in ref .
[ ] was also given to the characterization of properties of the hetero - structures consisting of mixed phases [ f@xmath158 , a@xmath159 ( @xmath160 and @xmath161 are the number of unit cells per period of the structure ) .
major conclusion drawn by the authors @xcite from data on magnetization , structural characterization of the modulated films and their in - plane conductivity have led them to the notion of stable fm and a - phase single layers , which preserve their integrity and stability even in very thin intervening structures such as @xmath162 .
this result supports the point that stability of the a - phase layers is due to the coherent octahedra shrinkage , @xmath11 , in the planes .
such an idea leaves room for speculation regarding the possible spin arrangements in the heterostructures .
it is interesting that the conductivity measurements for the samples with composition [ f@xmath163 , a@xmath163 ] gave lower values of conductivity as if it were due to the fm layers alone .
we ascribe this to the effect of charge re - distribution shown in fig .
[ potel](b ) : ferromagnetic layers having more carriers supply part of them to the a - phase layers .
ten layers of each phase is already a good approximation for the picture of a single interface , as it is seen in fig .
[ spins ] . with
@xmath160 and @xmath161 decreased , the role of coulomb effects becomes weaker and each layer preserves its nominal composition and , hence , the in - plane conductivity .
one more comment we would like to add to the latter point concerns a pronounced increase in magneto - conductivity toward the [ f@xmath165 , a@xmath165 ] samples . if the number of electrons on each fm- and
a - phase planes does not change , the positive magneto - conductivity effect is an indication of stronger canting of the moments in these samples .
( recall that large values of in - plane magneto - conductivity on fig .
[ cond](a ) is a result of changing fermi level at the transition between the 2d and 3d regimes ) .
data provided by @xcite qualitatively agree with our results for the values of in - plane conductivity for the canted a - phase ( fig .
[ cond](a ) ) . to summarize , we have determined the electronic spectrum for the canted a - phase and calculated both conductivity and magneto - conductivity of such a ground state .
calculations have been done in the framework of the two - band model @xcite .
disorder introduced by the substitutional doping results in an intimate correlation between conductivity and number of carriers .
defects are described in terms of random jahn - teller centers .
magneto - resistivity for both in - plane and out - of - plane directions are expressed in terms of magnetization , the latter can be measured independently .
negative in - plane magneto - resistivity turns out to be large .
our study of effects at the contact between the two phases showed rather large screening length of the order of five unit cells .
we suggest that in the transition to the a - phase the jahn - teller cooperative ordering plays a leading role .
the resulting magnetic structure determines the final electronic band spectrum . we applied the results of the present paper to discuss the data obtained in @xcite .
there is a qualitative agreement of experimental data with our results and conclusions above .
more experimental work is needed to further investigate interesting properties of the a - phase and heterostructures in the presence of an external magnetic field .
this work ( m.d . and l.p.g . ) was supported by the nhmfl through the nsf cooperative agreement dmr-9527035 and the state of florida and by darpa through the naval research laboratory grant no .
n00173 - 00 - 1 - 6005 .
the work of ( v.z.k . ) was supported by darpa under contract no . 01j543 .
here we will present the exact solution of the eq .
( [ h-1 ] ) in the continuous limit . in that case
, we can solve the poisson equation on each side of the contact will and , after using the boundary conditions are able to obtain a general solution .
if the plane of the interface coincides with @xmath167 plane , the potential will depend on @xmath16 only .
we assume the spectrum of the electrons to have a parabolic form : @xmath168 .
the concentration of electrons is given by : @xmath169^{3/2 } , \label{a-1s}\ ] ] where @xmath170 is a local chemical potential .
taking into account ( [ a-1s ] ) , we have to solve the poisson equation : @xmath171 , \label{a-3}\ ] ] where @xmath130 is a sr concentrations defined by : @xmath172 we introduce the following notations : @xmath173 , where @xmath174 is an elechtro - chemical potential and @xmath175 is : @xmath176 the boundary conditions for the potential preserving the charge conservation are : @xmath177 introducing @xmath178 and using ( [ a-1s ] ) and ( [ a-5 ] ) , eq .
( [ a-3 ] ) is re - written as : @xmath179 where @xmath180 the first integral of eq .
( [ a-6 ] ) has the following form : @xmath181 , \\
v(x ) = v(x)/v(\pm\infty ) , \end{split } \label{a-6p}\ ] ] where @xmath182 . equation ( [ a-6p ] ) can be presented as : @xmath183 where plus(minus ) sign corresponds to @xmath184 .
integral on the left hand side of eq .
( [ a-7 ] ) can be calculated exactly .
the final result reads : @xmath185 & = & \pm\frac{\kappa x}{\sqrt{10 } } + c , \nonumber\\ \phi\left[v(x)\right ] & = & \sqrt{a_1 - a_2x}\cdot\left\{2\sqrt{{a^*}_1 - { a^*}_2x}\cdot(x+\alpha)\cdot{f}[\phi_1,m_1 ] - \right .
\nonumber\\ & & \left .
b_1\cdot\sqrt{a_2\cdot(x+\alpha)^2\cdot(b_2 + b_3\cdot{x})}\cdot \pi[n_1;\phi_2,m_2]\right\}/\left\{\sqrt{(x+\alpha)\cdot(x^2+|\beta|^2)}\right\ } , \label{a-9}\end{aligned}\ ] ] where f and @xmath186 are elliptic integrals of the first and third kind correspondingly and @xmath187 are complex numbers .
making use of boundary conditions ( [ a-5 ] ) recovers the results obtained numerically on fig.[potel](a , b ) .
it is also useful to obtain the solution of the poisson equation ( [ a-3 ] ) in the linear approximation , assuming @xmath175 is small compared to @xmath174 .
after very simple algebra , ( [ a-3 ] ) takes the following form : @xmath188 , \label{a-10}\ ] ] where @xmath189 taking into account the boundary conditions ( [ a-5 ] ) , the integration of ( [ a-11 ] ) is straightforward .
thus the solution of ( [ a-11 ] ) reads : @xmath190 where the notation @xmath191 was introduced for brevity .
as it is easy to see , the result given by ( [ afinal ] ) reproduces all the features of our numerical solution shown in fig .
[ potel](a , b ) .
|
we use the double exchange ( de ) model via degenerate orbitals and tight - binding approximation to study the magnetoconductivity of a canted a - phase of pseudo - cubic manganites .
it is argued that the model is applicable in a broad concentration range for manganites a@xmath0b@xmath1mno@xmath2 with the tolerance factor , @xmath3 , close to one . as for the substitutional disorder , scattering on random jahn - teller distortions of mno@xmath4 octahedra
is chosen .
we emphasize an intimate correlation between the carrier concentration and resistivity value of metallic manganites .
magnetoresistance as a function of magnetization is calculated for a canted a - phase for both in - plane and out - of - plane current directions .
a contact between two manganite phases is considered and structure of the transition region near the contact is discussed .
numerical calculations show charge re - distribution near the contact and a large screening length of the order of five inter - atomic distances .
we employed our results to interpret data obtained in recent experiments on la@xmath5sr@xmath6mno@xmath2/la@xmath7sr@xmath8mno@xmath2 superlattices .
we also briefly discuss the relative importance of the cooperative jahn - teller distortions , double exchange mechanism and super - exchange interactions for the formation of the a - phase at increasing sr concentrations @xmath9 in la@xmath0sr@xmath10mno@xmath2 to suggest that the jahn - teller contraction of octahedra , @xmath11 , plays a prevailing role .
|
as the number of extrasolar planets detected by various ground- and space - based surveys grows , so too do the number of so - called super earths " , exoplanets with masses of 1 - 10 earth masses .
many of these super earths transit their host stars along our line of sight , which allow us to directly observe their atmospheres using the same techniques as for hot jupiters ( e.g. , * ? ? ?
such a case is true for gj 1214b , a 2.7 earth - radius , 6.5 earth - mass super earth detected by the mearth survey @xcite . because gj 1214a is an m - type star only 13 parsecs away , the system has proven to be a favorable target for follow - up observations ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
@xcite concluded that the measured mass and radius of gj 1214b is most consistent with an interior that is water - dominated , with a hydrogen - helium envelope that is 0.05% the mass of the planet .
@xcite modeled the planet s interior structure , and concluded that if water were present in the planet s atmosphere , it would be a supercritical fluid .
hence , gj 1214b should not have a solid surface .
@xcite also modeled the interior of gj 1214b assuming a two - layer ( homogeneous envelope overlying a rock core ) structure , and found their results favor a composition similar to that of @xcite .
@xcite ran a range of internal structure / evolution models ( h / he or @xmath3 envelope overlying an earth - like nucleus ) , finding that only a small amount of h / he is needed to explain the planet s mass and radius . in anticipation of follow - up observations of gj 1214b , @xcite modeled transmission and emission spectra for a range of atmospheric compositions , from hydrogen - dominated ( i.e. , those with a low mean - molecular - weight [ mmw ] atmosphere ) to @xmath1- and @xmath4-dominated ( i.e. , those with a high mean - molecular weight ) .
they found that if the planet s atmosphere were @xmath5-dominated , the primary transit depth would show larger variations with wavelength than if the planet had an @xmath4- or @xmath1-dominated atmosphere ; this is because of the larger atmospheric scale height for an @xmath0-dominated atmosphere as compared to a high - mmw atmosphere .
this would lead to enhanced spectral features that should be detectable by current ground- and space - based instrumentation .
transit spectroscopic observations by most groups , however , favor a flat transmission spectrum , consistent with a high - mmw ( e.g. , water ) atmosphere or an atmosphere with high - altitude clouds or hazes ( e.g. , * ? ? ? * ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
still , observations by other groups favor a low - mmw atmosphere @xcite , particularly if methane is depleted @xcite .
photochemical modeling by @xcite also support a methane depletion , consistent with methane photolysis , but note that this process is not efficient at the pressure levels probed by transmission spectroscopy .
the composition will affect not only the atmospheric opacities ( hence absorption of starlight and emission of infrared radiation ) but also the atmospheric scale height , dry adiabatic lapse rate , and hence the dynamical stability and circulation of the atmosphere .
the circulation will determine the location of hot and cold regions in the atmosphere , which , in turn , shapes lightcurve and spectral behavior at photospheric levels .
in light of these considerations , we model the atmospheric circulation of gj 1214b , testing a multitude of atmospheric compositions . the circulation of gj 1214b has been explored by other groups @xcite .
however , our circulation model incorporates the most rigorous radiative transfer scheme used to model the atmosphere thus far ( see below ) . in section 2
, we describe our general circulation model , the sparc / mitgcm , and describe our model integrations . in section 3
, we present results from our model integrations , and identify general trends in circulation and temperature structure with metallicity and composition . in section 4
, we generate emergent flux spectra and lightcurves in anticipation of future instrumentation aboard the _ james webb space telescope _ ( jwst ) , _ thirty meter telescope _ ( tmt ) and other other ground- and space - based facilities .
+ the atmospheric circulation of gj 1214b is modeled using the substellar and planetary atmospheric radiation and circulation ( sparc ) model @xcite , which couples the mitgcm , a general circulation model ( gcm ) maintained at the massachusetts institute of technology @xcite , with a two - stream implementation of the multi - stream , non - gray radiative transfer scheme developed by @xcite . to emphasize its heritage
, we refer to this model as the sparc / mitgcm .
the mitgcm solves the primitive equations , a simplification of the navier - stokes equations assuming local hydrostatic balance , which is valid in stably stratified atmospheres with horizontal length scales greatly exceeding vertical length scales .
the primitive equations are solved on a cubed sphere grid , allowing for longer timestepping and better accuracy near the poles as compared to a latitude - longitude grid .
the radiative transfer code solves for the upward and downward fluxes through a given vertical column of atmosphere in the gcm , which determines the heating rate used to update the temperature and winds . for each chosen atmospheric composition ( see below ) the opacities are divided into 11 frequency bins using the correlated-@xmath6 method ( goody et al . 1989 ; for more details on the sparc / mitgcm , including recent updates to the model , see showman et al .
2009 and kataria et al .
each model integration has a horizontal resolution of c32 ( @xmath7128 in latitude and longitude ) and 40 or 76 pressure levels .
the pressure levels extend from a mean pressure of 200 bars at the bottom to 0.2 mbar at the top , evenly spaced in log pressure .
the top level extends from a pressure of 0.2 mbar to zero .
the sparc / mitgcm has been successfully adapted to investigate a variety of aspects of the atmospheric dynamics of hot jupiters and hot neptunes @xcite .
while the mitgcm is classically an earth gcm , this is the first time the sparc / mitgcm in its entirety has been used to model the circulation of a super earth .
however , given the likelihood that gj 1214b does not have a solid surface based on its mass , radius and temperature , we can use the sparc / mitgcm with few adjustments . utilizing the sparc / mitgcm for rocky , terrestrial exoplanets will be a task for future studies .
lccc 1@xmath2 solar & 13000 & 230 & 2.228 + 30@xmath2 solar & 9440 & 175 & 2.936 + 50@xmath2 solar & 8213 & 150 & 3.424 + @xmath4-dominated ( 99% @xmath8 , 1% @xmath1 ) & 1981 & 28 & 18.026 + @xmath1-dominated ( 99% @xmath1 , 1% @xmath8 ) & 1016 & 12 & 43.974 + @xmath9 , @xmath10 & 1296 & 17 & 31.00 + [ rundata ] we model six atmospheric compositions for gj 1214b , adapted from @xcite .
first , we model @xmath0-dominated ( i.e. , low - mmw ) compositions at @xmath11 , and @xmath12 solar , which have mean molecular weights of 2.228 , 2.936 , and 3.424 @xmath13 , respectively .
these models assume molecular species are in chemical equilibrium abundances at the local temperature and pressure , accounting for rainout of species that have condensed .
for the high metallicity cases , all species except for @xmath14 are enhanced by their respective factors .
second , we model an @xmath3-dominated atmospheric composition , which is composed of 99% @xmath8 , and 1% @xmath1 .
third , we model a @xmath1-dominated atmospheric composition ( 99% @xmath1 , 1% @xmath8 ) .
lastly , we model an intermediate high - mmw case , with a composition of @xmath15 and @xmath16 . for each model integration , we assume the winds to be initially zero , and assign each vertical atmospheric column the global - mean radiative - equilibrium temperature - pressure profile calculated using a one - dimensional ( 1-d ) radiative transfer code . @xcite
have shown that hot , synchronously rotating exoplanets exhibit circulation patterns that are insensitive to initial conditions .
figure [ pt_plots ] shows the pressure - temperature ( p - t ) profiles used in these initial conditions .
the hydrogen - dominated 1-d p - t profiles were calculated using the radiative - transfer code of @xcite adapted from @xcite .
the @xmath4- and @xmath1-dominated 1-d profiles were generated using the code of @xcite .
both codes calculate the temperature structure self - consistently assuming radiative equilibrium .
the sparc / mitgcm self - consistently solves for the flow as dynamics and heating evolve . in changing the atmospheric composition
, we are also changing the mean - molecular weight , the specific heat , and the scale height .
we calculate the specific heat using the method described in @xcite .
this is given on a per mass basis as @xmath17 where @xmath18 and @xmath19 are the specific heat and mixing ratio of the @xmath20 atmospheric constituent , respectively .
the scale height , @xmath21 , is given by @xmath22 , where @xmath23 is the specific gas constant , @xmath24 is the effective temperature , and @xmath25 is the planetary gravity .
the values of molecular mass , @xmath26 and @xmath21 for each composition are listed in table [ rundata ] , and vary over an order of magnitude . for each simulation , we use a dynamical timestep of 25 or 10 seconds with a radiative timestep of 500 or 200 seconds .
the simulations were each run for approximately 5000 earth days , with outputs every 100 days .
for all three @xmath14-dominated models , the atmospheres possess an equatorial superrotating jet , with speeds exceeding @xmath27 .
each model also exhibits a pair of jets in the high - latitudes .
this is seen in figure [ lowmmw_plots ] , which plots the zonal - mean zonal wind averaged over a planetary orbit for the 1@xmath2 , 30@xmath2 and 50@xmath2 solar composition .
overplotted in red are zonal - mean isentropes , contours of constant potential temperature . for the 1@xmath2 solar case ,
the high - latitude jets are centered at roughly @xmath28 , with peak speeds comparable those at the equator . for the high metallicity cases ,
the high - latitude jets are centered at @xmath29 , with speeds of 700 @xmath30 .
+ + + two trends in circulation are seen as the metallicity is increased .
first , the peak speeds of the jet increase ; equatorial jet speeds range from @xmath31 in the 1@xmath2 solar case to greater than 1.7 @xmath32 in the 50@xmath2 solar case .
second , the depth of the high - latitude jets decrease with increasing metallicity ; jets in the solar case extend to pressures of approximately 1 bar , while the jets in higher metallicity cases extend to only @xmath33300 mbar .
similar trends are seen in circulation models of hot neptune gj 436b ( lewis et al . 2010 ) .
the trends in circulation are a result of enhanced opacity associated with higher metallicities , which leads to shallower heating in the atmosphere ( fortney et al .
2008 , dobbs - dixon & lin 2008 , showman et al .
2009 , lewis et al .
. this enhanced opacity with higher metallicity leads to differences in the planet s horizontal and vertical temperature structure . in our models ,
the temperature difference from dayside to nightside varies with height throughout observable regions of the atmosphere for all three metallicities .
however , at a given pressure , this day - night temperature difference is greater for higher metallicities .
we compare the temperature variations as a function of pressure in figures 3 and [ daynight_diff ] .
figure 3 plots the wind and temperature profiles for each atmospheric metallicity at three pressure levels : 1 mbar , 30 mbar , and 1 bar , which approximately bracket the range of pressures over which infrared photons escape to space ( figure 3 ) .
indeed , at the shallowest pressure , 1 mbar , the 50@xmath2 solar model exhibits the highest day - night temperature variations . at 30 mbar , day - night temperature differences are small , but the 50@xmath2 solar case nevertheless exhibits the largest temperature variation from equator to pole . at 1 bar ,
only the 1@xmath2 solar case exhibits significant temperature variation , as stellar energy is deposited deeper at low metallicity .
these trends are illustrated further in figure [ daynight_diff ] , which plots the maximum dayside - nightside temperature difference at each pressure level for each atmospheric composition .
this is calculated at each pressure level by first latitudinally - weighting the temperature at each longitudinal slice .
we then determine whether each slice is on the dayside or nightside , then subtract the minimum ( weighted ) temperature on the nightside from the maximum ( weighted ) temperature on the dayside to determine the maximum dayside - nightside temperature difference . as shown in figure 4 , above photospheric pressures ( less than @xmath3310 mbar ) ,
the day - night temperature variation at each pressure increases with increasing metallicity .
given the expectation that the day - night heating drives the equatorial superrotation @xcite , this would imply stronger superrotation with increasing metallicity , qualitatively explaining the trend seen in figure 2 . at pressures greater than 10 mbar , where radiative time constants are longer
, the temperature varies in longitude by less than @xmath3325 k. these trends in temperature and wind structure will affect resultant synthetic lightcurves and spectra ( see section 5 ) .
+ + + + + ( green ) , and pressure - induced absorption ( pia ) due to @xmath1-@xmath1 collisions ( red ) .
the dotted grey lines denote the boundaries of the 11 spectral bins used in the correlated-@xmath6 calculation .
note that the @xmath1-@xmath1 pia is only prominent in the longest wavelength ( shortest wavenumber ) bin .
, scaledwidth=50.0% ] a comparison of high - mmw atmospheric compositions yield major differences in the dynamical and temperature regimes of gj 1214b .
if the atmosphere is @xmath8-dominated ( figure [ highmmw_plots ] , top row ) , the atmosphere still possesses an equatorial superrotating jet , with peak speeds of @xmath34 , and high - latitude jets with speeds exceeding 500 @xmath30 .
for a @xmath1-dominated atmosphere , however , equatorial superrotation is much weaker ; instead , the dynamics are dominated by high - latitude jets , with peak speeds exceeding 500 @xmath30 ( figure [ highmmw_plots ] , bottom row ) .
the @xmath15 , @xmath16 case , as expected , exhibits an intermediate behavior , whereby the atmosphere is dominated by broad , high - latitude jets and moderate equatorial superrotation ( figure [ highmmw_plots ] , middle row ) .
however , all three cases have higher equator - to - pole and day - night temperature variations at photospheric pressures ( @xmath3310 mbar ) than the low - mmw models .
these @xmath35100 k variations extend as deep as 100 mbar , an order of magnitude greater than the low - mmw cases ( figure [ daynight_diff ] ) .
the changes in dynamical and temperature regimes between low- and high - mmw atmospheres and between water- and @xmath1-dominated atmospheres can be attributed to differences in the vertical opacity structure and hence heating budget . for a @xmath1-dominated atmosphere ,
the atmosphere is more transparent to visible radiation .
hence , the stellar energy is deposited deeper in the atmosphere as compared to hydrogen- and water - dominated atmospheres .
the qualitative picture can be further confirmed by plotting the heating / cooling rates and visible and ir fluxes on the dayside and nightside ( figure [ flux_plots ] ) .
the top left panel shows the dayside net visible flux , which has a net downward direction .
the water case absorbs the incoming stellar energy much higher in the atmosphere as compared to the solar and @xmath1 cases , which corresponds to a much larger specific heating rate at the top of the atmosphere where the atmospheric mass is much less ( top right panel ) .
note also that the heating and cooling rates are smallest for the @xmath1-dominated case , helping to explain the weak superrotation .
the large variation in visible flux with height for the water - dominated case leads to a large specific heating rate at low pressures , where the atmospheric mass is less .
the bottom two panels plot the net ir flux at the substellar and antistellar points , respectively .
they show that the water - dominated case also emits flux at lower pressures compared to the other two compositions .
overall , the plots show that the @xmath1-dominated case absorbs energy deepest , and the water - dominated case highest . based on the results presented in @xcite
, one would expect that the 1@xmath2 solar case , which has the strongest superrotation , should absorb visible flux at lower pressures compared to the other two compositions , where day - night temperature variations and forcing are largest .
however , as described above , the water - dominated case has the shallowest flux deposition .
this suggests that the differences in specific heat ( and therefore scale height ) might also play a role in the differences in energy budget and dynamical regimes . to test this hypothesis
, we ran two models , the first with 1@xmath2 solar atmospheric opacities but a specific heat , mean - molecular weight , and scale height set to the @xmath1-dominated value , and a second model which has the reverse ( @xmath1-dominated atmospheric opacities , 1@xmath2 solar specific heat , mmw and scale height ) .
the 1@xmath2 solar opacity case does show flow features similar to that of the @xmath1-dominated case in figure [ highmmw_plots ] , with high latitude jets and weak superrotation at the equator .
a detailed analysis of these differences , specifically for the @xmath1-dominated case , will be a task for future studies .
we can compare our results to the other circulation models of gj 1214b , particularly menou ( 2012 ) , which models three of the atmospheric compositions included in this paper ( water - dominated , 1@xmath2 and 30@xmath2 solar ) , though with a different circulation model ( the intermediate general circulation model ) , radiative transfer scheme ( double - grey ) , and model setup . in comparing the hydrogen - dominated models ( see figure 2 in menou 2012 ) , one can see broad agreement , with equatorial superrotation in the 1 - 2 @xmath32 range , and high - latitude eastward winds .
however , the jet structure is different the equatorial jets in menou ( 2012 ) extend to deeper pressures than our models .
these differences could stem from differences in radiative transfer scheme , but also differences in bottom boundary ( 10 bars vs. 100 bars in our model ) . in comparing water - dominated circulation models of gj 1214b
, we can also include results from zalucha et al .
( 2013 ) , who use a different setup of the mitgcm coupled to a newtonian relaxation scheme , including a surface at varying pressures . in all three models there is again broad agreement , with an eastward equatorial jet with a width of approximately 50 - 60 degrees .
however , the models again differ in jet speeds and structure .
equatorial wind speeds are greatest in menou ( 2012 ) , and zalucha et al .
( 2013 ) model the weakest .
furthermore , both menou ( 2012 ) and our results include eastward winds at high latitudes , while zalucha et al .
has westward winds at the same latitudes .
these differences are most likely due to differences in the bottom boundary and radiative heating schemes .
while we already include opacity effects due to pressure - induced absorption ( pia ) from @xmath0-@xmath0 and @xmath0-@xmath36 collisions ( see kataria et al .
2013 ) , here we test the importance of including opacity effects due to pia from @xmath1-@xmath1 collisions .
figure [ taus ] compares the optical depth , @xmath37 , of this opacity source with the optical depths for a pure water and pure @xmath1 atmosphere .
this optical depth is calculated as a product of the number of molecules per @xmath38 , @xmath39 , and the absorption coefficients , @xmath40 .
the value of @xmath39 is defined as @xmath41 , where @xmath42 is the number density ( in units of @xmath43 ) and @xmath21 is the scale height .
this can be further simplified using the ideal gas law as @xmath44 . here
we calculate the optical depths for each composition in each wavelength interval at a temperature of 725 k and a pressure of 1 bar ( @xmath45 cgs ) . as shown in figure [ taus ] , the @xmath1-@xmath1 pia is most important in the longest wavelength ( shortest wavenumber ) frequency bin ( denoted by grey dotted lines ) . only a small fraction of the planet s flux is emitted in this wavelength range , and therefore we expect that the inclusion of @xmath1-@xmath1 pia should not significantly affect the dynamical structure .
figure [ co2pia_plots ] compares the transient spin - up phase of two @xmath1-dominated runs with ( right column ) and without ( left column ) the inclusion of @xmath1-@xmath1 pia in zonal - mean zonal wind ( top row ) and wind / temperature profiles at 30 mbar ( bottom row ) .
there are minor differences between both cases ; the westward flow at the top of the atmosphere extends to deeper pressures at the equator when pia is included , and the pia case exhibits a slightly different flow pattern at @xmath3330 mbar .
however , the bulk features remain the same : the speeds and horizontal / vertical extent of the high latitude jets , and the temperature and shape of the hottest regions on the dayside do not differ significantly . therefore , while it is important to include this opacity source , it does not dramatically change the dynamical and thermal structure of the atmosphere .
using the outputs from our model integrations , we can generate lightcurves and spectra of gj 1214b for each atmospheric composition .
most ground- and space - based observations of gj 1214b have been obtained during transit , but their flat transmission spectra suggest the presence of clouds that prevent easy characterization of the atmosphere .
therefore , only dayside emergent flux spectra obtained at secondary eclipse and lightcurves will be able to constrain the planet s atmospheric composition .
observations with the _ spitzer space telescope _ were able to detect secondary eclipse ( fraine et al .
2013 , gillon et al . 2013 ) and future instrumentation on the _ james webb space telescope _
( jwst ) and the _ thirty - meter telescope _ ( tmt ) will improve on those observations . in anticipation of these and other future instruments ,
we generate theoretical spectra and lightcurves at wavelength bands not specific to any particular instrument ( see below ) . in this way
, observers may use these theoretical lightcurves and spectra to select the wavelengths that best suit their efforts .
we choose to focus on atmospheric compositions of 50@xmath2 solar and 99% @xmath8/1% @xmath1 ( water - dominated ) , as these two models best illustrate the differences in emergent flux spectra and lightcurves that arise from differences in circulation and temperature structures .
as discussed in sections 3.1 and 3.2 , at each pressure level the water - dominated case has a greater temperature difference from dayside to nightside than the 50@xmath2 solar case .
therefore , we expect the water - dominated case to exhibit larger flux variations with orbital phase as compared to the 50@xmath2 solar case ( figure [ daynight_diff ] ) .
however , this will vary widely as a function of wavelength , as the water spectrum is dominated by fundamental and combination vibrational bands in the near - infrared ( ir ) and mid - ir . within these water bands ,
the atmosphere is opaque , and hence observations at these wavelengths will probe lower pressures .
thus , according to figure [ daynight_diff ] , we would expect greater day - night temperature variations and larger flux variation with orbital phase . at wavelengths outside of the water bands ( i.e. , in spectral windows ) observations sense deeper , hotter regions of the atmosphere where day - night temperature variations are smaller
; hence , there should be less flux variation with orbital phase .
we see this behavior in theoretical emergent flux spectra for the water - dominated model ( figure [ spectra ] , top panel ) . for both the water - dominated case and the 50@xmath2 solar case ( bottom panel ) ,
the spectra are plotted at six orbital phases , from transit , where the nightside is visible ( black spectra ) , through to 120@xmath46 after secondary eclipse ( magenta spectra ) .
the deep absorption features seen in the water - dominated case are fundamental vibrational bands of water vapor at 2.66 , 2.73 and 6.27 microns , as well as combination bands at 1.13 , 1.38 , 1.88 , and 2.68 microns . inside the water bands where we are probing low pressure regions ,
the large day - night temperature variations lead to large flux variations with orbital phase . outside of the bands ( inside the spectral windows ) , we probe deeper pressures where there are small temperature differences and hence small phase variations . comparing the emergent flux spectra of the 50@xmath2 solar composition
, we see the absorption features are not as deep and less dominated by water features .
the difference between windows and non - windows is also less prominent .
however , the 50@xmath2 solar case also exhibits variations in emergent flux with phase and wavelength , indicating that for both cases , different atmospheric pressure levels are probed at different wavelengths .
this can be quantitatively shown by plotting the pressure probed in emergent flux , where the optical depth , @xmath37 , is equal to one .
we calculate the @xmath47 level by first determining the brightness temperature , @xmath48 , as a function of wavelength , and finding the pressure level at which the globally - averaged temperature is equal to @xmath48 .
the results are plotted in figure [ tau1 ] with a colorscale corresponding to the maximum temperature variation at each pressure level from figure [ daynight_diff ] . for both atmospheric compositions ,
the wavelength regions with small ( large ) phase variations correspond to deeper ( shallower ) pressures , where day - night temperature variations are smaller ( larger ) .
lightcurves of each composition further illustrate the differences between low- and high - mmw compositions .
we plot the planet / star flux ratio as a function orbital phase for the water- and 50@xmath2 solar compositions in figure [ lightcurves ] . in each case , an orbital phase of 0.0 corresponds to transit , while an orbital phase of 0.5 corresponds to secondary eclipse .
six lightcurves are plotted at the general wavelength bands @xmath49 listed in table [ wl_bands ] and denoted in figures [ spectra ] and [ tau1 ] . for the water - dominated case ,
flux variations are large in all but one band ( band @xmath50 , black line ) . at this band
the @xmath47 level corresponds to a pressure level of 0.1 bars , where day - night temperature variations are small .
all other wavelength bands probe high in the atmosphere , where day - night temperature variations are large ( figure [ tau1 ] ) . for a 50@xmath2 solar composition , the flux variations are large for bands @xmath50 , @xmath51 , and @xmath52 which probe low - pressure regions where day - night temperature variation is high . bands @xmath53 , @xmath54 and @xmath55 probe deeper pressures , and hence exhibit smaller phase variations .
our results demonstrate that one can break the degeneracy in determining the atmospheric composition of gj 1214b by observing the planet in thermal emission .
large phase variations within water absorption bands and small variations in its spectral windows would indicate a water - dominated atmosphere .
other high - mmw species that are highly absorbing , such as carbon dioxide , ammonia , or methane , might in principle exhibit their own characteristic pattern of lightcurve amplitude with wavelength , depending on the wavelengths of their absorption bands and spectral windows .
as shown in figure [ spectra ] and [ lightcurves ] , a hydrogen - dominated atmosphere should exhibit a pattern of lightcurve amplitude with wavelength that differs significantly from that of a high - mmw atmosphere such as one that is water - dominated .
while we present this method in a generalized sense , one should be able to utilize space - based instruments such as the near - infrared spectrograph ( nirspec ) aboard jwst or the wide field camera 3 ( wfc3 ) on hst , although the latter has less spectral coverage and would require a multitude of orbits to achieve sufficient signal - to - noise .
instruments on the next generation of ground - based telescopes might also be able to utilize this technique , such as the near - infrared spectrometer ( gmtnirs ) on gmt or the infrared multi - object spectrometer ( irms ) on tmt .
however , full - phase lightcurves would be difficult to obtain from the ground in a single observation , and one would have to contend with the water vapor in earth s atmosphere .
therefore , reduction of ground - based observations would be much more difficult . in either case , in order to probe inside and outside water bands effectively as the method requires , spectral observations are necessary .
observations in photometric passbands ( i.e. , broadband observations like those on the _ spitzer space telescope _ ) might be able to apply this method , but would smear out these spectral features .
these results are particularly favorable because they would generally be independent of the the presence of clouds , minor equilibrium condensates or photochemical haze . in transit ,
slant optical depths through the planet s terminator can be dozens of times larger than vertical optical depths @xcite , which can suppress absorption features . in emission , however , paths are closer to vertical , suggesting that it is much easier for hazes to flatten the transmission spectrum than the emission spectrum . still , if the clouds or hazes are sufficiently thick , they would absorb and scatter the emergent flux which could in turn suppress emission features and flux phase variations . given recent @xcite results that gj 1214b likely has clouds or hazes , future work will include exploring how clouds with varying compositions and particle sizes as well as photochemical hazes can affect the phase variations presented here .
+ + lcc @xmath50 & 2.2 & 2.45 + @xmath51 & 2.5 & 2.9 + @xmath52 & 3.2 & 3.5 + @xmath53 & 4.25 & 4.4 + @xmath54 & 5.5 & 7.0 + @xmath55 & 14.0 & 16.0 +
we present three - dimensional atmospheric circulation models of the super - earth gj 1214b , exploring changes in circulation as a function of metallicity and composition . for hydrogen - dominated atmospheres ,
atmospheric opacities are enhanced with increasing metallicity , leading to shallower atmospheric heating .
this yields strong dayside - nightside heating / forcing that increases with metallicity , which in turn produces the highest day - night temperature variations and hence the strongest equatorial superrotation in the 50@xmath2 solar model .
the water - dominated composition also exhibits superrotation at the equator and eastward jets at high latitudes , but the circulation of the @xmath1-dominated model is dominated mainly by high - latitude jets .
all three high - mmw models have higher horizontal temperature variations at a given ( low ) pressure than the low - mmw models .
these differences in temperature structure and circulation can be attributed to differences in opacity structure and scale height .
the theoretical dayside lightcurves and spectra presented here lead to a major prediction for how to break the current observational degeneracy in the composition of gj 1214b s atmosphere .
in particular , the water bands dominate the spectra of the 99% @xmath8 , 1% @xmath1 case . within water absorption bands ,
large day - night temperature variations lead to large flux variations with phase . outside of the water bands ( within atmospheric windows ) , these phase variations are small . in comparison ,
a 50@xmath2 solar atmosphere generally yields small phase variations at those wavelength bands and large phase variations at other characteristic bandpasses .
therefore , observing in emission would break the degeneracy to determining the atmospheric composition of gj 1214b .
one could potentially constrain the existence of water or other highly - absorbing species by selecting wavelength bands inside and outside of their atmospheric windows , and comparing the extent of phase variations with that of a low - mmw atmosphere .
this diagnostic is much less sensitive than transit spectra to clouds , condensates and hazes .
however , sufficiently thick clouds and hazes would absorb and scatter emergent flux , therefore diminishing emission features and flux variations with orbital phase .
this work was supported by origins grant nnx12ai79 g to aps .
t.k . also acknowledges support from the harriet p. jenkins pre - doctoral fellowship program ( jpfp ) .
resources supporting this work were provided by the nasa high - end computing ( hec ) program through the nasa advanced supercomputing ( nas ) division at ames research center .
we thank roxana lupu for comments and for providing opacity tables for @xmath1-@xmath1 pressure - induced absorption .
we also thank daniel apai , nikole lewis , robert zellem for insightful discussions .
lastly , we thank the anonymous referee for their helpful comments and suggestions . 10 adcroft , a. , campin , j .-
m . , hill , c. , & marshall , j. 2004 , monthly weather review , 132 , 2845 bean , j. l. , kempton , e. m .-
r . , & homeier , d. 2010 , , 468 , 669 bean , j. l. , dsert , j .-
m . , kabath , p. , et al .
2011 , , 743 , 92 berta , z. k. , charbonneau , d. , dsert , j .-
, et al .
2012 , , 747 , 35 charbonneau , d. , et al .
2009 , , 462 , 891 cooper , c. s. , & showman , a. p. 2006
, , 649 , 1048 croll , b. , albert , l. , jayawardhana , r. , et al . 2011 , , 736 , 78 crossfield , i. j. m. , barman , t. , & hansen , b. m. s. 2011 , , 736 , 132 de mooij , e. j. w. , brogi , m. , de kok , r. j. , et al .
2012 , , 538 , a46 fortney , j. j. 2005 , , 364 , 649 fortney , j. j. , marley , m. s. , lodders , k. , saumon , d. , & freedman , r. 2005 , , 627 , l69 fortney , j. j. , cooper , c. s. , showman , a. p. , marley , m. s. , & freedman , r. s. 2006 , , 652 , 746 fortney , j. j. , lodders , k. , marley , m. s. , & freedman , r. s. 2008 , , 678 , 1419 fraine , j. d. , deming , d. , gillon , m. , et al .
2013 , , 765 , 127 gillon , m. , demory , b .- o . ,
madhusudhan , n. deming , d. , et al . 2013 , arxiv:1307.6722 goody , r. , west , r. , chen , l. , & crisp , d. 1989 , , 42 , 539 kataria , t. , showman , a. p. , lewis , n. k. , et al . 2013 , , 767 , 76 knutson , h. a. , et al .
2007 , , 447 , 183 kreidberg , l. , bean , j. l. , dsert , j .-
m . , et al .
2014 , , 505 , 69 lewis , n. k. , showman , a. p. , fortney , j. j. , marley , m. s. , freedman , r. s. , & lodders , k. 2010 , , 720 , 344 liu , b. , & showman , a. p. 2013
, , 770 , 42 marley , m. s. , & mckay , c. p. 1999
, , 138 , 268 miller - ricci , e. , seager , s. , & sasselov , d. 2009 , , 690 , 1056 menou , k. 2012 , , 744 , l16 miller - ricci , e. , & fortney , j. j. 2010 , , 716 , l74 miller - ricci kempton , e. , zahnle , k. , & fortney , j. j. 2012 , , 745 , 3 narita , n. , nagayama , t. , suenaga , t. , et al . 2013 , , 65 , 27 nettelmann , n. , fortney , j. j. , kramm , u. , & redmer , r. 2011 , , 733 , 2 parmentier , v. , showman , a. p. , & lian , y. 2013 , , 558 , a91 redfield , s. , endl , m. , cochran , w. d. , & koesterke , l. 2008 , , 673 , l87 rogers , l. a. , & seager , s. 2010 , , 716 , 1208 showman , a. p. , fortney , j. j. , lian , y. , marley , m. s. , freedman , r. s. , knutson , h. a. , & charbonneau , d. 2009 , , 699 , 564 showman , a. p. , fortney , j. j. , lewis , n. k. , & shabram , m. 2013 , , 762 , 24 showman , a. p. , & polvani , l. m. 2011 , , 738 , 71 teske , j. k. , turner , j. d. , mueller , m. , & griffith , c. a. 2013 , , 920 valencia , d. , guillot , t. , parmentier , v. , freedman , r. s. 2013 , , 775 , 10 zalucha , a. m. , michaels , t. i. , & madhusudhan , n. 2013 , , 226 , 1743
|
we present three - dimensional atmospheric circulation models of gj 1214b , a 2.7 earth - radius , 6.5 earth - mass super earth detected by the mearth survey . here
we explore the planet s circulation as a function of atmospheric metallicity and atmospheric composition , modeling atmospheres with a low mean - molecular weight ( i.e. , @xmath0-dominated ) and a high mean - molecular weight ( i.e. water- and @xmath1-dominated ) .
we find that atmospheres with a low mean - molecular weight have strong day - night temperature variations at pressures above the infrared photosphere that lead to equatorial superrotation .
for these atmospheres , the enhancement of atmospheric opacities with increasing metallicity lead to shallower atmospheric heating , larger day - night temperature variations and hence stronger superrotation . in comparison ,
atmospheres with a high mean - molecular weight have larger day - night and equator - to - pole temperature variations than low mean - molecular weight atmospheres , but differences in opacity structure and energy budget lead to differences in jet structure .
the circulation of a water - dominated atmosphere is dominated by equatorial superrotation , while the circulation of a @xmath1-dominated atmosphere is instead dominated by high - latitude jets . by comparing emergent flux spectra and lightcurves for 50@xmath2 solar and water - dominated compositions , we show that observations in emission can break the degeneracy in determining the atmospheric composition of gj 1214b .
the variation in opacity with wavelength for the water - dominated atmosphere leads to large phase variations within water bands and small phase variations outside of water bands .
the 50@xmath2 solar atmosphere , however , yields small variations within water bands and large phase variations at other characteristic wavelengths . these observations would be much less sensitive to clouds , condensates , and hazes than transit observations .
|
the electronic structure of nanoparticles define their electrical , chemical and optical properties .
hence , investigating these forms the basis for developing new applications in nanotechnology .
a comprehensive understanding of the various possibilities to modify the electronic structure of a particle is required when aiming to tailor compounds for specific applications .
research in this area can advance , for instance , the development of electron photoemitters@xcite or nano - electronics.@xcite diamondoids are perfectly size- and shape - selectable , hydrogen passivated , _ _
sp__@xmath0-hybridized carbon nanostructures.@xcite as such , they are an ideal class of particles for the study of effects induced by manipulation of geometry and chemical composition on the electronic structure in nanoscale systems . in addition , functionalization of these particles presents another possibility to tune their electronic structure.@xcite apart from the exploration of their existence in crude oil,@xcite continuous improvements in the field of synthesis of diamondoids@xcite have led to a rise in their popularity during the last ten years .
diamondoids are of interest for the oil industry@xcite and environmental protection,@xcite and are also used for medicinal applications,@xcite among others .
of particular interest in nanoparticles is the size - dependence of their electronic structure,@xcite often referred to in recent literature@xcite as quantum confinement ( qc ) effects .
as they are perfectly size - selectable , diamondoids enjoy great popularity in the investigation of such effects.@xcite recently , schreiner _
_ synthesized diamondoid particles with extraordinary long cc - bonds.@xcite except for a study on the smallest@xcite singly bonded particle , their electronic structure have not been investigated . to our knowledge
this is the case also for such particles connected with cc - double - bonds,@xcite for which results for only one compound have been reported in the literature.@xcite the present work deals with the valence electronic structure of such singly and doubly bonded diamondoid particles .
the idea to combine lower diamondoids@xcite is comparable to a modular design principle .
it is motivated by the question whether such a combination of lower diamondoids can imitate the electronic properties of larger , pristine diamondoids .
we utilized photoelectron spectroscopy ( pes ) for all measurements in this work .
a scienta ses-2002 hemispherical electron analyzer was used .
the samples in the spectrometer focus region were ionized at @xmath1ev with a he - lamp ( specs uvs 10/35 ) .
a resistively heated oven was used to bring the samples into the gas phase , see table [ tab : experimental_parameters ] for an overview of the various samples and the experimental parameters . while the ambient chamber pressure was in the mid @xmath2 mbar range , chamber pressures during measurements
were always kept constant in the mid @xmath3 mbar range . to calibrate the spectrometer energy scale , and to test the resolution ,
xe gas ( air liquide , 99.995% purity ) was used .
the experimental resolution was @xmath4mev for all samples .
.[tab : experimental_parameters ] experimental parameters used for the measurements in this work .
vaporization temperatures are given for ambient chamber pressures in the mid @xmath2 mbar range . [ cols="^,<,^ " , ]
we have studied the valence electronic structure of diamondoid particles . the influence of the bonding partners and the types of connecting bond have been investigated . for singly bonded particles
, the central cc - bond has only little impact on the energy levels of the dimers under consideration .
moreover , we observe that a combination of the bonding partner orbitals describes the overall electronic structure well
. a consequence of this combination process can be seen through an analysis of the particle ips . while for the homo - dimers we measure ips well below the corresponding monomers , the change of ips for hetero - dimers strongly depends on the particle composition .
qc effects can be seen for the homo - dimer ips but seem to be nearly absent for ips of hetero - dimers .
dft computations show symmetrically distributed valence orbitals for dimers with equal bonding partners . as in pristine diamondoids ,
the homos are delocalized over the entire molecule . in hetero - dimers , the constituent orbitals are asymmetrically distributed .
the homo tends to be localized to the larger bonding partner due to its lower ip .
hence the overall ip resembles that of the larger bonding partner , and qc effects are less pronounced . for doubly bonded particles ,
the cc - double - bond has a clear and identifiable impact on the electronic structure .
it appears as an isolated electronic feature with vibrational fine - structure at the ionization onset in the photoelectron spectra .
the ips vary only slightly with the size of the constituents .
computations show a strong localization of the homo to the cc - double - bond and the influence on the homo of the surrounding carbon atoms is restricted to screening .
the homo-1 is delocalized over the @xmath5-bonds of the diamondoid cage structures and can be identified as a distinct shoulder in the photoelectron spectra .
to further understand the influence of the combination of single diamondoids to larger particles on the electronic structure of the resulting system , more studies on this class of compounds are highly desirable .
using singly bonded particles enables researchers to imitate the electronic properties of higher diamondoids , thereby circumventing the problem of extremely low synthesis / isolation yields for higher diamondoids .
furthermore , the study of singly bonded dimers is an interesting route to gain insight into the role of dispersive forces on the long central cc - bonds .
the further investigation of doubly bonded particles assists the comprehension of carbon nanostructures with both _ _ sp__@xmath0- and _ _
sp__@xmath6-hybridized moieties .
besides the already explored parameters size , shape , and functionalization , the combination of pristine diamondoids into particles with different bonding situations opens a completely new degree of freedom to explore with regards to electronic structure and band gap tuning .
we would like to thank the deutsche forschungsgemeinschaft dfg for financially supporting this work through grant for 1282 mo719/10 - 1 and the ministry of science and education of ukraine for support through grant 2520f .
we would also like to thank professor peter zimmermann for fruitful discussion and assistance with the experimental setup .
36ifxundefined [ 1 ] ifx#1 ifnum [ 1 ] # 1firstoftwo secondoftwo ifx [ 1 ] # 1firstoftwo secondoftwo `` `` # 1''''@noop [ 0]secondoftwosanitize@url [ 0 ]
+ 12$12 & 12#1212_12%12@startlink[1]@endlink[0]@bib@innerbibempty link:\doibase 10.1126/science.1141811 [ * * , ( ) ] link:\doibase 10.1063/1.4756893 [ * * , ( ) ] @noop `` , '' ( ) , link:\doibase 10.1002/anie.200701684 [ * * , ( ) ] link:\doibase 10.1063/1.3280388 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.108.267401 [ * * , ( ) ] link:\doibase 10.1088/0953 - 4075/46/2/025101 [ * * , ( ) ] link:\doibase 10.1063/1.4774268 [ * * , ( ) ] link:\doibase 10.1126/science.1078239 [ * * , ( ) ] \doibase http://dx.doi.org/10.1016/0378-3812(95)02949-4 [ * * , ( ) ]
\doibase http://dx.doi.org/10.1016/s0920-4105(96)00059-9 [ * * , ( ) ] link:\doibase 10.1038/19953 [ * * , ( ) ] link:\doibase 10.1021/es060675n [ * * , ( ) ] link:\doibase 10.1016/s0960 - 894x(01)00174 - 3 [ * * , ( ) ] link:\doibase 10.1063/1.447218 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.90.037401 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.95.096801 [ * * , ( ) ] link:\doibase 10.1103/physrevb.80.161411 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.95.113401 [ * * , ( ) ] link:\doibase 10.1103/physrevb.74.205432 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.103.047402 [ * * , ( ) ] link:\doibase 10.1038/nature10367 [ * * , ( ) ] link:\doibase 10.1021/ja302258q [ * * , ( ) ] link:\doibase 10.1039/p29770000505 [ ( ) ] @noop link:\doibase 10.1021/ja00432a007 [ * * , ( ) ] link:\doibase 10.1103/physrevb.70.045401 [ * * , ( ) ] link:\doibase 10.1103/physrev.140.a1133 [ * * , ( ) ] @noop `` , '' link:\doibase 10.1007/s00214 - 007 - 0310-x [ * * , ( ) ] link:\doibase 10.1080/00268970802649625 [ * * , ( ) ] link:\doibase 10.1103/physrevb.72.035447 [ * * , ( ) ] link:\doibase 10.1063/1.2773725 [ * * , ( ) ] link:\doibase 10.1063/1.438739 [ * * , ( ) ] link:\doibase 10.1063/1.363158 [ * * , ( ) ] link:\doibase 10.1103/physrevb.53.16302 [ * * , ( ) ]
|
we investigated the valence electronic structure of diamondoid particles in the gas phase , utilizing valence photoelectron spectroscopy .
the samples were singly or doubly covalently bonded dimers or trimers of the lower diamondoids .
both the bond type and the combination of bonding partners are shown to affect the overall electronic structure .
for singly bonded particles , we observe a small impact of the bond on the electronic structure , whereas for doubly bonded particles , the connecting bond determines the electronic structure of the highest occupied orbitals . in the singly bonded particles a superposition of the bonding partner orbitals determines the overall electronic structure .
the experimental findings are supported by density functional theory computations at the m06 - 2x / cc - pvdz level of theory .
|
@xmath0-hypernuclei ( hypernuclei in the following ) decay through weak interaction to non - strange nuclear systems following two modes , the mesonic ( mwd ) and the non - mesonic ( nmwd ) one . the mwd is further split into two branches corresponding to the decay modes of the @xmath0 in free space : @xmath5 @xmath6 @xmath7 indicates the hypernucleus with mass number @xmath4 and atomic number @xmath8 , @xmath9 and @xmath10 the residual nuclear system , usually the daughter nucleus in its ground state , and the @xmath11 s stand for the decay widths . since the momentum released to the nucleon in mwd ( p@xmath12100 mev / c , q@xmath1337 mev ) is much lower than the fermi momentum , the mwd is strongly suppressed by the pauli exclusion principle in all but the lightest hypernuclei . in nmwd
the hypernucleus decays through weak interaction involving the constituent @xmath0 and one or more core nucleons .
the importance of such processes was pointed out for the first time in @xcite .
if the pion emitted in the weak vertex @xmath14 is virtual , then it can be absorbed by the nuclear medium giving origin to : @xmath15 @xmath16 @xmath17 the processes ( [ gammap ] ) and ( [ gamman ] ) are globally indicated as one - nucleon induced decays ( one - proton ( [ gammap ] ) , one - neutron ( [ gamman ] ) ) while ( [ gamma2 ] ) as two - nucleon induced decay . by neglecting @xmath0 weak interactions with nuclear clusters of more than two nucleons , the total nmwd width is : @xmath18 the two - nucleon induced mechanism ( [ gamma2 ] ) was first suggested in @xcite and interpreted by assuming that the virtual pion from the weak vertex is absorbed by a pair of nucleons ( @xmath19 , @xmath20 or @xmath21 ) , correlated by the strong interaction . in ( [ gamma2 ] ) we have indicated for simplicity only the most probable process involving @xmath19 pairs .
note that the nmwd can also be mediated by the exchange of mesons more massive than the pion .
the nmwd mode is possible only in nuclei ; the q - value of the elementary weak reactions driving the decays ( [ gammap ] ) , ( [ gamman ] ) and ( [ gamma2 ] ) is high enough ( q@xmath22175 mev ) to avoid any pauli blocking effect and the final nucleons thus have a large probability to escape from the nucleus .
indeed it is expected that nmwd dominates over mwd for all but the @xmath23-shell hypernuclei and only for very light systems the two decay modes are expected to be competitive .
the total decay width of an hypernucleus , @xmath24 , is thus given by : @xmath25 in order to compare data from different hypernuclei , partial @xmath11 s are usually given in units of @xmath26 , the total decay width of the free @xmath0 .
the nmwd of hypernuclei has been scarcely studied up to a few years ago .
experimentally it is not only necessary to produce and to identify hypernuclei in their ground state by means of a performing magnetic spectrometer , but also to detect in coincidence the nucleons emitted in ( [ gammap ] ) , ( [ gamman ] ) and ( [ gamma2 ] ) and to measure their energy .
furthermore , there is a big difficulty in extracting the observables related to the above mentioned weak processes due to the strong distorsion introduced by final state interaction ( fsi ) on the spectra of the nucleons emitted in the elementary processes corresponding to ( [ gammap ] ) , ( [ gamman ] ) , ( [ gamma2 ] )
. the information on the initial bare momenta may be completely lost , and their contributions can be mixed , with possible additional quantum - mechanical interference effects @xcite . following the first pioneering experiments @xcite , the sks collaboration measured at the 12 gev kek ps the spectra of protons and neutrons from the nmwd of @xmath27he and @xmath28c produced by the ( @xmath29 , @xmath30 ) reaction at 1.05 gev / c @xcite . at the da@xmath31ne ( @xmath32 , @xmath33 ) collider the finuda collaboration measured the spectra of protons and neutrons from @xmath27he and from 7 @xmath34-shell hypernuclei . a full account of all the papers on the subject may be found in @xcite .
many theoretical papers have also been produced , motivated by the strong interest and discovery potential of the study of nmwd of hypernuclei ; they are listed @xcite as well .
in an early paper @xcite proton spectra from nmwd of @xmath27he , @xmath35li and @xmath28c , measured by finuda , were presented and discussed . as a second step ,
proton spectra from nmwd of @xmath36be , @xmath37b , @xmath38c , @xmath39n and @xmath40o were produced and analyzed @xcite ; the experimental energy resolution was @xmath41 at 80 mev .
figure [ pspectra ] shows all the experimental spectra ; they are up to now a unique data bank for @xmath34-shell hypernuclei with @xmath1 , from which several interesting considerations and conclusions were drawn . in @xcite
we developed a method for disentangling the contributions from the decay ( [ gamma2 ] ) without using intranuclear cascade ( inc ) calculations , as in @xcite .
the first step was to fit the eight experimental spectra of figure [ pspectra ] above 80 mev ( @xmath42 ) to gaussians with free central values , widths and areas . in table
[ table1 ] the values of the gaussian s centers , @xmath43 , with their statistical error are reported .
.kinematics and gaussian fit parameters .
first column : hypernucleus ; second column : proton kinetic energy , @xmath44 , from a 2-body kinematics of one proton induced nmwd , with no daughter nucleus recoil ( see text for more details ) ; third column : gaussian fit mean value from @xcite , @xmath43 ; fourth column : present analysis gaussian fit mean value , @xmath45 ; fifth column : present analysis gaussian fit standard deviation , @xmath46 ; sixth column : gaussian fit mean value starting form 60 mev , @xmath47 .
statistical errors only are quoted . [ cols="^,^,^,^,^,^ " , ] they are given in table [ table2 ] , which reports also the experimental values of @xmath48 we used , when available ; for the other hypernuclei we adopt the parametrization @xmath49 proposed in @xcite .
the errors on @xmath3 are calculated by considering statistical errors for @xmath50 and @xmath51 , the errors reported in column two for @xmath52 and the statistical error for @xmath53 , which is the largest one ( 15@xmath54 ) .
table [ table2 ] reports , in the sixth column , the theoretical values of @xmath3 calculated recently in @xcite .
it is interesting to note that the new evaluation , following the determination of @xmath53 given by ( [ alpha_a ] ) , and the former indirect calculation for both @xmath27he and @xmath28c from ( [ explicit ] ) are compatible within the error : thus it appears that the method used to evaluate @xmath53 does not introduce any systematic error in the @xmath3 value . on the other hand ,
a systematic error on @xmath3 can be estimated by repeating all the previous procedure with the fits from 60 mev and from 80 mev : it corresponds to @xmath55 for @xmath56 , @xmath57 , @xmath58 , @xmath59 and @xmath60 for @xmath61 , @xmath62 , @xmath63 and @xmath64 .
another systematic error on @xmath3 can be obtained from the systematic error on @xmath65 : it amounts to @xmath66 only ; the total systematic error amounts to @xmath67 for @xmath56 , @xmath57 , @xmath58 , @xmath59 and @xmath68 for @xmath61 , @xmath62 , @xmath63 and @xmath64 and is small compared to the statistical one .
values as a function of @xmath4 for @xmath56 , @xmath57 , @xmath58 , @xmath61 , @xmath62 , @xmath63 , @xmath64 and @xmath59 from the present analysis ( blue stars , errors from table [ table2 ] ) .
theoretical calculations of @xmath3 @xcite ( violet squares ) are also shown .
@xmath3 from @xcite for @xmath56 ( brown full circle ) , from @xcite for @xmath37b and @xmath28c ( green full circles ) and from @xcite for @xmath28c ( orange full circle ) have also been plotted .
b ) @xmath69 values as a function of @xmath4 for @xmath56 , @xmath57 , @xmath58 , @xmath61 and @xmath64 ( red stars ) from @xcite and for @xmath62 ( orange cross ) from @xcite .
theoretical calculations of @xmath69 for @xmath56 ( gray up triangle ) from @xcite , for @xmath35li , @xmath36be , @xmath37b , @xmath28c and @xmath39n ( down violet triangles ) from @xcite and for @xmath56 , @xmath57 , @xmath58 , @xmath61 and @xmath64 ( cyan diamonds ) from @xcite are also reported
. , title="fig:",width=377 ] values as a function of @xmath4 for @xmath56 , @xmath57 , @xmath58 , @xmath61 , @xmath62 , @xmath63 , @xmath64 and @xmath59 from the present analysis ( blue stars , errors from table [ table2 ] ) .
theoretical calculations of @xmath3 @xcite ( violet squares ) are also shown .
@xmath3 from @xcite for @xmath56 ( brown full circle ) , from @xcite for @xmath37b and @xmath28c ( green full circles ) and from @xcite for @xmath28c ( orange full circle ) have also been plotted .
b ) @xmath69 values as a function of @xmath4 for @xmath56 , @xmath57 , @xmath58 , @xmath61 and @xmath64 ( red stars ) from @xcite and for @xmath62 ( orange cross ) from @xcite .
theoretical calculations of @xmath69 for @xmath56 ( gray up triangle ) from @xcite , for @xmath35li , @xmath36be , @xmath37b , @xmath28c and @xmath39n ( down violet triangles ) from @xcite and for @xmath56 , @xmath57 , @xmath58 , @xmath61 and @xmath64 ( cyan diamonds ) from @xcite are also reported .
, title="fig:",width=377 ] .
figure [ figgp]a ) shows the comparison among the values from this experiment ( blue stars ) , previous data @xcite ( brown full circle , green full circles and orange full circle respectively ) and theoretical values @xcite ( violet squares ) ; in the figure statistical errors only are indicated on the present results to match with previous data .
a general agreement between our data and the theoretical ones is evident , even though the experimental errors are quite large , with the exception of @xmath36be , which is lower by 1.5 @xmath70 and of @xmath40o which is lower by 1.3 @xmath70 .
figure [ figgp]b ) shows the experimental determinations of @xmath69 from @xcite ( red stars ) for @xmath56 , @xmath57 , @xmath58 , @xmath61 and @xmath64 and from @xcite ( orange cross ) for @xmath62 ; theoretical calculations of @xmath69 from @xcite ( gray up triangle ) for @xmath56 , from @xcite ( down violet triangles ) for @xmath35li , @xmath36be , @xmath37b @xmath28c and @xmath39n and from @xcite ( cyan diamonds ) for @xmath56 , @xmath57 , @xmath58 , @xmath61 and @xmath64 are also indicated .
the purpose is to show the first experimental verification , at least for the @xmath11 s relative to charged particles , of the long - time advocated complementary behaviour of mwd and nmwd of hypernuclei in the relevant @xmath4 range ( @xmath71 ) .
we have determined the partial decay widths of the one - proton induced nmwd from measured proton spectra for eight hypernuclei ( @xmath72 ) : it is the first systematic determination ever done for @xmath34-shell @xmath0-hypernuclei .
the measured values , though affected by errors ranging from 20@xmath54 to 30@xmath54 , agree reasonably with those predicted by a recent precise theoretical calculation @xcite . to make a better comparison smaller experimental errors are necessary , at a 10@xmath54 level or less , on both @xmath48 and @xmath50 ( @xmath51 ) ( see ( [ alpha ] ) ) . at present , the laboratory equipped with the beams and detector s arrays necessary for this kind of measurements is j - parc .
it is also necessary to develop a simple inc calculation , without two - nucleon induced contributions , to verify the validity of the hypothesis of linearity with @xmath4 of the fsi correction . with all these requirements satisfied
, it could then be possible to try to face the problem of the experimental study of the @xmath73 weak interaction from the hypernuclear data .
indeed , a precise study of the nmwd of hypernuclei will be the only way to get information on the four - baryon weak process @xmath74 and to realize then the idea of using a nuclear system as a `` laboratory '' for the study of interactions between elementary particles not otherwise accessible in vacuo .
|
previous studies of proton and neutron spectra from non - mesonic weak decay of eight @xmath0-hypernuclei ( @xmath1 ) have been revisited .
new values of the ratio of the two - nucleon and the one - proton induced decay widths , @xmath2 , are obtained from single proton spectra , , and from neutron and proton coincidence spectra , , in full agreement with previously published ones . with these values
, a method is developed to extract the one - proton induced decay width in units of the free @xmath0 decay width , @xmath3 , without resorting to intra nuclear cascade models but by exploiting only experimental data , under the assumption of a linear dependence on @xmath4 of the final state interaction contribution .
this is the first systematic determination ever done and it agrees within the errors with recent theoretical calculations .
@xmath0hypernuclei , two - nucleon and proton - induced non - mesonic weak decay width 21.80.+a , 25.80.pw
|
translating linear temporal logic ( ltl ) formulas to their equivalent automata ( usually bchi automata ) has been studied for nearly thirty years .
this translation plays a key role in the automata - based model checking @xcite : here the automaton of the negation of the ltl property is first constructed , then the verification process is reduced to the emptiness problem of the product .
gerth et al .
@xcite proposed an on - the - fly construction approach to generating bchi automata from ltl formulas , which means that the counterexample can be detected even only a part of the property automaton is generated .
they called it a tableau construction approach , which became widely used and many subsequent works @xcite for optimizing the automata under construction are based on it .
however , as we shall see , in the tableau framework the formulas are translated into the generalized bchi automata ( gba ) in an intuitive way so early that deeper relationships between ltl and bchi automata are ignored .
the muller automata is a generalization of bchi automata , and it preserves all the properties of bchi automata theoretically . for instance , the muller automata can also be used in model checking @xcite . moreover , in automata theory , the muller automata have nicer properties than bchi automata .
for instance , the determinization of muller automata can be as expressive as the non - deterministic ones , whereas it does not hold for the bchi automata .
the deterministic bchi automata can not be closed under complement , while the muller automata can be .
however there is not much research works on directly translating the ltl formulas to the muller automata , and one often obtains the muller automata from the existed automata @xcite . as a result , translating the ltl formulas to equivalent muller automata ( non - deterministic )
is also significant for model checking .
although the deterministic muller automata can not be obtained directly , it is essential for further research of the applications of muller automata . in this paper
, we propose a novel construction by making use of the notion of _ disjuctive - normal forms _ ( dnf ) .
for an ltl formula @xmath3 , its dnf normal form is an equivalent formula of the form @xmath4 where @xmath5 is a finite conjunction of literals ( atomic propositions or their negations ) , and @xmath6 is a conjunctive ltl formula such that the root operator of it is not a disjunction .
we show that any ltl formula can be transformed into an equivalent dnf normal form , and refer to @xmath7 as a clause of @xmath3 .
it is easy to see that any given ltl formula induces a labelled transition system ( lts ) : states correspond to formulas , and we assign a transition from @xmath3 to @xmath6 labelled with @xmath5 , if @xmath7 appears as a part of the dnf form of @xmath3 .
figure [ ltl_ts_and_ba ] demonstrates our idea in which the transition labels are omitted .
transition system as shown in the figure , we can first get the transition system ( ts ) from the formula @xmath8 , in which each node corresponds to a formula .
then we can extend the ts by doing the instantiation for formulas in each node to acquire the final bchi automaton ( ba ) .
the details will be talked about in section [ construction ] . + . ]
the lts is the starting point of our construction .
surprisingly , for until - free ( or release - free ) formulas , the bchi automaton can be obtained directly by equipping the above lts with the set of accepting states , which is illustrated as follows .
consider the formula @xmath9 , whose dnf form is @xmath10 .
the corresponding bchi automaton for @xmath9 is shown in figure [ fig : buchiforaub ] where nodes @xmath9 and @xmath11 represent formulas @xmath9 and @xmath12 respectively .
the transitions are self - explained . by semantics , we know that if the run @xmath13 satisfies a release - free formula @xmath3 , then there must be a finite satisfying prefix @xmath14 of @xmath13 such that any paths starting with @xmath14 satisfy @xmath3 as well .
thus , for this class of formulas , the state corresponding to the formula @xmath11 is considered as the single accepting state .
the until - free formulas can be treated in a similar way by taking the set of all states as accepting .
the main contribution of the paper is to extend the above construction to general formulas . as an example we consider the formula @xmath15 , which has the normal form @xmath16 .
note here the formula @xmath11 will be even not reachable . the most challenging part of the construction will then be identification of the set of accepting states . for this purpose
, we identify subformulas that will be reached infinitely often , which we call looping formulas .
only some of the looping formulas contribute to the set of accepting states .
these formulas will be the key to our construction : we characterize a set of atomic propositions for each formula , referred to as the _ obligation set_. the set contains properties that must occur infinitely often to make the given formula satisfiable . in our construction , we add an additional component to the states to keep track of the obligations , and then define accepting states based on it an illustrating example can be found in section [ example ] .
our construction for general formula has at most @xmath0 states with @xmath1 denoting the number of subformulas .
the number of states for the release / until cases is bounded @xmath2 .
recall the complexity of @xmath17 @xcite of the classical tableau construction . to the best of our knowledge
, this is the first time that one can give a precise bound on the exponent for such construction .
summarizing , the contributions of the paper are as follows : 1 .
we propose a _ disjunctive - normal form _
( dnf ) for the ltl formula , based on which we investigate the relationship between dnf normal forms and bchi automata . as a result we present a novel construction for translating ltl formulas to bchi automaton . especially for release
/ until - free formulas , our construction is very simple in theory .
we discuss some interesting properties of ltl formulas , which demonstrate another view on the satisfiability of ltl formulas . using these properties
, we can find the translation from ltl to bchi automaton is far more intuitive .
3 . as far as we know , compared to the complexity of @xmath17 @xcite for the tableau construction , our approach gives the more precise one of @xmath0 , and even more a better one of @xmath2 when the formula is restricted into release - free ( until - free ) .
etessami and holzmann @xcite have pointed out the goal of keeping the size of generated automata from ltl formulas small may not result in reducing the cost of model checking .
thus sebastiani et al .
@xcite suggested to make the generated automata `` as deterministic as possible '' and it is well accepted as a heuristic strategy by the researchers since then .
however , it should be noted that keeping the generated automata small is still significant and it is still one of the key goals for the generated automata together with their determinism .
thus , the new construction proposed here also adopts this criteria that makes the number of the states and transitions of the generated automata as small as possible .
the experiments in this paper also consider the `` determinism '' criteria and use the approach in @xcite to make the generated automata as deterministic as possible , which makes the experimental comparisons are more meaningful . as we know , there are two main approaches to bchi automata construction from ltl formulas .
the first approach generates the alternating automaton from the ltl formula and then translates it to the equivalent bchi automaton @xcite .
gastin et al .
@xcite proposed a variant of this construction in 2001 , which first translates the very weak alternating co - bchi automaton to generalised automaton with accepting transitions which is then translated into bchi automaton . in particular , the experiments show that their algorithm outperforms the others if the formulas under construction are restricted on fairness conditions .
recently babiak et al .
@xcite proposed some optimization strategies based on the work @xcite .
they make the observation on the formula where each branch of its syntax tree containing at least one eventually and always operators . compared to ltl2ba
, their implementation tool , ltl3ba , performs more fast and deterministic .
the second approach was proposed in 1995 by gerth et al .
@xcite , which is called the _ tableau _ construction .
this approach can generate the automata from ltl on - the - fly , which is widely used in the verification tools for acceleration of the automata - based verification process .
introducing the ( state - based ) _ generalized bchi automata _ ( gba ) is the important feature for the tableau construction .
daniele et al .
@xcite improved the tableau construction by some simple syntactic techniques .
giannakopoulou and lerda @xcite proposed another construction approach that uses the transition - based generalized bchi automaton ( tgba ) . and some optimization techniques @xcite have been proposed to reduce the size of the generated automata .
for instance , etessami and holzmann @xcite described the optimization techniques including proof theoretic reductions ( formulas rewritten ) , core algorithm tightening and the automata theoretic reductions ( simulation based ) .
the experiments in the literature also pointed out the smaller generated automaton does not always result in the better performance of the model checking . based on this observation , sebastiani et al .
@xcite focused on the determinism of the generated automaton and pointed that the generated automaton from the ltl formula should be `` as deterministic as possible '' , and this insight was accepted by other researchers , i.e. the tool spot @xcite integrated this heuristics into its recent implementation .
for muller automata our paper proposes a unified framework , thus without first generating the corresponding bchi automata , as done in @xcite .
although jong @xcite proposed a translation directly from the ltl formulas to muller automata based on the classic automata construction , the translation involves automaton complementation and intersection which makes the complexity of the approach higher .
section [ example ] illustrates our approach by a running example .
section [ sec : pre ] introduces preliminaries of bchi automata and ltl formulas and then introduces the _ disjunctive - normal form _ for ltl formulas ; section [ construction ] specifies the proposed _ dnf - based _ construction ; section [ discussion ] discusses how our approach is related to the tableau construction in @xcite .
section [ conclusion ] concludes the paper .
[ example ] we consider the formula @xmath18 as our running example .
the dnf form of @xmath8 is given by : @xmath19 where @xmath20 , @xmath21 , @xmath22 .
it is easy to check that the above dnf form is indeed equivalent to formula @xmath8 .
interestingly , we note that @xmath23 all have the same dnf form above . the bchi automaton for the formula @xmath8.,title="fig : " ] + the corresponding bchi automaton for @xmath8 is depicted in fig .
[ fig : buechi ] .
we can see that there are four states in the generated automata , corresponding to the four formulas @xmath24 .
the state corresponding to the formula @xmath8 is also the initial state .
the transition relation is obtained by observing the dnf forms : for instance we have a self - loop for state @xmath25 with label @xmath26 .
if we observe the normal form of @xmath8 , we can see that there is a term ( @xmath27 ) , where there is a conjunction of two terms @xmath26 and @xmath28 , and @xmath8 in @xmath29 operator corresponds to the node @xmath25 and @xmath26 corresponds the loop edge for @xmath25 .
if we look at the edge @xmath30 between @xmath25 and @xmath31 in the generated automaton , we still can find a corresponding relation between the normal form of @xmath8 and the automaton , e.g. , the second term in the normal form of @xmath8 gives the hint .
thus , the _ disjunctive - normal form _ of the formula has a very close relation with the generated automaton .
the most difficult part is to determine the set of accepting states of the automaton .
we give thus here a brief description of several notions introduced for this purpose in our running example .
the four of all the formulas @xmath32 have the same _ obligation set _ , i.e. @xmath33 , which may vary for different formulas . in our construction , every _
obligation _ in the _ obligation set _ of each formula identities the properties needed to be satisfied infinitely if the formula is satisfiable . for example
, the formulas @xmath34 are satisfied if and only if all properties in the obligation @xmath35 are met infinitely according to our framework
. then , a state consists of a formula and the _ process set _ , which records all the properties that have been met so far . for simplicity , we initialize the _ process set _ @xmath36 of the initial state @xmath25 with the empty set . for the state @xmath31 ,
the corresponding process set @xmath37 is obtained by taking the union of @xmath36 and the label @xmath38 from @xmath25 .
the label @xmath39 will be omitted as it is not contained in the obligation
. similarly one can conclude @xmath40 and @xmath41 : here the property @xmath42 implies no property has been met so far .
when there is more than one property in the _ process set _ , the @xmath43 can be erased , such as that in state @xmath44 .
moreover , the _ process set _ in a state will be reset to empty if it includes one _ obligation _ in the formula s _ obligation set_. for instance , the transition in the figure @xmath45{c\wedge d}}s_1 $ ] is due to that @xmath46 , which is actually in @xmath47 .
so @xmath48 is reset to the empty set .
one can also see the same rule when the transitions @xmath45{c\wedge e}}s_1 $ ] , @xmath49{c\wedge e}}s_1 $ ] , @xmath50{b\wedge e}}s_1 $ ] occur . through the paper
, we will go back to this example again when we explain our construction approach .
.the formulas expanded from @xmath51 and their normal forms [ cols= " < " , ] the next five formulas in the table are selected for the reason that they are often used in property description . again , _ aalta _ is better than ltl2ba or ltl2buchi without optimizations .
compared with the optimized versions of the two tools , our tool generates the automata of the same size for three formulas , and is worse for the other two formulas . for the formula @xmath52 , the size of the generated automaton from ltl2ba or ltl2buchi with optimizations
can be reduced to be linear with the length of the formula . however , the size of automaton generated by _ aalta _ for this formula is exponential to the length of the formula , which is the same case with ltl2ba and ltl2buchi when they do not optimize the results .
once the result automata from _ aalta _ are optimized by the same techniques as those in ltl2buchi ( + sbf ) , we can see almost the same results for the formulas in the table between the two tools , i.e. shown in the fifth and last column .
moreover for the last formula shown in the table , we can also find the result ( 5s , 14 t ) from _ aalta _
( + sbf ) can outperform the one ( 6s , 19 t ) from ltl2buchi ( + sbf ) .
now in table [ result2 ] and table [ result3 ] we present the experimental results on the constructions for the benchmarks , compared with ltl2buchi . to make the comparison more reasonable , we transform the generated automata from two tools to more deterministic versions as described in @xcite by enumerating all the combination of transitions for the automata .
this is mainly motivated by the fact that non - deterministic automata with the same transitions and states do not mean that they have the same performance for the model checking @xcite @xcite . in the two tables the field @xmath53 lists the percentage of the automata number from _ aalta _ in the total one whose both states and transitions are smaller than those from ltl2buchi .
the other fields denote the similar percentages .
note that the automata from _ aalta _ and ltl2buchi may not always meet the situations in column 2 , 3 and 4 in table [ result2 ] and [ result3 ] , so the total percentage of the three columns can not be 100@xmath54 .
first we scratch all the 2221 formulas without @xmath55 subformulas in benchmarks with a total number of 4046 to confirm the conjecture of `` the _ dnf - based _ construction can generates the bchi automata as small as possible when the formulas do not contain any @xmath55-operators '' . as table [ result2 ] shows , when considering the formulas without any @xmath55 operators and both _ aalta _ and ltl2buchi do not use optimizations ( column 2 , 3 , 4 ) , we have 27.0@xmath54 of the generated automata from _ aalta _ are smaller than those from ltl2buchi and 57.0@xmath54 of the ones from _ aalta _ are equal to those from ltl2buchi , i.e. totally about 84@xmath54 of the automata from _ aalta _ are as small as those from ltl2buchi .
meanwhile , there is just about 4.0@xmath54 of the results from _ aalta _ are bigger than those from ltl2buchi , which would not affect the truth of the _ dnf - based _ construction can outperform the original tableau construction any more . when comparing to the optimizing results from ltl2buchi ( + sbf ) , we have 5.1@xmath54 of the generated automata from _ aalta _ are smaller than those from ltl2buchi
( + sbf ) and 71.0@xmath54 of the ones are equal to those from ltl2buchi ( + sbf ) , i.e. totally 76.1@xmath54 of automata from _ aalta _ are as small as those from ltl2buchi ( + sbf ) , which is still a huge advantage . finally we present the comparison results for all the 4046 formulas between _ aalta _ and ltl2buchi .
the process time of 4046 formulas in the computer for the experiments is about 3 minutes . for the 2023 formulas in the first row in table [ result3 ]
, we have 37.4@xmath54 of the bchi automata generated by _ aalta _ are smaller than those generated by ltl2buchi ( the second column ) , while 46.2@xmath54 of the generated automata from the two tools have the same size ( the third column ) .
compared to ltl2buchi ( + sbf ) , 73.7@xmath54 of the bchi automata generated by _ aalta _ ( + sbf ) have the same size .
surprisedly , observe that 10.2@xmath54 of the generated automata by _
aalta _ ( + sbf ) are better .
the second formulas set is generated by negating the formulas in the first set ( second row ) and the statistics are similar . from this table
we can see our approach also has a better performance than the tableau construction when both of them use the same optimizations .
let @xmath3 be a formula @xmath56 such that the root operator of @xmath57 is not a disjunctive : then we define the disjunctive formula set as @xmath58 .
when @xmath59 does not include a disjunctive as a root operator , @xmath60 only include @xmath59 itself .
we first can directly use the rules in lemma [ lemma : expansion ] to generate an intermediate normal form for @xmath3 , whose format is @xmath61 where @xmath5 is an propositional formula and @xmath57 is an ltl formula without any constraint in definition [ def : dnf ] .
we denote the set of this intermediate normal form of the formula @xmath3 as @xmath62 ; second we prove any intermediate normal form can be changed to the + _ disjunctive - normal form_. intuitively , one can easily find for each @xmath5 and @xmath57 the corresponding @xmath63 and @xmath64 can be obtained trivially .
then we can get the final _ disjunctive - normal form _ through the following two steps : * basic step : if @xmath59 is the case of the literal @xmath72 , then since @xmath73 , so obviously @xmath74 . *
inductive step : if the formulas @xmath57 ( @xmath75 ) satisfy @xmath76 , then : 1 .
if @xmath77 , we know @xmath78 .
according to lemma [ lemma : expansion].5 we have @xmath79 , then by induction hypothesis we have @xmath80 , so @xmath81 ; 2 . if @xmath82 , we know @xmath83 .
according to lemma [ lemma : expansion].2 we have @xmath84 , so @xmath85 ; 3 . if @xmath86 , we know @xmath87 .
according to lemma [ lemma : expansion].6 we know @xmath88 . then by induction hypothesis
we have @xmath89 and @xmath90 , so @xmath91 ; 4 . if @xmath92 , we know @xmath93 . according to lemma [ lemma : expansion].3 if @xmath94 then @xmath95 directly by induction hypothesis , else if @xmath96 then by induction hypothesis we have @xmath97 ; 5 .
if @xmath98 one can also prove in the similar way that @xmath99 .
* base step : if @xmath70 then according to lemma [ lemma : dnf : finite ] we know @xmath71 . *
induction step : if @xmath102 where @xmath103 and @xmath71 hold , then according to lemma [ lemma : dnf : finite ] we know for all @xmath104 we have @xmath105 .
then according to lemma [ lemma : expansion].6 we know @xmath106 holds .
that is , if @xmath101 can be reached from @xmath3 in @xmath107 steps and @xmath81 holds , then any @xmath108 can be reached from @xmath3 in @xmath109 steps also has @xmath110 . now come to prove theorem [ thm : expand : bounded ] . from lemma [ lemma :
expand : finite ] we know for all @xmath111 if @xmath112 then we have @xmath113 .
so the elements number in @xmath114 can not exceed the number of @xmath68 , i.e. @xmath115 . thus @xmath116 . *
basic step : if @xmath117 , then @xmath118 , and @xmath119 obviously true .
* inductive step : if for the formulas @xmath120 ( @xmath121 ) , @xmath122 holds .
then we have : 1 . if @xmath123 , then @xmath124 . since for each @xmath125 in @xmath126 , the predicate @xmath127 according to its definition , and since @xmath124 so @xmath128 . then by induction hypothesis
we know @xmath129 holds thus @xmath130 holds .
if @xmath131 , then @xmath132 , so we know @xmath133 . then since @xmath134 , and by induction hypothesis @xmath135 holds when @xmath128 while @xmath136 holds when @xmath137 . due to @xmath138
so @xmath139 is true .
3 . if @xmath140 , then @xmath141 . then @xmath142 . by induction hypothesis that @xmath143 and @xmath144 are true , thus @xmath145 holds .
if @xmath146 , then @xmath147 . since for each @xmath125 in @xmath126 @xmath148 , and by induction hypothesis @xmath136 holds , so @xmath130 also holds .
similarly one can prove the situation when @xmath149 and we omit it here .
* basic step : if @xmath117 , then @xmath118 , and @xmath150 .
so obviously @xmath151 .
* inductive step : if for the formulas @xmath120 ( @xmath121 ) , @xmath152 holds .
then we have : 1 . if @xmath123 , then we know @xmath124 and @xmath153 . since by induction hypothesis @xmath154 , and @xmath155 , so @xmath156 .
thus @xmath157 holds .
2 . if @xmath131 , then we have @xmath132 and @xmath158 . by induction hypothesis @xmath154 and @xmath159 , so @xmath160 , in which @xmath161 is exactly @xmath126 .
thus @xmath157 holds .
3 . if @xmath140 , then @xmath141 . since @xmath162 , and by induction hypothesis we have @xmath163 , where @xmath164 , so @xmath165 .
obviously @xmath166 , so @xmath157 holds .
if @xmath146 , then we know @xmath147 and @xmath167 . by induction hypothesis @xmath168 , and since @xmath169 so @xmath125 is also in @xmath126 . thus @xmath157 holds .
similarly one can prove the case when @xmath149 and we omit it here .
for each @xmath173 in @xmath174 let @xmath175 , then we know easily @xmath176 . if @xmath177 , then we know @xmath178 and @xmath179 .
since @xmath180 is also in @xmath174 , then according to the assumption @xmath181 and @xmath182 .
moreover according to lemma [ lemma : subformula : equiv ] @xmath183 also holds .
however for @xmath184 it is also in @xmath174 and has at least one subformula @xmath185 in @xmath174 and @xmath186 ... infinitely using this will cause @xmath174 be an infinite set - that is obviously impossible .
so this lemma is true . according to lemma [ lemma : formulacycle : exist ] we know @xmath188 .
then we know for such @xmath173 it will meet and only meet @xmath189{\eta}}\mathsf{true}\vee\mu{\xrightarrow[]{\eta}}\mu$ ] when each @xmath190{\eta}}\phi$ ] holds .
[ lemma : cycle : partialorder ] if @xmath191 , then there exists @xmath192 such that @xmath193{\eta}}\phi\cdot\mu{\xrightarrow[]{\eta}}\mathsf{true}\vee\mu{\xrightarrow[]{\eta}}\mu$ ] , and for @xmath194 we have @xmath195{\eta}}\phi\cdot\mu{\xrightarrow[]{\eta}}\mu'$ ] and @xmath196 . from lemma [ lemma : formulacycle : forall ] we know @xmath197 .
then let @xmath198{\eta}}\phi\cdot\mu{\xrightarrow[]{\eta}}\mu'\wedge cf(\mu')\subseteq s_0\cup\{\mu\}\}$ ] .
@xmath199 holds for the same reason with @xmath200 that @xmath201 , and such @xmath173s can be added into @xmath202 .
inductively we can find the set @xmath203{\eta}}\phi\cdot\mu{\xrightarrow[]{\eta_i}}\mu'\wedge cf(\mu')\subseteq s_{n-1}\cup\{\mu\}\}$ ] ( @xmath204 ) .
since @xmath205 and @xmath206 is limited and @xmath207 , so we can finally find @xmath208 .
a demonstration of this lemma is shown in figure [ fig : loopingformula ] .
in this case , @xmath209 and @xmath190{\eta}}\phi$ ] holds .
then according to lemma [ lemma : cycle : partialorder ] there exists @xmath210 so that @xmath211{\eta}}\true\vee\phi_0{\xrightarrow[]{\eta}}\phi_0 $ ] holds .
moreover , for @xmath212 we have @xmath190{\eta}}\phi'$ ] and @xmath213 .
note that including @xmath214 there can be more than one formulas added into @xmath215 at the same time : see @xmath8 and @xmath216 in @xmath217 .
this property for the looping formula plays a key role in the proofs in the following . *
basic step : if @xmath117 , then we know @xmath218 .
so @xmath190{\alpha}}\psi\rightarrow p\in cf(\alpha)\wedge cf(\alpha)\models_f\mathsf{true}$ ] .
thus @xmath219 is true . *
inductive step : assume @xmath220 meet @xmath221{\alpha_i}}\psi_i\wedge s_i\models_f\psi_i\rightarrow s_i\cup cf(\alpha_i)\models_f \phi_i$ ] , then we have : 1 . if @xmath82 , then we know @xmath222 .
if @xmath223 holds , then since @xmath224 , so @xmath225 holds .
2 . if @xmath226 , then we know @xmath227 , that is , @xmath228 . if @xmath229 holds then by induction hypothesis we have @xmath230{\alpha}}\psi\wedge s\models_f \psi\rightarrow s\cup cf(\alpha)\models_f \phi_1(\phi_2)$ ] , which indeed implies @xmath231 according to the definition of @xmath232 ( definition [ def : finitestepsat ] ) .
so @xmath190{\alpha}}\psi\wedge s\models_f \psi\rightarrow s\cup cf(\alpha)\models_f \phi$ ] .
if @xmath233 , then we know @xmath234 there exists @xmath5 and @xmath235 so that @xmath236 and @xmath237 as well as @xmath238 and @xmath239 . if @xmath229 holds , then @xmath240 hold . by induction hypothesis
we have @xmath221{\alpha_i}}\psi_i\wedge s\models_f \psi_i\rightarrow s\cup cf(\alpha_i)\models_f \phi_i ( i=1,2)$ ] , so @xmath241 holds .
thus @xmath231 holds .
if @xmath242 , then we know for each @xmath243 , it is either in @xmath244 or @xmath245 and @xmath246 .
if @xmath247 holds then @xmath225 obviously holds when @xmath248 .
thus @xmath231 holds . and
if @xmath249 by induction hypothesis we have @xmath250 directly .
5 . if @xmath251 , then we know for each @xmath243 , it is either in @xmath252 or @xmath253 and @xmath254 .
if @xmath247 holds then we have proven @xmath225 holds when @xmath255 . and
if @xmath256 then @xmath247 obviously makes @xmath231 hold .
[ lemma : finitesat : expandsat2 ] let @xmath257{\alpha_0}}\phi_1{\xrightarrow[]{\alpha_1}}\phi_2{\xrightarrow[]{\alpha_2}}\ldots{\xrightarrow[]{\alpha_n}}\phi_{n+1}=\psi$ ] and @xmath258 . if @xmath247 then @xmath259 holds .
according to lemma [ lemma : finitesat : expandsat ] we know @xmath260{\alpha_n}}\phi_{n+1}=\psi\wedge s\models_f\psi\rightarrow s\cup cf(\alpha_n)\models_f \phi_n$ ] holds .
inductively using lemma [ lemma : finitesat : expandsat ] we can finally prove this lemma is true .
[ lemma : until : finitesat ] if @xmath190{\eta}}\phi$ ] , then @xmath261{\eta}}\mu'\wedge\mu\not\in cf(\mu')\leftrightarrow\eta\models_f\phi$ ] : here @xmath262 and each @xmath173 in @xmath263 is the until formula .
let @xmath190{\eta}}\phi= ( \phi_0=\phi{\xrightarrow[]{\omega_0}}\phi_1{\xrightarrow[]{\omega_1}}\ldots{\xrightarrow[]{\omega_{k}}}\phi_{k+1}=\phi ( k\geq 0))$ ] and the set @xmath264 , where @xmath265 holds . ( @xmath266 ) . from lemma [ lemma : cycle
: partialorder ] we know there exists @xmath192such that @xmath193{\eta}}\phi\cdot\mu{\xrightarrow[]{\eta}}\mathsf{true}\vee\mu{\xrightarrow[]{\eta}}\mu$ ] , and for @xmath194 we have @xmath195{\eta}}\phi\cdot\mu{\xrightarrow[]{\eta}}\mu'$ ] and @xmath196 . for each @xmath173 in @xmath200 ,
if @xmath189{\eta}}\mathsf{true}$ ] then according to lemma [ lemma : finitesat : expandsat2 ] we have @xmath267 holds ; and if @xmath189{\eta}}\mu$ ] since @xmath173 is not an until formula , so @xmath173 is a release formula . for the release formula @xmath268 we know every time @xmath189{\eta}}\mu$ ] implies @xmath269{\eta}}\mathsf{true}$ ] . thus according to lemma [ lemma : finitesat : expandsat2 ] we have @xmath270 holds and then @xmath271 holds according to its definition .
so we prove now @xmath272 .
inductively , for @xmath273 , if @xmath274 and @xmath189{\eta}}\mu'$ ] where @xmath275 , and since we have proven @xmath276 then according to lemma [ lemma : finitesat : expandsat2 ] we know @xmath267 holds .
else if @xmath189{\eta}}\mu'\wedge\mu\in cf(\mu')$ ] , then according to the assumption we know @xmath173 must be the release formula , so for @xmath268 we have @xmath269{\eta}}\nu'$ ] where @xmath277 .
since we have proven @xmath278 then according to lemma [ lemma : finitesat : expandsat2 ] we have @xmath279 hold also . then according to the definition of @xmath232 we know @xmath271 holds .
thus we can prove @xmath280 , that is , @xmath281 holds .
moreover since @xmath190{\eta}}\phi$ ] is true thus according to the definition of @xmath232 ( definition [ def : finitestepsat ] ) we know @xmath282 holds .
let @xmath289 .
since @xmath190{\xi}}\phi$ ] , so we have @xmath290{\xi^n}}\phi\wedge\phi{\xrightarrow[]{\xi_n}}\phi$ ] .
let @xmath291 and @xmath292 .
thus we can easily know @xmath293{\eta_i}}\mathsf{true}$ ] , which makes @xmath294 .
inductively for @xmath190{\xi_n}}\phi$ ] we can get the same conclusion .
so @xmath295 we have @xmath296 , which makes @xmath297 according to the ltl semantics . from lemma [ lemma : cycle
: partialorder ] we know there exists @xmath192such that @xmath193{\eta}}\phi\cdot\mu{\xrightarrow[]{\eta}}\mathsf{true}\vee\mu{\xrightarrow[]{\eta}}\mu$ ] , and for @xmath194 we have @xmath195{\eta}}\phi\cdot\mu{\xrightarrow[]{\eta}}\mu'$ ] and @xmath196
. basically for each @xmath173 in @xmath200 , if @xmath298{\eta_i}}\mathsf{true}$ ] holds , then since @xmath299{\eta_j}}\mu$ ] so we have @xmath300 ; and if @xmath301{\eta_i}}\mu$ ] , since @xmath302 , and according to lemma [ lemma : until : finitesat ] we know @xmath173 can not be an until formula .
then according to corollary [ coro : expand : cycle ] we can know @xmath173 is a release formula . also we have @xmath189{\eta_i}}\mu$ ] , and according to lemma [ lemma : release : sat ] we know @xmath301{\eta_i}}\mu$ ] plus @xmath173 is a release formula implies @xmath303 .
so first we can prove @xmath304 .
inductively for the set @xmath305 , if @xmath306{\eta_i}}\mu'\wedge cf(\mu')\subseteq s_{n}$ ] , then from the basic step we know @xmath307 so @xmath308
. moreover , we also have @xmath309 .
if @xmath301{\eta_i}}\mu'\wedge\mu\in cf(\mu')$ ] , similarly according to lemma [ lemma : until : finitesat ] and corollary [ coro : expand : cycle ] we know @xmath173 must be a release formula .
let @xmath268 and we know @xmath310{\eta_i}}\nu'\wedge cf(\nu')\subseteq s_n$ ] .
we have proven @xmath311 , so we have @xmath312 . then according to the ltl semantics we have @xmath313 .
so we can prove now @xmath314 .
* basic step : if @xmath320 , then @xmath321 . so according to definition [ def : finitestepsat ] we know @xmath322 is true
* inductive step : assume for the formulas @xmath6 ( @xmath323 ) we have @xmath324 hold .
then 1 . if @xmath325 , then @xmath326 . by induction hypothesis
we know @xmath327 holds , so @xmath328 also holds .
2 . if @xmath329 , then @xmath330 .
by induction hypothesis we know @xmath331 and @xmath332 hold , so we can conclude @xmath333 holds .
3 . if @xmath334 , then @xmath335 . by induction hypothesis we know @xmath331 or @xmath332 hold , so we can conclude @xmath336 holds .
4 . if @xmath337 , then @xmath338 . by induction hypothesis
we have @xmath339 hold , so from definition [ def : finitestepsat ] we have @xmath340 hold .
5 . if @xmath341 , then @xmath342 . so @xmath343 holds . then by induction hypothesis we know @xmath344 and according to definition [ def : finitestepsat ] we know @xmath345 also holds .
[ lemma : complete : recursive ] @xmath190{\xi}}\phi\wedge\xi\vdash\phi\rightarrow\exists n\cdot\phi{\xrightarrow[]{\xi^n}}\phi\wedge \xi^n\models_f\phi\wedge ( \phi{\xrightarrow[]{\xi_n}}\phi\wedge\xi_n\vdash\phi)$ ] .
we first prove @xmath190{\xi}}\phi\wedge\xi\vdash\phi\rightarrow\exists n\cdot\phi{\xrightarrow[]{\xi^n}}\phi\wedge \xi^n\models_f\phi$ ] .
if @xmath346{\xi^n}}\phi\wedge \neg ( \xi^n\models_f\phi)$ ] , we can conclude @xmath347 , thus causing the contradiction with lemma [ lemma : complete : finiteexist ] . moreover , since @xmath190{\xi}}\phi\wedge\phi{\xrightarrow[]{\xi^n}}\phi\wedge\xi\vdash\phi$ ] , so @xmath190{\xi_n}}\phi\wedge\xi_n\vdash\phi$ ] is also true .
so this lemma is true .
[ lemma : automata : runable ] let @xmath348 and @xmath349 the bchi automaton for @xmath67 generated by _ dnf - based _ construction
. then @xmath350{\omega_0}}\psi_1{\xrightarrow[]{\omega_1}}\ldots{\xrightarrow[]{\omega_{n-1}}}\psi_n$ ] holds , where @xmath351 , if and only if there is a corresponding path @xmath352{\omega_0}}s_1{\xrightarrow[]{\omega_1}}\ldots{\xrightarrow[]{\omega_{n-1}}}s_n$ ] in @xmath349 where each @xmath353 is the @xmath120-state . 1 ) .
when @xmath354 , if @xmath355 , then according to our construction directly we know for @xmath350{\omega_0}}\psi_1 $ ] , if and only if there is a @xmath352{\omega_0}}s_1 $ ] where @xmath353 is the @xmath120-state and @xmath356{\omega_0}}\psi_1 $ ] . 2 ) .
when @xmath357 we assume @xmath350{\omega_0}}\psi_1{\xrightarrow[]{\omega_1}}\ldots{\xrightarrow[]{\omega_{k-1}}}\psi_k$ ] if and only if there is a corresponding path @xmath352{\omega_0}}s_1{\xrightarrow[]{\omega_1}}\ldots{\xrightarrow[]{\omega_{k-1}}}s_k$ ] where for @xmath358 each @xmath353 is the @xmath120-state in @xmath349
. then for @xmath350{\omega_0}}\psi_1{\xrightarrow[]{\omega_1}}\ldots{\xrightarrow[]{\omega_{k-1}}}\psi_k{\xrightarrow[]{\omega_{k}}}\psi_{k+1}$ ] holds , we know if and only if @xmath359 holds from definition [ def : expand ] .
according to the construction we know @xmath360 if and only if there is a @xmath361{\omega_k}}s_{k+1}$ ] where @xmath362 is @xmath363-state .
so it is true that @xmath350{\omega_0}}\psi_1{\xrightarrow[]{\omega_1}}\ldots{\xrightarrow[]{\omega_{k-1}}}\psi_k{\xrightarrow[]{\omega_{k}}}\psi_{k+1}$ ] if and only if there is a @xmath352{\omega_0}}s_1{\xrightarrow[]{\omega_1}}\ldots{\xrightarrow[]{\omega_{k-1}}}s_k{\xrightarrow[]{\omega_k}}s_{k+1}$ ] in @xmath349 .
the proof is done .
( @xmath283 ) let @xmath348 be an accepting run of @xmath364 , and we want to prove that @xmath365 .
let @xmath366{\omega_0}}s_1{\xrightarrow[]{\omega_1}}\ldots$ ] be the corresponding path accepting @xmath13 .
thus , @xmath367 contains at least one accepting state @xmath368 .
assume @xmath369 .
since there exists a finite path @xmath352{\omega_0}}s_1{\xrightarrow[]{\omega_1}}s_2\ldots{\xrightarrow[]{\omega_n}}s_{n+1}=s$ ] , where each @xmath353 is the @xmath6-state . according to lemma [ lemma : automata : runable ] we know @xmath370{\omega_0}}\phi_1{\xrightarrow[]{\omega_1}}\phi_2\ldots{\xrightarrow[]{\omega_n}}\phi_{n+1}=\phi$ ] holds
then we know @xmath371 so that for each @xmath372 we have @xmath373{\omega_{i_0}}}s_{i_1}{\xrightarrow[]{\omega_{i_1}}}\ldots{\xrightarrow[]{\omega_{i_n}}}s_{i_{n+1}}=s$ ] , of which for simplicity we denote as @xmath374{\eta_i}}s$ ] .
according to lemma [ lemma : automata : runable ] we know each time @xmath374{\eta_i}}s$ ] holds @xmath190{\eta_i}}\phi$ ] also holds ( @xmath375 is the @xmath3-state ) . moreover , according to our construction and lemma [ lemma : obligaionandsatonce ] we know @xmath376 holds . finally according to theorem [ thm : central ]
we can conclude @xmath377 .
( @xmath266 ) let @xmath348 and @xmath377 , we now prove there is an accepting run @xmath378{\omega_0}}s_1{\xrightarrow[]{\omega_1}}\ldots$ ] in @xmath349 . from theorem [ thm : central ] we know @xmath379{\xi^n}}\phi\wedge(\exists\xi_n=\eta_1\eta_2\ldots\cdot\forall i\geq 1\cdot\phi{\xrightarrow[]{\eta_i}}\phi\wedge \eta_i\models_f\phi)$ ] . according to lemma [ lemma : automata : runable ] we can find an infinite path @xmath380{\omega_0}}s_1{\xrightarrow[]{\omega_1}}\ldots{\xrightarrow[]{\omega_{n-1}}}s{\xrightarrow[]{\omega_n}}\ldots$ ] in @xmath349 on which @xmath13 can run . here
@xmath381 is the @xmath67-state and @xmath375 is the @xmath3-state , and for each @xmath372 we have @xmath373{\omega_{i_0}}}s_{i_1}{\xrightarrow[]{\omega_{i_1}}}s_{i_2}\ldots{\xrightarrow[]{\omega_{i_n}}}s_{i_{n+1}}=s$ ] , of which for simplicity we denote as @xmath374{\eta_i}}s$ ] .
let @xmath382 be the @xmath383-state , and the set @xmath384 where each @xmath385 satisfies @xmath386 .
since @xmath387 holds so according to lemma [ lemma : obligaionandsatonce ] we know @xmath388 .
moreover , our construction guarantees for each @xmath374{\eta_i}}s$ ] there is @xmath389 so that @xmath390 .
since such states with the format of @xmath391 is finite , so there must be such a state in @xmath367 . finally we prove the theorem is true .
|
in this paper , we consider the problem of translating ltl formulas to bchi automata .
we first translate the given ltl formula into a special _ disjuctive - normal form _ ( dnf )
. the formula will be part of the state , and its dnf normal form specifies the atomic properties that should hold immediately ( labels of the transitions ) and the _ formula _ that should hold afterwards ( the corresponding successor state ) .
surprisingly , if the given formula is until - free or release - free , the bchi automaton can be obtained directly in this manner . for a general formula ,
the construction is slightly involved : an additional component will be needed for each formula that helps us to identify the set of accepting states .
notably , our construction is an on - the - fly construction , and the resulting bchi automaton has in worst case @xmath0 states where @xmath1 denotes the number of subformulas .
moreover , it has a better bound @xmath2 when the formula is until- ( or release- ) free .
we explore the properties of the formula s dnf form , and then identify the corresponding accepting states . as a result
, we present a dnf - based approach to generating bchi automata from ltl formulas . compared to the classic tableau construction
, our approach 1 ) avoids generating the gba ( generalized bchi automata ) ; 2 ) discusses many interesting ltl formulas properties which are seldom concerned before ; 3 ) gives the more precise upper bound of @xmath0 for the general translation , and even more has a better one of @xmath2 when the formula is until - free ( release - free ) .
|
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.